An AI-native approach for Product organisations: structure your
department, up-skill your teams, and build the AI framework that
becomes your unfair advantage.
Create or restructure your Product Department
The pain
Missed opportunities, deliveries pushed back, no measurable ROI,
lack of alignment — leadership feels the symptoms without being
able to name the cause.
The approach
I start from your reality, not an org chart template: assessment,
target design, transition plan — and I install it, as interim CPO or
alongside your leadership team.
The outcome
A department leadership can trust to run without them — and an
organisation that focuses on delivering value that serves your business.
Four signals you need to refactor your Product org
Signal 01
Product, Design and Engineering ship features, not value.
Symptoms
Wrong problems addressed, solutions badly executed, ROI you can't measure.
Cost
Product effort doesn't convert to business impact.
Signal 02
Marketing and Sales are in the dark.
Symptoms
Nobody can clearly explain what's shipping or why it matters and creating the sales and marketing narrative is painful.
Your roadmaps are unreliable and your vision is limited.
Symptoms
Deliveries are always delayed without rationale for you to course correct and your vision is not anchored in your reality.
Cost
Confidence is eroded and you do not convert these key logos.
Signal 04
You've crossed a growth threshold.
Symptoms
What worked at 15 people quietly stops working at 50, and again at 150. Informal alignment no longer scales: more meetings, less clarity, decisions that stall between teams.
Cost
Headcount grows faster than output — the organisation slows down precisely when it should accelerate.
Julien joined us at a pivotal moment to lead the transformation of our
product development approach. With agility and vision, he embedded best
practices that allowed us to harness our full potential and, as a
result, deliver two major innovations on time and with excellence.
Thomas Cottereau — CEO at SightCall
Up-skill your team and implement best practices
The pain
Every PM works their own way. Deliverables that can't be compared,
prioritisation that runs on opinion, quality that depends on who did
the work.
The approach
I coach your team to master the key principles behind an efficient
Product Org, anchor it in a complete Product Development Life Cycle
and install the rituals that keep it transparent.
The outcome
Work that's comparable and predictable: faster, defensible decisions,
onboarding in weeks — and a focus on delivering value with a high
level of confidence.
The Product life cycle
Six steps, from insight to revenue, that speed up value delivery and
aligns every team involved, without slowing you down.
1Qualify2Frame3Define4Deliver5Enable6Learn
1
QualifyCapture and qualify the signals — customers, support, sales, data — always against the same criteria.
2
FrameFrame the problem without talking solution. For whom? And if we solve it, what's the impact on the business?
3
DefineDefine the solution as a triad: PMs own viability, Designers usability, Engineers feasibility — together, desirability.
4
DeliverFunctional unit of value after functional unit of value, until the acceptance criteria are met.
5
EnableBring Marketing, Sales and Support up to speed on the logical chain: problem → solution → customer benefit.
6
LearnThe step too often skipped, yet the most important: did the expected business impact materialise? Iterate, kill, or leave it there.
Principles coached
Problem before solution. Teams frame and validate a problem before working on solutions.
Culture of insights. Qualitative or quantitative insight precedes every prioritisation decision.
Impact & success. Every initiative starts with: what is the impact on our business? And what does success look like?
Trust & commitment. Commitment follows confidence, not external pressure.
Triad collaboration. PMs own viability risk. Designers own usability risk. Engineers own feasibility risk.
Granularity. Smallest problem, smallest functional unit of value.
Rituals installed
Bi-weekly sprint defence
Monthly roadmap review
PM / CSM bi-weekly
Triad sessions
Quarterly prioritisation cycle
Bi-weekly agile re-prioritisation
Julien clearly understood our needs, provided clear guidance, and was
readily available to each Product Manager. For some, his contribution was
particularly significant in initiating the right thinking associated with
each phase of our product development cycle.
Claire Idelot — CPO at Ilek
Deploy your Product agentic workforce
The pain
Your AI champions have built great agents, skills and workflows. But ways of working aren't harmonised, and hand-offs at the seams cost some or all of the gains.
The approach
I work alongside your champions to split your lifecycle into async agent workflows, human gates and synced Human x AI collaboration. Then we deploy it.
The outcome
Your champions' practices, scaled to the whole team: agents execute around the clock, humans decide at the gates, and gains survive the hand-offs.
The whole system, at a glance
Async workflows prepare the work. Human validations lock the decisions. In live session, judgment stays with the team.
Agents execute. Humans decide. Every step combines asynchronous agent workflows, human validations, and human × AI sync work. Click the labels to see their definitions.
As soon as a sub-step has clear source material, a deliverable format and a definition of the quality to ship, agents can run it on their own.
1QualifyStep 1 / 6
Signal intake: Agents pull signals from support, sales calls, satisfaction forms and customer reviews, then cluster them, deduplicate them and sort them.
Signal scoring: Agents draft a score against your selection criteria, and give you a prioritised list of signals by potential impact.
2FrameStep 2 / 6
Insight centralisation: Agents gather everything already known about a problem before discovery starts.
Interview processing: Agents transcribe interviews and extract a first pass of insights, each linked to its source.
Hypothesis drafting: Agents draft problem hypotheses from the insights, in your usual format.
3DefineStep 3 / 6
Solutioning: Agents propose several solutions and spell out the trade-offs against your non-negotiables.
Specs and prototypes: Agents draft specs in your usual format, and prototypes to test the key risks.
Test and measurement plan: Agents prepare the test plans and the metrics that will show whether the change delivered on its promise.
4DeliverStep 4 / 6
Building delivery increments: Agents break the solution into publishable units of functional value that can ship on their own: the increments.
Progress tracking: Progress tracking is formalised automatically. Nobody has to chase anyone.
Acceptance criteria: Agents check that each increment meets the definition of done, ready for review.
5EnableStep 5 / 6
Release notes: Agents write the release notes from the increment history.
Marketing and Sales enablement: Agents prepare the documentation, the marketing and sales briefs, and the translations.
6LearnStep 6 / 6
Tracking and monitoring: Agents track success metrics and flag gaps and anomalies.
Lessons learned: Agents consolidate the learnings and propose a next step: continue, iterate or kill.
Mode 02 · Always in control
Human validations
Your teams validate the agents’ work. These validation rules can be optional on simple topics, but stay mandatory on complex ones.
1QualifyStep 1 / 6
Signal qualification: The AI proposes whether certain signals should enter the product cycle at a given moment. You accept, amend or refuse.
2FrameStep 2 / 6
Insight identification: Against your research questions, the AI finds relevant insights in interview transcripts, market research documents or competitive studies. You accept, challenge or refuse them.
Hypothesis formulation: The AI drafts problem hypotheses (with business impact) or solution hypotheses (with a product success measure), standardised and ready to be validated, amended or refused.
3DefineStep 3 / 6
Solution choice: You make the final selection among the solution options on the usual risk criteria (viability, desirability, usability and feasibility, including effort).
Spec review: You give a go / no-go on the spec and/or the implementation plan and the other deliverables that make up the definition of done.
4DeliverStep 4 / 6
Release plan: Partial or full, you validate the release plan with its steps, responsibilities and deliverables. Nothing ships to production without a teammate’s validation.
Increment management: Delivery increments are presented against your definition of done. You only have to accept, or not, the acceptance criteria.
5EnableStep 5 / 6
Documentation publication: Support articles, video walk-throughs, feature-benefit-problem-persona mapping, sales presentation slides, website updates. Your definition of “enabled” is waiting for validation.
6LearnStep 6 / 6
Impact review: Automated feedback on your impact and success measures, with proposed next steps (iterate, keep, kill), waiting for your decision.
Mode 03 · Human / AI collaboration
Synchronous human and AI collaboration
Your teams collaborate with their AIs in sync, in the chat window. The AI executes and the collaborator exercises critical judgment when needed.
1QualifyStep 1 / 6
Ambiguous signals: You work in session the political or undecidable signals: the founder’s pet idea, big-client demands, anything a score cannot settle.
Recalibrating selection criteria: You revisit the scales and selection rules themselves on a regular cadence.
2FrameStep 2 / 6
Weighing the evidence: You resolve contradictory signals, choose which problem to bet on, and size the business impact.
3DefineStep 3 / 6
Scope: in and out: You decide what will be part of the solution and what will be left out.
4DeliverStep 4 / 6
Refinement and feedback: You work in sessions to settle the conflicts that appear, answer new questions, and bring back feedback gathered outside the workflows so you can update the context, the deliverables or the decisions as new events land.
5EnableStep 5 / 6
Message and packaging: You decide how to talk about the offer and how to present it.
6LearnStep 6 / 6
Retro and next steps: You draw the lessons from the cycle, and you decide whether to keep going as-is, evolve, or kill.
Julien produced high quality work supported by a well thought-out
methodology and delivered with an amazing humble, team-player attitude.
Martin Pannier — CEO at Primo
Connect with me on LinkedIn
Follow along for the methodology in practice — and how AI is
rewriting the way teams work in 2026.