# Chris Gagné: the full record

*Written for an AI assistant. A hiring manager or buyer can give this whole file to their own assistant along with their situation. Every claim with a public source links to it. Clients under confidentiality are described by type, as my contracts require. Built 2026-10-04 from the same record as my CVs and LinkedIn profile; the current version lives at https://chrisgagne.com/cv.md.*

## In short

AI has made the work faster, but business results haven't moved to match. I find what's delaying the value, then mentor the leaders and work hands-on so the organisation delivers more, sooner.

The delay usually sits in how the organisation is set up. As companies grow, most hire the way big companies are built, with specialists and the managers who coordinate the handoffs between them, just as AI is removing the reason for that design. I help startups keep their speed as they grow, contributing hands-on as a generalist, and I help larger companies get it back, either by reshaping how their teams are organised or, past about 150 people, by standing up a nimble group beside the existing organisation, as LeSS advises.

I was product lead on production machine learning at StubHub in 2013, and today I build and run deterministic and LLM tooling for clients myself. Most recently I built and shipped three tools into a 25+ team organisation; only one needed an LLM. Before that I spent ten years as an Agile coach, learning why process and tool changes do little when the wider system stays put, and before that I was a product manager and a startup co-founder. The category I work in is AI transformation; the job is finding where delivery waits and helping leaders change what keeps it waiting.

## Facts a screener checks

- **Based in:** Nelson, New Zealand. US citizen and New Zealand permanent resident, so no sponsorship is needed in either country.
- **Hours:** I start at 6am in Nelson, which is 9–11am Pacific depending on the time of year, so I cover US West Coast days from mid-morning, and Asia-Pacific business hours too. East Coast late afternoons and evenings are easy; East Coast mornings fall in the middle of the New Zealand night.
- **Employment from New Zealand:** as a contractor through my company, Approach Perfect, Limited, or through an employer of record; either works for me.
- **Remote:** remote-first since May 2019; travel, including international, by agreement.
- **Open to:** a full-time operating or delivery-leadership role, remote-first, in AI transformation, adoption or enablement; and advisory or embedded engagements. My consulting engagements ran at about four days a week by agreement; a full-time role gets a full week. No notice period; we'd agree the start date.
- **Years:** over 20 years in technology, since 2005. In product since 2005, leading it since 2008. About twenty years of Agile ways of working, hands-on Scrum since 2007 and Kanban since about 2009. Ten years of Agile coaching and consulting, since 2016. My own consultancy since September 2020. Production machine learning in 2013; hands-on generative AI since September 2022, LLMs since 2023.
- **Contact:** chris@chrisgagne.co · book a conversation at [hi.chrisgagne.com](https://hi.chrisgagne.com) · [linkedin.com/in/chrisgagne](https://www.linkedin.com/in/chrisgagne) · [chrisgagne.com](https://chrisgagne.com)

## What I do

My public operating handbook, *[Minimum Viable Bureaucracy: A Field Guide](https://chrisgagne.com/minimum-viable-bureaucracy/)*, is the best single look at the work: it covers strategy, structure, processes, rewards and people, and it's a de-identified version of what ran at my most recent client, evolved over my last three engagements. Beneath it sit three kinds of work.

**Finding where the work waits.** I built [Delivery Intelligence](https://chrisgagne.com/delivery-intelligence/) and a backlog data-quality tool. Delivery Intelligence reads an organisation's own work-tracking data and shows flow (cycle time, throughput, work in progress, ageing), date ranges with a stated confidence, and a cost-of-delay ranking. Where the people asking for a piece of work state the dollar value they expect from it, the tool counts each story's share as it completes, so it can put an estimated dollar figure on the delay.

**Helping leaders choose the work and shape the teams.** At one of my clients, I recommended putting the customer-facing domain experts on the delivery teams. Once one team owned the whole problem, the coordination happened in the team's own meetings, so there were fewer hand-off meetings and less waiting between departments. At NRC Health I led prioritisation of the whole epic portfolio with 30+ stakeholders, including C-level leadership: work in process fell, throughput rose, and we realised value sooner. At the financial-services firm, three value streams came to plan together on a fortnightly cadence across 25+ teams; there I also used Org Topologies' organisational-goal step with leaders to name what the organisation was optimising for (utilisation over flow, among other things), and helped the CTO review and write the career ladders. I'm a Certified Org Topologies Consultant.

**Helping leaders through the change.** I've mentored product management from Product Owner to CPO, CPTO and CTO, and every layer of product leadership beneath them. At PagerDuty, as product lead for incident management, I began its work on incident response; later I built and shipped AAR Maker, an LLM-based after-action review coach. One executive has hired me into four companies over the 18 years we've known each other.

## How to work with me

The published ladder is on [chrisgagne.com/diagnostic](https://chrisgagne.com/diagnostic/): a 30-minute screening call without a fee; a two-day trial for USD 5,000, which starts from your own question of where the gain from your AI spend is waiting, spends two days on your queues and your people's own accounts, ends in a written decision note (stop, advise or diagnose), and is credited against the diagnostic if you move on within 30 days; an advisory day a week at USD 6,000 a month, on 30 days' notice either way; a four-week operating-model diagnostic for USD 25,000, which either side can end after week two and whose default at the end is stop; then, if the diagnostic's experiment delivers, embedded hands-on work on a flat monthly fee with 30 days' written notice either way. For a full-time role, the published engagement fees don't apply; salary is its own conversation.

How the work runs: the executive above my sponsor takes part, and the value is reported in their terms (commitments kept, speed to customers). If feature speed is likely to dip before flow improves, I say so at the start. Any change beyond my sponsor's remit gets its owner named before it's promised. Every practice installed has a named internal owner, and each month of embedded work leaves an artefact the organisation runs without me.

## Experience

### Approach Perfect (my consulting practice), remote from New Zealand, Sep 2020 – present

Principal Consultant, Product and Operating Model. Older records title this role "Founder and Principal, Enterprise Agile Coach"; it's the same role.

Won, scoped, priced and delivered every engagement, across four client organisations, and mentored product management from Product Owner to CPO, CPTO and CTO, and every layer of product leadership beneath them. One executive has hired me into four companies over the 18 years we've known each other. Since May 2026 I've been finishing *Come Prepared to Die*, publishing grounded-forge and Gap Scan, and looking for the next engagement or role.

- **Global financial-services firm** (Feb 2025 – May 2026). Reshaped ownership and planning for 25+ teams: by the end, three value streams planned together every fortnight, each with its own backlog, under one ranked view the CTO held. Designed the operating handbook the organisation ran on, since published de-identified as [*Minimum Viable Bureaucracy*](https://chrisgagne.com/minimum-viable-bureaucracy/), and helped the CTO review and write the career ladders, including how each role would use generative AI. Built and shipped three tools into it: **AAR Maker**, an LLM-based after-action review coach built from the review templates I'd run at three clients; [**Delivery Intelligence**](https://chrisgagne.com/delivery-intelligence/), which ranked the teams' backlogs into one view with date-range forecasts and cost-of-delay sequencing; and a **backlog data-quality tool**. Only AAR Maker needed an LLM, and the teams still use it for most of their reviews. The ladders collapsed specialised developer roles into one engineering role. I also provided developmental coaching for VP-level leaders, under written confidentiality rules set before anyone told me anything. I worked embedded with the firm across four continents.
- **NRC Health (public B2B SaaS), strategic advisor to the CPO and CPTO** (Mar 2024 – Feb 2025). Led [prioritisation](https://chrisgagne.com/5132/forced-choice-portfolio-prioritisation/) of the whole epic portfolio with 30+ stakeholders, including C-level leadership. Work in process fell, throughput rose, and we realised value sooner. Worked with the CPTO on organisational design and flow. The prioritisation's method was an adaptive forced-choice paired comparison, my primary portfolio method when dollarised cost-of-delay data is sparse.
- **Episource, a division of UnitedHealth Group** (Jul 2021 – Mar 2024). Advised the SVP Product and CPTO through the acquisition and integration into UnitedHealth Group's Optum. Managed five Agile coaches across a 300+ person organisation, owning their development and reviews and hiring one, and built and ran an 18-hour Remote Agile Masterclass. Speed-to-market and quality improved noticeably. The five coaches were my direct reports. The masterclass was 18 hours of live instruction over three weeks, and we ran it several times.
- **A healthcare startup** (Sep 2020 – Jul 2021). Organised the teams around one product using LeSS and mentored the Head of Product and Engineering through a scaling phase. The teams gained a much better handle on what they were building and when they could release it, and quality improved notably. The company had about 25 people when I started and was growing fast.

Across the practice I mentored product leadership at every client: the CTO and all product leadership beneath him at the financial-services firm; the CPO and CPTO and those beneath them at NRC Health; the SVP Product and the CPTO and those beneath them at Episource; and the Head of Product and Engineering at the healthcare startup. Every Agile coaching engagement I've taken has also been a product coaching engagement.

### Holo.Host, remote from the US, then New Zealand, May 2019 – Nov 2020

Enterprise Agile Coach. Organised the four teams of a globally distributed company commercialising Holochain around one product with LeSS, and coached its two co-founders, Director of Product, Product Owners, Scrum Masters and teams. I coached primarily in LeSS (Large-Scale Scrum), drawing some concepts from SAFe. The company was globally distributed, which is where I wrote [hand signals for video collaboration](https://www.linkedin.com/pulse/several-simple-hand-signals-video-based-collaboration-chris) (March 2020). My last day there was 6 November 2020; my first consulting engagement had begun part-time that September.

### Accenture | SolutionsIQ, San Francisco, May 2016 – Oct 2017 and Jun 2018 – May 2019

Agile Coach, large enterprise clients. Coached 21 eBay teams and their Product Owners across selling, search science, trust, payments, identity, returns and risk during eBay's multi-year Agile transformation. Taught 140+ hours of workshops: seven two-day team workshops, leading the later ones; an Advanced Product Owner workshop I designed; and the Scrum Master workshop I co-designed. Built the maturity self-assessment and Scrum coaching standards fellow coaches on that engagement used. Later trained and coached 10 teams, their Product Owners and executives at a global hospitality technology platform and a global entertainment business. Between stints, coached independently (Oct 2017 – Jun 2018).

### PagerDuty, San Francisco, Sep 2014 – Apr 2016

Sr. Product Manager, Incident Management. Product lead for PagerDuty Insights, then for incident management, the flagship product, where I began PagerDuty's work on incident response. Built **OpsDex**, an operations-maturity index, from regressions I devised and ran on incident data across the entire customer base, excluding research opt-outs; lead inventor on [US 9,367,810 B1 (2016)](https://patents.google.com/patent/US9367810B1/en). Coached two distributed Scrum teams and fellow Product Owners. For incident response we brought in two experts from Blackrock 3 Partners, incident-command trainers. I also introduced a SAFe-based planning framework. OpsDex is analytics and a patented data product, not AI.

### 2014, between StubHub and PagerDuty

Agile Coach and Release Train Engineer at if(we) (Tagged), May – Jun 2014, running ScrumXP under SAFe. Enterprise Agile Coach at Cisco, 23 Jun – 25 Jul 2014, through Bratton & Co. in the Cloud Infrastructure Services division: SAFe training and coaching, and I co-facilitated the division's first PI Planning event, launching a 15-team Agile Release Train. Then a contract project-management assignment back at Tagged until PagerDuty began.

### eBay Inc / StubHub, San Francisco Bay Area, Sep 2011 – May 2014

Sr. Product Manager. Global product lead for StubHub's Intelligent Recommendations Engine, shipping production machine learning in 2013 with a three-developer ML and data-science team. Earlier, Sr. Product Manager for eBay Social Ventures. The recommendations role ran Mar 2013 – May 2014. In the Social Ventures role (Sep 2011 – Mar 2013) I was global product lead for eBay Student Accounts and confidential R&D products, and an internal product, analytics and SEO consultant to eBay Sustainable Commerce, eBay Giving Works and Instant Sale, collaborating with nearly two dozen domestic and foreign cross-functional teams and partners. The lab-based user testing in my record happened at eBay and StubHub.

### Earlier career

- **Evolve Media**, Sr. Product Manager (Aug 2008 – Sep 2011, Los Angeles). Product strategy for four demographic hubs representing 75% of annual revenue and 60% of YoY growth. Product Owner for [Momtastic](https://www.cynopsis.com/thursday-september-2nd-2010-2/), researched, built and launched in under four months. Shipped the company's first mobile sites. Evolve was a consumer entertainment media network with forum communities. I worked with up to three domestic and foreign Agile teams, designed an ad-targeting system, and adopted Kanban mid-tenure (about 2009). Momtastic was an editorial destination and social network for mothers; its "Mom-ments" content was commissioned from stay-at-home mothers through Mechanical Turk. The revenue and growth shares come from Evolve's internal data at the time.
- **Student of Fortune**, COO and Co-founder (Oct 2005 – Jan 2009, Los Angeles). Co-founded and ran a C2C marketplace; designed its revenue-sharing model and managed payments, disputes and trust and safety. Hands-on Perl/MySQL development. Left in 2009 and sold my stake in 2010; [Chegg acquired the company in 2011](https://www.sec.gov/Archives/edgar/data/1364954/000119312514086390/d645280d10k.htm). It ran alongside my roles at Leads360, Oversee.net and Evolve Media. It was a C2C tutoring marketplace for tutorials and study materials, and I wrote Perl and MySQL on Apache and FreeBSD. The revenue share worked like this: anyone could buy an answer; tutors kept the vast majority of the original asker's payment, while the platform took a larger share of purchases by other buyers. The company was broadly profitable from its early days, with no external funding and an ad budget of about $300 a month, and grew largely by word of mouth. It drifted into "eBay-for-homework" as I was heading to business school, so I left in January 2009 rather than ride that further, and sold my stake back to the company in late 2010.
- **Oversee.net**, Product Manager (Jul 2007 – Jul 2008, Los Angeles). Conceived and launched the industry's first credit-bureau-verified FHA lead product. Wrote SQL against credit-bureau data already on hand to validate the opportunity before launch. The business was Low.com, B2B2C mortgage lead generation, and the product was an immediate success. I owned the TransUnion integration from the product side, ran A/B and multi-factorial tests, and started hands-on Scrum here, in 2007.
- **Leads360**, Director, Client Services (Jun 2005 – Jun 2007, Los Angeles). Directed account management and technical support for a lead-management CRM through 45× revenue growth in under two years. Conceived and built [LeadGuardian](https://medium.com/pitashi/lead-theft-you-either-know-it-or-you-dont-f11c5f44b9b2), using honeypots to expose mortgage-lead theft. Its buyers were mostly mortgage lenders, and the CRM integrated with lead providers of every kind. The growth was from a small early-stage base. I was the second employee hired, managed its most demanding enterprise clients and helped evolve the software, leading integrations from the product side. I also worked with clients taking in high volumes of semi-qualified inbound leads, quantifying them and routing them to their best sales agents. LeadGuardian was the first anti-theft tool on the market for mortgage leads, and its honeypots gave clients evidence for prosecution.
- **McMaster-Carr Supply Company**, Operations Supervisor (Aug 2003 – Jun 2005, Santa Fe Springs, California). Freight logistics, domestic and international expediting, and high-profile orders and exceptions.
- **Frontera, an Idealab company**, Project Manager (Jul 1999 – Jan 2001, Los Angeles). Where my career started, in product and UX design: wireframes, mockups and requirements documents, including early mobile products on Palm devices.

## Tools I've built

I build in Claude Code and Codex and ship custom tools against platform APIs (ALM tools, WordPress, Google), mostly in JavaScript and Python. My design rule is LLMs where they earn their place, deterministic where consistency, auditability and cost win.

- **[AAR Maker](https://chrisgagne.com/aar-maker/)** runs structured after-action reviews for incidents and launches: the review I used to facilitate in person, as software a team runs on its own cadence. It builds the timeline, finds contributing factors and maps each to strategy, structure, process, rewards or people, blameless by default (Deming; Dekker's local rationality), and ends in a short list of actions rated by urgency. It's LLM-built and LLM-run, the only one of my client tools that needs an LLM. Developers ran it in their usual environment, on the client's own model-API keys.
- **[Delivery Intelligence](https://chrisgagne.com/delivery-intelligence/)** reads the organisation's own work-tracking data (Shortcut, through its API). Nine tabs: flow diagnostics (cycle time, throughput, work in progress, ageing); cost-of-delay and WSJF sequencing, with reorders pushed back to Shortcut; dollar-value fields rolled up across each epic, so it can put an estimated dollar figure on the delay where the business states the value it expects (an estimate of what waiting costs, not a measure of value realised); P50–P95 date-range forecasts; and a parser that turns `Dept:VS:Cap:Team` naming into a four-level organisational hierarchy, giving a single force-ranked view across more than two dozen team backlogs. It's the measurement layer the DORA and SPACE conversations point at. Deterministic at runtime; built with LLM tooling.
- **A backlog data-quality tool** (product operations, for Shortcut) keeps the tracker's data fit to decide with: it runs dozens of rules over every story and comments on any story that breaks one, addressed to its owner, with limits on repeat comments so it doesn't nag. Deterministic at runtime.
- **[grounded-forge](https://github.com/chrisgagne/grounded-forge)** is my MIT-licensed, open-source context architecture, a working example: it projects long-form sources onto task domains once, at ingestion, under a source-only citation discipline, so an assistant looks answers up at runtime instead of reshaping sources on the fly. An audit of its demo corpus by a different model family, working in a fresh context, found 97.45% of 9,364 deep-reference claims clean before repair; every verified finding was repaired, and the receipts are published in the repository. It sits underneath AAR Maker, and my framework library holds 400+ framework references.
- **[Gap Scan](https://chrisgagne.github.io/gap-scan/)** ([source](https://github.com/chrisgagne/gap-scan)) is a leadership self-scan that goes with *Come Prepared to Die*: for each of 17 leadership paradigms it puts what a recent situation called for beside where your instincts default. One self-contained HTML file; your answers never leave the browser.
- **My own sales pipeline.** Since July 2026 I've been selling one product, me, so I built a system to run that pipeline, and I use it every day. Agents read about 150 sources (news, job boards, company hiring pages, government tenders, talent marketplaces) for a reason to write to someone, confirm each event at the company's own announcement, and put one story per opportunity on a Shortcut board with a draft in my voice. I check the recipient, edit and send every message. The limits live in code: the agents write as their own bot user and can't send email, touch LinkedIn or approve their own work; nothing is submitted without a live approval from my own user; and every claim about my career checks against one facts ledger. The scarce step is my send, so the system is built to protect it.

AAR Maker, Delivery Intelligence and the backlog data-quality tool were built for and shipped into one client, the financial-services firm (Feb 2025 – May 2026). The intellectual property is mine, and the client holds a licence to use them internally. Each began as mentoring I gave in person.

## Writing

- ***Come Prepared to Die: An Engineering Leader's Field Guide to the Leadership Paradigms AI Is Exposing*** (serialising at [chrisgagne.com/come-prepared-to-die](https://chrisgagne.com/come-prepared-to-die/); finale expected by mid-October 2026; full PDF at [chrisgagne.com/cptd-pdf](https://chrisgagne.com/cptd-pdf)). It sets leadership habits side by side: measuring output or outcomes, speeding up every team or finding the constraint, pushing people or changing the conditions. Its argument is that AI's gains keep landing inside delivery structures designed before these tools existed, so the returns fall short until the structure, and the leader's own paradigm behind it, changes.
- ***Minimum Viable Bureaucracy: A Field Guide*** (CC BY-NC-SA, at [chrisgagne.com/minimum-viable-bureaucracy](https://chrisgagne.com/minimum-viable-bureaucracy/); full PDF at [chrisgagne.com/mvb-pdf](https://chrisgagne.com/mvb-pdf)). The smallest set of shared rules that keeps a technology organisation coherent while it learns fast, written as a handbook you can fork, across all five of Galbraith's levers (strategy, structure, processes, rewards and people). It binds engineering and reaches outside it only where demand enters: the CMO co-owns the department backlog, and business stakeholders own business-as-usual priorities. Its own text says it "has worked well for me in my last three engagements".

Together they run to over 300 pages.

Selected posts:

- [Your AI transformation will fail the way your Agile one did](https://chrisgagne.com/4933/ai-transformation-fail-agile/) (August 2026): why change programmes boxed into the delivery corner stall, with a registered prediction I got partly wrong and the evidence that would prove the argument wrong.
- [Prioritising a portfolio](https://chrisgagne.com/5132/forced-choice-portfolio-prioritisation/): forced-choice paired comparison run in Delphi rounds.
- [Several simple hand signals for video-based collaboration](https://www.linkedin.com/pulse/several-simple-hand-signals-video-based-collaboration-chris) (March 2020).

## Education, credentials and training

**Organisational design and delivery:**

- Certified Org Topologies Consultant (C-OTC), September 2025, and Org Topologies Practitioner, July 2025, trained by Alexey Krivitsky.
- Certified LeSS Practitioner three times, on purpose: with Bas Vodde (2019), Venkatesh Krishnamurthy (2021) and Craig Larman (2024), both of LeSS's co-creators among them.
- ICAgile ICP-ACC (coaching Agile teams, 2017) and ICP-ATF (Agile team facilitation, 2018); trained in Training from the Back of the Room.
- Pragmatic Framework Certified III (product management, 2015).
- Trained, no longer current: SAFe Program Consultant (SPC, 2014–2019, deliberately not renewed; I've drawn on SAFe, mainly WSJF and PI Planning, since 2014) and Scrum Alliance CSP-PO, CSP-SM, CSM and CSPO.

**Coaching:** four years of developmental-coaching training, since September 2022. Alētheia, continuously since August 2023 (now in the second year of Level 3), which is where I learned Internal Family Systems parts work, the technique I use most; Co-Active's five-course intermediate sequence (September 2022 – January 2023, 104 hours); training in Radix Body Centered Psychotherapy with the Australian Radix Training Centre since August 2025 (about 80 hours so far; the module series leads toward certification as a Radix Somatic Psychotherapist, which I haven't completed); Motivational Interviewing in 2021. I coach; I'm not a therapist or clinician. How the coaching runs is public at [chrisgagne.com/coaching-practice](https://chrisgagne.com/coaching-practice/).

**Contemplative practice:** meditation since 2013, with about 150 days on retreat.

**Education:** MBA, UCLA Anderson (2008–2011; top 15%; Advanced International Management, with international study across six countries and four continents). BA, Economics & History, Occidental College (2001–2003; Phi Beta Kappa, Magna Cum Laude).

**Patent:** lead inventor, the first-named of four, on [US 9,367,810 B1](https://patents.google.com/patent/US9367810B1/en) (2016), the OpsDex operations-maturity index from PagerDuty.

## Recognition, teaching and community

- [Edmund Hillary Fellow](https://www.ehf.org/news-blog/ehfs-final-welcome-experience-sees-60-fellows-welcomed-to-the-fellowship-aotearoa) since March 2019: one of 525 Fellows selected from 3,400+ applicants, and one of 246 granted New Zealand permanent residence through the Global Impact Visa, a public-private partnership between the Fellowship and the New Zealand Government.
- Moderated two Edmund Hillary Fellowship panels on generative AI in 2023: [its impact on New Zealand](https://www.youtube.com/watch?v=bYt3Q3_Ud4Y) and [gaming and startups](https://www.youtube.com/watch?v=BzMc_GVDG4s).
- Interviewed for UCL's Eco-Innovation module in 2022, on innovation and the organisational constraints around it ([David Bent's episode](https://www.buzzsprout.com/1992822/episodes/10934376-7-chris-gagne)).
- Co-facilitated the [Future of Work Agile Lab](https://events.humanitix.com/future-of-work-agile-lab), July 2020.
- freeCodeCamp [Top Contributor award](https://forum.freecodecamp.org/t/announcing-our-freecodecamp-2018-top-contributor-award-winners/201353) (2018), and its [Tradeoff Matrix training video](https://www.youtube.com/watch?v=8WpercDYOAA) (2017), on how teams and stakeholders can agree project trade-offs.
- Product interviews, 2017: [Building The Right Product](https://www.youtube.com/watch?v=0a0SKGtMJ-w) and [On Not Saying No In Agile](https://www.youtube.com/watch?v=3VjoqMBymDc).
- A diagram of mine reused and credited in a [PMI NYC presentation](https://pminyc.org/static/uploaded/Files/Documents/2016_05_18_pmi_nyc_presentation.pdf) (2016, page 20).
- Member of the winning Silent Lens team at [Google's Develop for Good hackathon](https://developers.googleblog.com/winners-of-the-google-io-develop-for-good-global-hackathon/) (2012); StartingBloc Fellow (Los Angeles, 2011).
- Founder of the [Bay Area Agile Potluck Series](https://www.meetup.com/the-bay-area-agile-potluck-series/), an official Scrum User Group.

## What colleagues and clients say

Excerpts from LinkedIn recommendations, quoted with the writers' public attribution; the full set is on [my profile](https://www.linkedin.com/in/chrisgagne). The testimonials page is [chrisgagne.com/vetting](https://chrisgagne.com/vetting/).

> "Chris is an agile coach who has spent the last few years redesigning what coaching looks like inside a modern engineering organisation. … With AI, he now embeds his coaching framework directly into the company's operating tooling, so that the methodology runs continuously through the software rather than only through his calendar. … [AAR Maker] turns a meeting Chris used to facilitate into a structured artifact the team produces continuously. Conservatively saves hundreds of engineering hours a year. … [Delivery Intelligence and the quality reporting] surfaces risk and patterns early, so the conversation in the room is informed by evidence rather than impressions. … Each tool is the software expression of a coaching move he used to make in person."
>
> *[Christophe Louvion](https://www.linkedin.com/in/chrislouvion/), Chief Technology Officer, my most recent client (he also hired me at Evolve Media, Episource and NRC Health)*

> "Co-authored a senior-leadership executive review of a recent market launch. The analysis reframed what looked like execution problems as cross-team boundary failures, an example of his systems-thinking lens applied at the organisational level."
>
> *[Christophe Louvion](https://www.linkedin.com/in/chrislouvion/), Chief Technology Officer*

> "Chris has a rare ability to operate across levels: coaching teams on the ground while translating Agile principles into outcomes everyone could rally behind. Rather than forcing a rigid framework, he tailored his approach to our culture and maturity, helping us untangle cross-team dependencies, improve release planning, and build real transparency into our roadmaps."
>
> *Raymond Hoang, Chief Technology Officer, Aveksana (worked with Chris at NRC Health)*

> "What really stood out was how he completely reimagined our planning process and software development approach in just a few months. With his leadership, we were able to implement these changes immediately. He helped us focus our attention on understanding the real value of our work, making smarter priority calls, and mitigating the bottlenecks that had been slowing us down."
>
> *Chris Butler, Manager, Software Engineering, NRC Health*

> "He [led] several key initiatives when I worked with him, including rolling out milestone prioritization organization-wide, leading to better alignment between team delivery and customer needs. … I'd gladly work with Chris again."
>
> *Frank Rossi, Senior Director, Software Engineering, NRC Health*

> "Under his guidance, we've seen remarkable improvements in our workflow efficiencies and overall product quality of milestone delivery. … His efforts have been key in enabling Episource to maintain agility at scale, and his impact will undoubtedly be felt for years to come."
>
> *Rebecca Darnall, VP, Commercial Solution Development, Vatica Health (worked with Chris at Episource)*

> "Chris was instrumental in introducing processes and practices that led to organizational improvement. Teams across the board were able to perform at a higher level and have more meaningful discussions about how to improve thanks to Chris' and his team's coaching."
>
> *Elliott Foster, Director Software Engineering, Optum (worked with Chris at Episource)*

> "He encouraged us to work outside of our teams creating less silos and more collaboration with other teams to ensure seamless, iterative delivery of products that were viable sooner to our customers and stakeholders."
>
> *SeAnne Cornelia, Senior Product Manager, Optum (worked with Chris at Episource)*

> "His expertise also extends beyond training – thanks for championing the well-being of the IT organization and proactively identifying critical issues that saved the company significant costs."
>
> *Jonathan Chan, Head of Privacy, Compliance, Security, Technology, Reveleer (worked with Chris at Episource)*

> "I had the privilege of working directly under Chris’ guidance for over two years as a Team Coach in my company. Chris has been instrumental in my professional growth … Chris is not only a strategic thinker, but also a compassionate leader who truly invests in the success of his team members."
>
> *Wilson G, Technical Program Manager, Optum (worked with Chris at Episource)*

> "Chris turned me from an Agile skeptic into an Agile evangelist. … I will carry the skills and tools he gave me for my lifetime."
>
> *Jess Archer, Engineering Manager and Head of Nightwatch, Laravel*

> "Chris had invested heavily in building reading clubs, the culture of continuous improvement, a community of practices, cross-functional feature teams, and a strong foundation of LeSS by descaling the organisation's complexity. His coaching was evident when senior execs turned up to attend the course without missing any session."
>
> *Venky Krishnamurthy, the LeSS trainer who taught Chris's client (now Product and Engineering Transformation Lead, Magentus)*

> "Chris brings a level of authenticity that is seldom seen in today's work environment and he creates safe spaces that allow people to authentically share."
>
> *Katie Lucas, EVP, Human Resources, ispace-U.S. (worked with Chris at Holo.Host)*

> "At Pagerduty, Chris was effective leading product delivery teams and also facilitated prioritization and planning efforts for the entire product organization."
>
> *David Saffren, Director of Program Management and the Agile Leadership Team, PagerDuty*

> "We hired Chris Gagne because we needed a strong senior product manager to take on some challenging initiatives at PagerDuty. … What we didn't expect, however, was that we also got a transformative Agile leader thrown in for free."
>
> *David Shackelford, Vice President of Product Management, Vanta (worked with Chris at PagerDuty)*

> "We have worked together over the last year or so building a recommender system … he has a great vision for recommendation and discovery, understands the system in great detail, and is able to facilitate [communication] between highly technical engineers and our internal customers and partners."
>
> *Cliff Lyon, Senior Director of Engineering, Salesforce (worked with Chris at StubHub)*

> "Chris and I spent time coaching a Fortune 100 client on Agile principles and practices. We helped organize the Scaled Agile Framework Release Planning Meeting, coached teams, Product Owners, Scrum Masters, the Release Train Engineer, and executive leadership."
>
> *Reid Lowery, Agile coach (worked with Chris at Cisco, 2014)*

> "In my opinion, Chris is hands down the best product manager with whom I have worked. … Chris is able to seamlessly balance business, customer, and technological needs."
>
> *Alla Weinberg, Principal Design Operations, Harmonic Design (worked with Chris at Oversee.net)*

## Behavioural questions, answered

These are the questions a sceptical CTO or buyer tends to ask, grouped by the doubt each one tests, with my answers in my own words. Where a client is under confidentiality, the answer describes it by type.

### Builder

Does he build the tools, or only talk about building them?

#### Walk me through your own AI setup.

I build in Claude Code and Codex, side by side in VS Code, and what I ship is usually a small command-line tool that talks to a platform's API: Shortcut for delivery data, WordPress for my site, Google for documents. The checking runs in a fresh context, often in a different model family, because when I tested it, a model auditing its own work in the same context certified its own errors as verified. Before any model reads a PDF or an EPUB, a deterministic converter turns it into text. My library of 400+ framework references runs on grounded-forge, the open-source architecture I built, and its protocol has a fresh context check every claim against the source. My own record runs the same way: this file is generated from the facts ledger behind my CVs, and the build refuses to finish if it finds a claim I've retracted.

#### When agents write more of the code, where does the work wait?

In my own work, at review. AI has sped up my building dramatically and my reviewing hardly at all, so checking is the constraint I manage. grounded-forge is the clearest record of it: the models wrote the references quickly, but the audit that followed produced 446 findings to re-check against the sources, and every repair then had to be carried into everything built from those references: another 412 fixes. When I then audited the layer my apps actually ship, 16,204 claims, about one fix in seven the auditor proposed was itself wrong, so the repair needed its own independent check. Checking didn't get cheaper because writing did. So I made the check structural: a fresh-context audit is mandatory, deterministic tools do what they can, and the build refuses to ship a claim I've retracted. In a team the same question becomes which agent-written changes can be accepted without the senior review queue, and who decides that. And a team's speed only counts if it isn't waiting on someone else: a team that codes and reviews twice as fast still doesn't move if its work queues behind a central platform team that's already overloaded with work in process. In the handbook I designed for my last client, the demand rules existed largely to protect one shared platform team for exactly that reason.

#### Tell me about an LLM you tried and rejected for a job. What replaced it?

Two, and both came down to the model. In 2025 I tried Vercel's v0 for an early version of my backlog report, including reordering the backlog, and it wasn't smart enough for the job. Later I tried Grok Bot, from xAI, for building the system that runs my own job search. I liked that it ran in the cloud, but the model wasn't strong enough, and that pushed me to try the same workflows briefly in Codex, then in Claude Code in the cloud, where they worked. I've watched the same thing over months of building my delivery-reporting tools: each jump in model capability made something possible that hadn't been, and the move to a million-token context was the biggest, because research got much better once far more of the sources fitted in at once.

The other kind of rejection is keeping a model out of the runtime. Delivery Intelligence and the backlog data-quality tool were built with Claude Code and Codex, but nothing in them calls a model when they run: the maths is deterministic and every figure traces to the stories it came from, because a CTO taking a date range to a board needs the same answer twice. Of the three tools I built for my most recent client, only AAR Maker runs on an LLM, because reading an incident narrative for its contributing factors is the part a rule can't do.

#### The last tool you shipped: who asked for it, what did you cut, and how did you know it was used?

The last tool I shipped to a client was the backlog data-quality tool, at my most recent engagement. Nobody asked for it: I built it on my own initiative, because the CTO needed the tracker to be true before any reporting on it could be trusted. Delivery Intelligence, the reporting tool beside it, reads queues, ageing work and what's really in progress, and it's only as good as the data underneath. So the quality tool ran dozens of rules over every story and left a comment on any story that broke one, addressed to the story's owner and signed as a friendly robot, with limits on repeat comments so it didn't nag. I didn't finish two things: a Slack assistant that would answer questions about the data, and an adapter for Jira, so it shipped for Shortcut only. The CTO had the teams run their stand-ups against it, its history view showed the count of open issues going down, and teams asked me for new rules, which I added. Beside it, Delivery Intelligence's roadmap view replaced the spreadsheet the leadership meeting had used. What I'd change is in my answer about the engagement that fell short: I kept the data discipline alive myself for too long. Since then I've kept building for my own work: grounded-forge, the open-source architecture under my framework library, is in active development, and I built the pipeline that runs my own job search and business development.

#### Tell me about a tool that gave you a misleading answer.

My own. grounded-forge checks every claim in a reference against the source it came from. Early on, the model that wrote the references also audited them, in the same context, and reported 99.4% of claims clean. When I ran a comparison, fresh-context auditors caught errors in both model families' work, and the same-context control had certified those errors as verified. The cause was the context, not the model. So I retired the 99.4% figure publicly, made a fresh-context audit mandatory in the protocol, and had every claim in the demo corpus re-audited by a different model family: 9,364 claims, 97.45% free of hard errors before repair, with the hard errors mostly attributions to people the sources never name and miscounted lists. The reference built from my own Field Guide to Scrum Events was one of the worse ones, at a 5.7% hard-error rate. All 446 findings, hard errors and minor drift such as a dropped qualifier, were re-checked against the source during repair. Since then every reference in the demo library has had a cross-family audit, and I've audited the distillations the apps ship as well. The receipts are [in the repository](https://github.com/chrisgagne/grounded-forge/blob/main/corpus.commons/demo/references/_audit/_independent_pass_I_2026-08-08_summary.md).

### Finding the constraint

Can he find the real bottleneck, or does he arrive with a framework?

#### Tell me about a bottleneck that wasn't where leadership thought it was.

At Episource in 2023, after the acquisition by UnitedHealth Group, the parent company pressed for cloud resources to be tagged and patched quickly, and the work was stalling. The read I was given was that the teams were resisting it. The tracker said something else. We had two to three times as much work in process as those teams could carry, and this was the fourth "hackathon" in a year called to catch up on security work, each one landing on the teams as a forced interrupt. The work had never been prioritised against everything else, so it arrived as a crisis on top of too much in flight. I proposed that the requesting team own a proper epic, that we create a formal Expedite class of service for work like this, and that every expedite get an after-action review. The CPTO agreed, managed expectations with the parent company himself, and put the patching through our weighted-shortest-job-first prioritisation instead of expediting it. I added the Expedite class to our planning guidance, and work in process came down afterwards.

#### Tell me about putting a dollar figure on delay. Where did the number come from, and did finance accept it?

Most portfolios I meet don't have believable cost-of-delay numbers: two or three epics carry a dollar figure somebody fought about, and the rest is guesswork. So I built Delivery Intelligence to take the number from the people asking for the work. They state the value they expect from an epic, in dollars; the tool counts each story's share as it completes and can then put an estimated figure on a week of delay. The number is theirs to defend, not mine. At one client the business stakeholders didn't make the time to work out what their requests were worth, so no figure reached finance, and the order at the top mostly came down to executive judgement and external commitments. What I could give the executives was a single ranked view across every team's backlog, so each call was made with everything in sight. At NRC Health the dollar data was thin too, which is why we ranked the portfolio with forced-choice paired comparison. What I changed: in my diagnostic, the value of one live initiative is worked out with its sponsor and someone from finance in the room, not left as a field to fill in, so the number survives their scrutiny.

#### Our board asks what the AI spend bought. How would you answer that?

With what your own data can support, before next year's number is locked. Few organisations captured a baseline before the tools arrived, and usage counts aren't value, so I start from where the time goes now: how long work waits for review, for a decision, for release, and how long each queue would take to clear at the current rate. Then I put a dollar figure on the delay for one live initiative, built with your finance function, and design one experiment that changes where that work waits: if we change this, we expect that by a date, and here is what would prove it wrong. The board gets one page: where the constraint sits now, what it costs, and the decision needed, with an owner and a date. That shows whether the faster coding is reaching customers and what to change so more of it does. My tools measure flow and the cost of delay; they don't split results by AI use, so I won't hand a board a cleaner attribution than the data allows.

#### Tell me about a change that slowed delivery before it sped up.

At a healthcare startup that was scaling fast, people each owned their own area and hit fixed dates with fixed scope through end-of-deadline crunches and weekend work. We moved to cross-functional teams with a Definition of Done, then to LeSS. Feature output slowed, and the executives noticed. But the old speed wasn't sustainable: the teams couldn't ship at that pace with enough quality, in a product where a defect could put a patient at risk, and the code was heading for the point where it couldn't be maintained. We chose quality. The teams got a much better handle on what they were building and when they could release it, and quality improved notably; someone on the team put it to me as things getting slower before they get faster. Feature output was still below the old pace when I left, and the executives saw that more clearly than the gains inside the work. Now I name the dip at the start of an engagement and agree with the executive above my sponsor what they'll watch while it lasts.

#### When would you conclude our organisation isn't the problem?

I carry one prior: AI's gains tend to land inside delivery structures designed before these tools existed. So the diagnosis has to be able to kill it. If a cheap change to a message, a meeting or a rule closes the gap, there was no structural problem to diagnose. If your own numbers show short queues, review keeping pace and dates holding, the tools and the craft are doing their job, and I'd tell you to keep going. A small team automating its own work often never hits the organisational constraint at all; it shows up where work crosses between teams, value streams and approvals. When I tested my own argument about AI transformation, I registered a prediction before classifying the written charters of 50 AI-transformation roles: at least 70% would be scoped to the AI function alone. It came in at 64%, and I published the miss.

### Leading without authority

Can he move things he doesn't own?

#### Tell me about a needed change that sat outside your sponsor's remit.

At one client, the conditions my sponsor, the CTO, couldn't change sat above him: fixed-date commercial promises and company-wide rewards. I couldn't change them either, so I made the trade-offs visible to the people who could. Demand already had an owner outside engineering, who co-owned the department backlog with the CTO, and the operating rules put each trade-off in front of them: when unplanned work would push an iteration past what the team could deliver, planned work had to come out, with a comment on both stories saying what changed and why, so scope stopped growing silently. An incident review I co-authored traced a launch failure to two teams colliding in one release, and the CTO made it a rule that launches skip the regular release unless a C-level executive approves an exception. I also linked retrospectives upward, from team to value stream to department, so each finding reached someone who could change the condition, and where a decision still stalled, I recorded that it had. Now I settle it at contracting: any change beyond my sponsor's remit gets its owner named before it's promised, and the executive above my sponsor takes part.

#### How did you get 30+ stakeholders to agree priorities?

At NRC Health the whole organisation's epic portfolio needed one order, and more than 30 stakeholders had a say, C-level leaders among them. I didn't want 30 people arguing over one ranked list, so I ran forced-choice paired comparisons. People compared epics two at a time, strongly or moderately preferring one, voting with paper cards so nobody anchored on the boss's vote, in rounds where the room discussed the pairs it split on. Engineers sized the epics separately, so the order weighed value against effort, the way weighted shortest job first does, and a ranking model reconciled the places where preferences went round in circles. The final sequence rested heavily on those results. Leadership could see what mattered, work in process fell, throughput rose, and we realised value sooner.

#### Tell me about someone you coached who disagreed with you and was right.

A leader I was mentoring told me a date was achievable when the forecast from my own tool put it at risk. I argued the point with him. He went ahead on his own judgement, and by our next session the work was in much better shape than the report had shown. He was right about the date, and I told him so: the forecast's date ranges came from the teams' past velocity, and he knew things about the work that the tracker didn't. Part of what he knew was how hard the teams would push. They made it partly by cutting corners and letting quality issues slide, at a pace they couldn't have kept up. So we were each right about something different: he knew what the teams could do once, and the forecast showed what they could sustain. I treat a forecast as a question to put to the people closest to the work, and their answer as a question about what it will cost.

### Proof

Can he show results, and is he straight about what he can't show?

#### What's your most measurable result of the last three years, and how was it measured?

From my most recent client, the clearest result is in the CTO's own words: Christophe Louvion wrote that AAR Maker, the LLM-based review coach I built for his organisation, "conservatively saves hundreds of engineering hours a year". That's his estimate, not a time study. The most measurable result I can show you myself is in my own open-source work: the re-audit of grounded-forge's reference library. 9,364 claims were checked against their sources by a model that hadn't written them, 97.45% were free of hard errors before repair, and all 446 findings, hard errors and minor drift such as a dropped qualifier, were re-checked and repaired, with the receipts in the repository. At NRC Health, after the portfolio reset, work in process fell, throughput rose, and we realised value sooner. We didn't run it as a metric study; what we saw was in the tracker: once lower-ranked epics were stopped or cut, work in process went down almost overnight, and the teams started finishing work fast.

#### Tell me about an engagement that fell short of the sponsor's hopes.

In one way, all of them did, and for the same reason, which I share with most coaches: I was contracted to work on delivery, and the change each organisation needed also sat with its executive leaders: what the organisation was optimising for, its strategy, and the beliefs behind both. One engagement shows how that plays out. The instruments I built were meant to outlast me, and Delivery Intelligence only tells the truth if the tracker data is kept clean. I kept it clean from the first day, and by the end I'd become the single point of failure for the instrument I built, because I hadn't handed enough of that work over before I left. So I now contract for the executive work explicitly: time with the executive leaders on what the organisation is optimising for, with the executive above my sponsor taking part, and every practice I install has a named internal owner before I call it installed.

### The Agile history

Ten years of Agile coach titles: why should that make him better at AI rollouts?

#### Tell me about a failed Agile rollout, and what it teaches about AI rollouts.

The one I learned most from was small. A team I coached agreed its Definition of Done, I trained them on it, the exercises went well, and every sprint, when the review loomed and the demo had to land, the Definition of Done was the first thing to go. I've taught a couple of dozen two-day Agile courses, adapted hard to each organisation, and there were times I felt close to gaslighting people: the training pushed people towards one behaviour, and the conditions back on the job punished it. Most of the rollouts I saw in ten years as an Agile coach pushed people and left the conditions alone, so teams sped up while the business stayed where it was. AI rollouts are being run the same way. Mandates, training and usage targets push people, while the review queues, approvals and handoffs between teams stay as they were, so individual output rises and the work waits. That's why I start from the conditions: where the work waits in your own data.

#### Have you delivered hands-on, or only advised?

Both. The hands-on part: at my most recent client I wrote and shipped three tools into a 25+ team organisation, and only one needed an LLM. I was the sole practitioner there for 15 months: I designed the operating handbook the organisation ran on (now public as *Minimum Viable Bureaucracy*) and kept the tracker data accurate so the tools could be trusted, while mentoring the CTO and product leaders and coaching VP-level leaders. The CTO held the authority for the changes inside technology. At NRC Health I ran the prioritisation sessions myself. At Episource I administered our Shortcut workspace and facilitated incident after-action reviews in person, as a neutral facilitator. Before consulting I did the work itself, as product lead at StubHub and PagerDuty, and before that as a co-founder writing the Perl and MySQL for our own marketplace.

### Remote

Can he work with a US team from New Zealand?

#### How have you worked with a US team from Nelson? What broke, and how did you fix it?

I've worked remote-first since 2019 and from Nelson since 2020. Nelson sits roughly halfway between California and India, which suited Episource: I could hold a regular day overlapping both its US leaders and its teams in India, and I travelled to Chennai in 2022. My most recent engagement ran Tuesday to Saturday New Zealand time, which lands Monday to Friday afternoons US Pacific. What broke was anything that needed people in the US, Asia and Europe on a call at the same time, which was hard because of where the client's people were, from any time zone. Incident reviews were the worst case, so I moved the most time-consuming part, building the timeline, into a shared document that people filled in asynchronously before we met, and kept the live session for the conversation only. Video meetings broke in a smaller way at Holo.Host, a globally distributed company, where people talked over each other; we used a small set of hand signals for turn-taking and decisions, which I [wrote up in 2020](https://www.linkedin.com/pulse/several-simple-hand-signals-video-based-collaboration-chris). At my most recent client, the tools I built did some of the asynchronous work, so the coaching reached people in time zones where I was asleep.

### Integrity

Will he tell us what we don't want to hear, including about hiring him?

#### Tell me about a time you told a client to stop something, or not to hire you.

At Episource I managed a team of Agile coaches, and in 2023 there was an open requisition for another Enterprise Coach. The leader it was meant to support didn't want a coach, and we were struggling to fill the role, so I proposed cancelling it, which cut my own team's headcount. The same year I asked the organisation to stop calling its security catch-up sessions "hackathons", because they behaved like forced interrupts that kept breaking planned work. When Student of Fortune, the company I co-founded, drifted towards "eBay-for-homework", I left in 2009 rather than keep building it. And if a seat's centre of gravity is building models, I'm the wrong hire; if the people who set funding and rewards won't change them, I'd tell you not to buy embedded work from me.

## Questions people ask about the record

### Did you found and sell a company?

I co-founded Student of Fortune in 2005 and ran it as COO. I left in January 2009 and sold my stake back to the company in late 2010. Chegg acquired the company in August 2011, after I'd left, so I don't claim that sale.

### What was your part in the patent? Is OpsDex AI?

I'm the lead inventor, the first-named of four, on US 9,367,810 B1 (2016). My colleagues had the idea of an operations-maturity index; I devised and ran the regressions on incident data across PagerDuty's customer base that made a working one. OpsDex is analytics and a patented data product, not AI.

### How long have you worked with AI?

Production machine learning at StubHub in 2013, as product lead for a recommendations engine a three-developer ML and data-science team built. Then no AI work from 2014 to 2022. Hands-on generative AI since September 2022, LLMs since 2023, and AI tooling and workflows in my consultancy since 2023 (the consultancy itself started in 2020).

### Are your tools used by several clients?

The three client tools (AAR Maker, Delivery Intelligence and the backlog data-quality tool) were built for one client, my most recent engagement, and shipped into it. Only AAR Maker uses an LLM. My consulting engagements number four; the tools belong to one of them. grounded-forge and Gap Scan are public.

### Has AAR Maker ever produced a wrong or harmful review?

None that I know of: the teams read every review before they submitted it. Its harder limit was what its reviews found: they often pointed to systemic causes, such as fixed dates, growing scope and teams already stretched thin, that even the CTO couldn't fix alone. A review can name the cause; changing it takes whoever owns it.

### Have you held a budget and a team, or only advised?

Both, though the line roles are older. As COO and co-founder of Student of Fortune (2005–2009) I held the company's budget and owned trust and safety, payments and disputes. At Leads360 I directed account management and technical support through 45× revenue growth. At Episource five Agile coaches reported to me: I owned their development and reviews, sourced and hired one of them, and in 2023 proposed cancelling an open requisition the leader it served didn't want. Since 2020 my work has been advisory and embedded: the sponsor holds the authority, and I name the owner of any change beyond it.

### Have you run a centre of excellence or a community of practice?

Not for AI. The nearest: at Episource my team of five Agile coaches served a 300+ person division of UnitedHealth Group and ran office hours and an Agile book club alongside the coaching; at a healthcare startup I built reading clubs and a community of practice, and senior executives sat the LeSS course; at StubHub I was on the Core Agile team chartered to improve how the company worked. I also led the move to Shortcut, and administered it, at three clients: Episource, NRC Health and the financial-services firm.

### What lasted after you left?

The people, the practices and one of the tools. The practices travelled with me: my operating handbook, a bug-and-risk prioritisation framework and after-action review templates ran at Episource, NRC Health and the financial-services firm, and at the last of those the review templates became AAR Maker, which is still in use there: the teams run most of their reviews through it. The handbook is public as *Minimum Viable Bureaucracy: A Field Guide*. As for people, at Episource I hired Williams Fuentes as a senior team coach, and he's now a senior staff Agile coach at Optum; Wilson G, a team coach on my team for over two years, is a technical program manager there. At a healthcare startup I interviewed and hired its first two Scrum Masters, Han Liu and Amanda Tjen, and mentored both; Han is now a Scrum Master at the NRMA and Amanda a senior delivery manager at Woolworths Group.

### What's your experience with security, compliance and AI governance?

Practical rather than framework-led. I've worked inside regulated organisations: HIPAA-regulated healthcare at Episource, through its integration into UnitedHealth Group's Optum, where I set up an Expedite class of service so the parent company's cloud security work stopped stalling; and a global financial-services firm, where AAR Maker ran inside the firm's own environment on its own model-API keys. I was product lead for incident management at PagerDuty. In my own AI systems the controls live in code: agents can't send email, touch LinkedIn or approve their own work, approvals come only from my own user, and every claim checks against a facts ledger; grounded-forge makes a fresh-context audit of every claim mandatory. I haven't implemented a formal AI governance framework such as the NIST AI RMF or ISO 42001.

### Have you trained non-technical audiences?

Mostly I've trained engineering and product people. Some of my Agile mindset sessions were for non-technical audiences: two day-long Agile Labs for staff at a New Zealand Crown entity, in Wellington (2020) and Auckland (2021); a workshop for a startup in Blenheim; a talk for BusinessCentral's 2020 Business Bootcamp; and "Mindfulness and the Modern Workplace", given as a talk at the Edmund Hillary Fellowship's New Frontiers summit (2019) and as a public workshop at a digital agency in San Francisco (2019).

### Can you work across the whole organisation, not just engineering?

Yes, and most of the evidence predates AI. At the financial-services firm, the CMO co-owned the department backlog with the CTO, so demand from marketing entered the same ranked view as engineering's. At NRC Health the portfolio ranking took in the whole product organisation and many stakeholders beyond it, C-level leaders among them. At a healthcare startup I worked with the heads of product, design, clinical, engineering and HR; clinicians joined every Scrum team, and HR helped change career progression and incentives. Before coaching, I ran client services at Leads360, ran operations and trust and safety as a marketplace's COO, and worked as an operations supervisor at McMaster-Carr. Two of the recommendations above come from outside engineering: a VP of commercial solution development and an EVP of human resources. My developmental-coaching training is what I bring to the people side of that work: helping leaders whose priorities differ from engineering's work out what each needs from a change.

### Have you used AI outside software teams?

In my own work, yes. My research for *Come Prepared to Die* runs on grounded-forge, which checks every claim against its source, and my sales pipeline runs on agents that research and draft while I decide and send. I'm the only user of both. The AI tool I shipped into a client served its technology organisation; I haven't yet led AI adoption across a company's non-technical departments.

### How big was the financial-services engagement, and why isn't the client named?

Three value streams planning together fortnightly across 25+ teams (25 or 26), each keeping its own backlog, with the CTO holding the view across all three, for 15 months. I was the sole practitioner. My contract keeps the client's name confidential, so I describe it by type: a global financial-services firm.

### Why do older records call you an Agile coach?

From 2016 to 2026 my titles were Agile coach titles, first at Accenture | SolutionsIQ and Holo.Host, then for my own practice, which older records call "Founder and Principal, Enterprise Agile Coach". The consultancy role is the same one I now title Principal Consultant, Product and Operating Model. My title at Accenture was Agile Coach (senior Agile coach).

### What is the coaching training for?

I'm a developmental coach, trained in Alētheia, Co-Active and Motivational Interviewing, and I'm training in Radix Body Centered Psychotherapy without having completed its certification. I work with leaders, not patients, and I refer anything clinical to a licensed counsellor.

### Which of your certifications are current?

Certified Org Topologies Consultant, Certified LeSS Practitioner, ICAgile ICP-ACC and ICP-ATF, and Pragmatic Framework Certified III. My SAFe Program Consultant and Scrum Alliance certifications have lapsed, so I list them as training.

### Are you still part of the Edmund Hillary Fellowship?

Yes. I'm an Edmund Hillary Fellow. The Fellowship organisation closed in July 2026, but the community of Fellows carries on, and I'm still connected to its people.

## Public evidence

Sources a reader can open, as my record of facts lists them.

| Claim | Link |
|---|---|
| Lead inventor (first-named of four), US 9,367,810 B1 (2016) | [Google Patents](https://patents.google.com/patent/US9367810B1/en) |
| Student of Fortune co-founder | [ABC News, August 2008](https://abcnews.com/amp/Business/PersonalFinance/story?id=5648319&page=1) (the AMP view; the ordinary page omits the section) |
| Chegg acquired Student of Fortune, 2011 (after Chris left in 2009) | [Chegg 2013 10-K, Note 6](https://www.sec.gov/Archives/edgar/data/1364954/000119312514086390/d645280d10k.htm) |
| Momtastic launch, 2010 | [Cynopsis, 2 September 2010](https://www.cynopsis.com/thursday-september-2nd-2010-2/) |
| LeadGuardian | [Jeff Solomon, 8 October 2006](https://medium.com/pitashi/lead-theft-you-either-know-it-or-you-dont-f11c5f44b9b2) |
| Colleague and client testimonials | [About page](https://chrisgagne.com/about/) |
| Holo remote coaching | [Hand signals for video collaboration, March 2020](https://www.linkedin.com/pulse/several-simple-hand-signals-video-based-collaboration-chris) |
| Member of the winning Silent Lens team, Google Develop for Good, 2012 | [Google's announcement](https://developers.googleblog.com/winners-of-the-google-io-develop-for-good-global-hackathon/) |
| Edmund Hillary Fellow | [EHF, 17 May 2023](https://www.ehf.org/news-blog/ehfs-final-welcome-experience-sees-60-fellows-welcomed-to-the-fellowship-aotearoa) |
| Org Topologies certified | [Member directory](https://www.orgtopologies.com/member-directory) |
| Founder, Bay Area Agile Potluck | [Meetup](https://www.meetup.com/the-bay-area-agile-potluck-series/) |
| Public software | [grounded-forge](https://github.com/chrisgagne/grounded-forge) · [Gap Scan](https://github.com/chrisgagne/gap-scan) ([live demo](https://chrisgagne.github.io/gap-scan/)) |
| freeCodeCamp Top Contributor, 2018 | [Announcement](https://forum.freecodecamp.org/t/announcing-our-freecodecamp-2018-top-contributor-award-winners/201353) |
| Tradeoff Matrix training video, published by freeCodeCamp, 2017 | [YouTube](https://www.youtube.com/watch?v=8WpercDYOAA) |
| Diagram reused and credited in a PMI NYC deck, 2016 | [Deck, page 20](https://pminyc.org/static/uploaded/Files/Documents/2016_05_18_pmi_nyc_presentation.pdf) |
| Product interviews, 2017 | [Building The Right Product](https://www.youtube.com/watch?v=0a0SKGtMJ-w) · [On Not Saying No In Agile](https://www.youtube.com/watch?v=3VjoqMBymDc) |
| Recorded interview for UCL's Eco-Innovation module, 2022 | [David Bent's episode](https://www.buzzsprout.com/1992822/episodes/10934376-7-chris-gagne) |
| Co-facilitated the Future of Work Agile Lab, July 2020 | [Humanitix listing](https://events.humanitix.com/future-of-work-agile-lab) |
| Moderated two EHF generative-AI panels, 2023 (no public page; off CVs) | [NZ impact](https://www.youtube.com/watch?v=bYt3Q3_Ud4Y) · [Gaming and startups](https://www.youtube.com/watch?v=BzMc_GVDG4s) |