FRDownload CVEmail me

Omar El Hachimi / Projects / LCL GenAI platform

Project 03 · LCL · Crédit Agricole group · Dec 2023 – now

LCL's self-service GenAI platform: 90 teams build their own AI assistants

Omar El Hachimi architected and industrialised LCL's central GenAI platform, where 90 teams and 500+ people create, deploy and operate their own AI assistants on their own documents, under the bank's access, cost and security rules.

Role Architect and technical lead

LCL GenAI platform, how it works: 1 Create, 2 Connect, 3 Control, 4 Answer, 5 Monitor. 90 teams with their own assistants; 500+ users; 6 families of use cases.
LCL GenAI platform: how it works and key figures.
90teams with their own assistants
500+users
6families of use cases

Context

Every team at the bank wanted an assistant that knew its own documentation, tickets and data. Each request was becoming a separate project, with its own security review and its own costs.

The bank needed one safe place where any team could build what it needed, without waiting for a project.

How it works

  1. CreateA team creates its assistant in the web interface.
  2. ConnectIt links Confluence, Jira or a database.
  3. ControlThe gateway checks access rights and limits on every request.
  4. AnswerThe model answers from the team's own content.
  5. MonitorEvery conversation is traced for quality and cost.

What I built

  • Self-service assistants. Teams create assistants in a web interface, without code, with retrieval (RAG) they configure themselves.
  • Connectors. Native connectors to Confluence, Jira and internal databases, so assistants answer from the team's own knowledge.
  • A control layer. A model gateway applies role-based access, API quotas, cost tracking and rate limiting to every request.
  • Observability. Every conversation is traced, so quality and cost can be checked and assistants debugged. Load tests run before each scale-up.
  • Use cases delivered with the teams. Knowledge and productivity: documentation assistants across sources, business content generation. Engineering and DevSecOps: questions to databases in plain language (Text-to-SQL), code analysis, Kubernetes manifest review, help with CVE remediation.

Design choices

Autonomy inside rules

Teams stopped waiting for a project each because the rules (access, quotas, cost) were built into the platform rather than reviewed case by case.

Stack

Python · LangChain · LangGraph · LlamaIndex · Amazon Bedrock (Claude) · Open WebUI · LiteLLM · Langfuse · PostgreSQL · Elasticsearch · Kubernetes · Helm · ArgoCD · GitLab CI/CD · Dynatrace · Prometheus

Questions

How do you let 90 teams build their own AI assistants safely?

Offer a self-service platform where teams configure assistants and their sources themselves, and put a gateway in front of the models that applies access rights, quotas, cost tracking and rate limits to every request, with tracing on every conversation. Omar El Hachimi built this for LCL.