Twenty years building and leading products, 20+ products launched as Technical PO or founder. An everyday AI operator, I put agents into production for my clients and for my own tooling. Results are measured in time saved and cost.
I take vertical responsibility for delivery: from market insight to the moment it sells. Market validation (Mom Test, Business Model Canvas, Lean Startup), spec-driven roadmap, architecture, execution and quality governance: one accountable counterpart, committed to the outcome.
Twenty years fully remote, async-first, with distributed international teams. Founder coaching included: I transfer the method rather than create dependency.
I transformed my own software development lifecycle with agentic AI before recommending it to anyone. The method is proven on myself, then applied to your organization.
Process mapping, triangulation of interviews / documents / data, and a prioritized build-vs-buy action plan co-validated with decision-makers. Operated solo and heavily tooled, where a consulting bench takes 8 to 12 weeks. Minimal internal load: one decision-maker and one domain expert per entity.
For teams already using AI daily, but without method or guardrails. Every participant leaves with their own automations and supervised agents in production, built on their real cases. Program (in French) at transformation-ia.warfog.gg.
Framing, team upskilling, solid implementation of critical use cases, and security of delivered implementations (APIs, tokens, OAuth, production robustness). Steer AI without hiring a Chief AI Officer.
Executable specs, TDD, E2E tests, CI/CD, vertical traceability from requirement to release. The Jared pipeline orchestrates Claude agents across the whole cycle: product management, development, QA. How it works is detailed at jared.warfog.gg.
For funded founders without an engineering team: I deliver the outcome fully autonomously, a third way between building a team and steering an agency.
Blocking legacy code, vendor dependency, overpriced SaaS: the source code is the source of truth. I rebuild the specification, then take over, recode or internalize.
A build is not a one-shot. Operations are covered by a maintenance subscription, held to the same standard as the initial build.
Four projects shipped to production: measured velocity gains from ×5 to ×20, costs divided by 3 to 4. Including one delivered in 3 days against a 2-month internal estimate.
Each engagement has its own method; the terms of engagement stay the same.
Fully remote, async-first, with occasional travel. Three parallel engagements at most: availability is part of quality. Start is immediate when a slot is open, within two to four weeks otherwise. And whatever the engagement, AI output goes through the same gates as human-written code: review, automated tests, CI.
Each activity's own method is detailed on its page (in French): the week-by-week AI transformation audit process, the Fractional CPTO scope, the application modernization and POC-to-production approaches and the hands-on training pedagogy.
Very broad functional scope coverage … fully testable.
PO · EdTech, delivered in 3 days
Very clear, and in line with best practices.
PO · EdTech, on the delivered specifications
Recommended on Malt by Philippe Acquier, Salim Laimeche, Nicolas Lapointe and 5 others.
A Fractional CPTO provides a company's product and technical leadership on a part-time basis, without the cost or commitment of a full-time C-level hire. They take vertical responsibility for delivery, from market insight through to go-to-market: market validation, spec-driven roadmap, architecture, execution and quality governance, with a single point of contact accountable for the outcome.
The approach starts with an AI transformation audit: process mapping, triangulation between statements, documents and data, then a prioritised build / buy action plan co-validated with decision-makers. If needed, it continues with a Fractional Head of AI engagement: framing, team upskilling and hardening of critical use cases, without hiring a Chief AI Officer.
Three to four weeks, run solo and tool-assisted, where a team of consultants typically needs 8 to 12 weeks. The internal workload is minimal: one decision-maker and one subject-matter expert per entity involved.
Yes. The hands-on AI training runs over two days remotely, with a maximum of ten participants, and is fundable through French OPCO schemes via a Qualiopi-certified body. Each participant leaves with their own automations and agents supervised in production, built on their real use cases.
Through a retro-spec: the source code is the source of truth. The specification is rebuilt from the existing code, then the software is taken over, recoded or internalised. This addresses situations of blocking legacy code, vendor lock-in or a SaaS that has become too expensive.
Across four projects delivered to production, measured velocity gains range from ×5 to ×20 and delivery costs are cut by 3 to 4, including one engagement delivered in three days where the internal team had estimated two months. Every delivery passes the same gates as human-written code: review, automated tests, CI.