For founders, CTOs & tech leads · Fixed-price · 100% remote
That's what an agent-driven delivery looks like in 2026 — not slides, not POCs, not "AI-augmented dev". Real production code, shipped on a tight timeline, by one engineer running a fleet of agents.
I'm Loïc Ponchon. 14 years of fullstack engineering as the foundation, full-time agentic workflow since 2024. Fixed-price engagements with direct decision-makers.
Selected clients & ventures
Two profiles I work best with. If you recognize yours, we'll probably click on the first call.
You have a feature backlog, an MVP idea, or an AI angle you want to validate. You don't want to hire a full team for it. You want it shipped, measured, and out the door.
What I deliver here: MVPs, AI features in production, internal tools, automation pipelines.
You have a .NET application — back-office, business platform, internal tool — that's slowing the team down. Refactor or full rebuild has been on the roadmap for quarters. Classical estimates scare the budget away.
What I deliver here: legacy rebuild (WebForm, MVC monolith → .NET 9 + React 19), agentic CI/CD setup, E2E test coverage, team handover.
If you've ever tried to rebuild a legacy app, ship an AI feature, or get an MVP out the door — you know the real cost is rarely the code itself. It's the time.
12-18 months estimated, multiple devs needed, scope drift, regression risk on every push.
2 months delivered solo with a complete Playwright E2E suite running in parallel. (See E-TIPI case below.)
Endless POC loops, LangChain stack overkill, no measurable ROI, project shelved after 6 weeks.
Minimal stack on real data from day 1, scoped fixed-price package, measurable output in 10-20 days.
Wireframes for weeks, hire a team, miss the window, MVP ships when the market has already moved.
Direct from brief to working prototype, scoped in weeks not months, you own the codebase.
Manual reviews bottleneck the team, flaky tests ignored, regressions slip through, ops burden grows.
Agentic pipelines: lint, test, smoke, regression checks with auto-fix loops on common failures.
The point isn't "AI replaces developers". It's that with 14 years of senior fullstack judgment running the show, agents eat the boilerplate so the work that ships per day multiplies.
Want this on your project?Featured case study · 2026
Returned to E-TIPI (LMS B2B international) in 2026 as Tech Lead Freelance, seven years after the end of my CTO mandate (2018-2025). The mission: rebuild the legacy back-office — WebForm + MVC monolith, dispersed business logic, custom DAL — into a clean .NET 9 API (Clean Architecture / DDD, CQRS + MediatR) + React 19 stack, migrated progressively with the Strangler pattern (façades + feature flags, no prod breakage), on a multi-tenant LMS serving major accounts across banking, food-service and heritage / foundation sectors.
12 to 18 months with a small team. The kind of project that often gets shelved because the cost doesn't match the business case.
500 pages delivered in 2 months. Code-review-fix cycle reduced by about 70% on this kind of bulk modernization work. Internal team rebuilt and onboarded to own the next phase.
Four engagement types. Defined scope, measurable output, direct decision-maker. The agentic workflow is the common denominator.
Refactor or full rebuild of legacy apps into modern stacks.
From brief to production. SaaS MVPs, BO custom, business tools.
Production-grade AI features that ship and measure ROI.
CI/CD piloted by agents. Lint, test, smoke, auto-fix.
No 6-week sales cycle. Most projects close in 2 calls.
You describe the project: context, goal, current pain. I ask the right questions to scope it. If it's not a fit, I say so upfront and redirect.
Written proposal: scope, deliverables, timeline, fixed price. You either accept, push back, or pass. Zero ambiguity.
I ship in short feedback loops with weekly sync. You see the code, you see the tests, you see the demo. No "trust me bro".
If the first delivered milestone doesn't match the agreed scope, you don't pay for it. Plain and simple. I bet on the agentic workflow — the risk is mine, not yours.
B2B SaaS for cartographic prospection. 4 million French companies on an interactive map, filterable by industry code, headcount, GPS radius. AI scoring, integrated kanban pipeline. 70+ Playwright E2E tests green.
AI-for-ops consultancy for SMBs. Custom GPTs (Storytelling, Prospecting, Content Strategy) deployed at client sites. +300% measured content output on first pilots.
Tech lead of the LMS B2B international platform. Modernization path WebForms → MVC → React + API-first. DDD architecture, GitLab CI/CD, daily production ownership.
API-first .NET 8 services with strong focus on maintainability. Dynamic React UIs. OpenAI integration with structured outputs and governance.
Enterprise expense management platform on SOA for a Private Equity fund. ~50 internal users. SLA held for the full duration in autonomy.
Internationalized the B2B platform and CMS workflow. Belgian B2B website ownership. SQL-to-Cassandra data migration.
4.8 / 5 on Malt across 11 verified reviews. Quality 4.8 · Deadline 4.9 · Communication 4.9.
"Loïc has done excellent work. He showed real adaptability and was a precious help in turning our project into reality. Thank you Loïc!"
"Always very happy to work with Loïc. Clean, organized, fast and efficient work. I strongly recommend him… but for now, I'm keeping him for myself ;)"
"Loïc fully invested himself in this small development, asked the right questions, and ultimately delivered high-quality code using libraries and patterns already in place perfectly."
"Short mission but very well executed. Loïc quickly understands the need, is reactive and proactive with suggestions. Until soon for another project."
Being honest upfront saves us both a discovery call. Here's when we shouldn't work together.
I focus on short fixed-price engagements where I can deliver fast and move on. For long-term embedded positions, you're better served by an agency or a hire.
The agentic workflow doesn't map to hours. I sell outcomes — clear scope, fixed price, weekly demos. Hourly tracking would just slow us both down.
I move best with one direct decision-maker on the other side. Committees and validation chains kill the velocity you'd be paying me for.
I don't sell slides. I sell shipped code. If you need strategic deliverables (board decks, architecture whitepapers) without delivery attached, I'm not the profile.
If you recognized your situation in any of these, I'd rather you find someone who fits better — and I'm happy to redirect you in our discovery call.
Because the agentic workflow changes the time/output ratio. Paying me by the day would mean you're paying for hours not shipped. Fixed-price means we agree on a scope and an outcome, and the price reflects the value — not the time logged. A daily rate (€800) is available as a bridge if your procurement requires it.
Fair question. In my case it means a concrete daily practice: Claude Code in the CLI, autonomous agents running tests and CI/CD, a dedicated AI dev machine with background agents, custom MCP servers. The 500 pages in 2 months at E-TIPI is the most recent measurable proof.
Most fixed-price packages I run sit between 5 and 60 days of work (€5k to €60k). Smaller than that, a discovery call usually shows we can scope it down to a focused one-week sprint. Bigger than that, I propose a phased approach with clear milestones.
Yes. Full IP transfer, clean repo, documented codebase, CI/CD set up. Your team can take over at any milestone.
Mainly .NET 8/9 (Clean Architecture / DDD, CQRS + MediatR, EF Core), React 19, TypeScript, Node.js, Python (for AI). PostgreSQL, SQL Server / DacPac. CI/CD (GitLab, GitHub Actions), OpenTelemetry observability, Playwright E2E. Claude / OpenAI / Anthropic Agent SDK / MCP for the AI layer, including AI chat assistants wired to your business APIs. Comfortable on legacy too (Strangler-pattern migrations without breaking prod). If your stack is different, ask — the agentic workflow adapts fast.
Within 24 business hours. You'll get an honest take — feasible, scope, ballpark, or a redirection if it's not a fit.
Day 1: kick-off call, repo access, scope reconfirm. Days 2-4: first vertical slice delivered (one full feature end-to-end, including tests). Day 5: demo + next-week planning. From there, weekly demos until done.
Yes, mutual NDA upfront if you need it. Standard practice for sensitive scopes.
Two paths to start. Both lead to the same place: a 30-minute discovery call where I tell you honestly if we should work together.