Full-lifecycle agentic delivery for French finance and accounting firms

The chapter: the frontier

Tanbits is a 120-person software house with offices in Lahore and Doha and a Boston association. I direct roughly 80 engineers through their team leads, across every web and mobile discipline, and own client handling, delivery, approvals, product decisions and market research alongside the directors. Second-line leadership: I manage the leads who manage the teams.

Over the past year the whole pipeline has run agentically. PRDs → AI development → AI Playwright testing → AI CI/CD → UAT → final delivery, on compound engineering with file-based durable state. Not AI tools bolted onto a team, but a development operating model.

The stakes. Velocity is worthless with quality drift. The system exists to make speed safe: PRDs as contracts, reviewer agents, and near-total, CI-enforced AI test suites.

The clients

Primexis is a French financial-management firm providing consolidation and reporting services to large corporate groups and international subsidiaries. Groupe Aplitec is a Paris-based accounting, audit and advisory group. Both run on internal tooling that had accumulated over decades.

The scope

A team of five engineers, led by me at Tanbits, built internal platforms for both firms:

• Automated their legacy systems. Accounts and bookkeeping automation replacing manual, spreadsheet-and-legacy-app workflows.
• AML/CFT compliance checks. Anti-money-laundering and counter-terrorist-financing screening automated in a regulated domain, where a missed check is a regulatory event.
• Automated data updates and event propagation throughout the ecosystem. An event-driven architecture keeping each firm’s internal tools in sync instead of relying on people to re-enter data.
• Government API integrations with French regulatory and administrative APIs.

The delivery model

This account is the proof that the full-lifecycle agentic pipeline works end to end:

PRDs → AI development → AI Playwright testing → AI CI/CD → UAT → final delivery.

Compound engineering throughout: the requirements, decisions and roadmap lived in versioned files, so the work accumulated across weeks instead of resetting with each session. Builder agents implemented features in parallel, reviewer agents validated them, tester agents produced the unit and end-to-end suites, and CI enforced coverage before anything reached UAT.

Why it is the counterpart to Mortgage Automator

Mortgage Automator proved the system is fast. Primexis and Groupe Aplitec prove it is complete and compliant. A small team, a regulated financial domain, European enterprise clients, and every phase of the SDLC run agentically, not just code generation. When someone asks whether AI-orchestrated delivery is only good for prototypes, this is the answer.