AbbVie logo

AI Marketing Assets & Content Automation

Pharmaceutical – AI Marketing Assets Content Automation

Agentic RAG pipeline generating traceable pharmaceutical marketing and regulatory-content drafts.

An agentic document chain

Regulatory documents, marketing assets and slides drafted by agents, with traceable sources

Project Details

Client

AbbVie

Sector

Pharmaceutical – AI Marketing Assets Content Automation

Year

2024

Key Technologies
LangchainLangGraphOpenAIHuggingFace

Mission

Lead Data Scientist / AI Architect - Agentic Content Generation

RAG, LLM & Agentic Content Generation

Project assignment, Paris / remote.

Advanced automation of the generation of regulatory documents (MA), marketing materials and PowerPoint presentations using an agentic RAG/LLM pipeline orchestrating the entire document production cycle, from multi-source ingestion to final contextualized generation. From-scratch development of a modular application on Palantir Foundry / Code Workspace: ingestion, chunking, embeddings (OpenAI/HuggingFace), vector search (PostgreSQL, SQLAlchemy, Alembic) and fine-grained contextual memory management for agents truly autonomous in their actions. Integration of Langchain & LangGraph to build a true collective of agents: dynamic task chaining, multi-step planning, initiative-taking on complex scenarios (regulatory validation, material structuring, ad hoc answers) with proactive tracking (Langfuse monitoring). Automation: PowerPoint slide creation via an agentic templating engine: agents able to dynamically generate, organize and adapt content into reusable blocks/functions according to business or regulatory compliance constraints. Advanced agentic prompt engineering: automated structural tuning (temperature, tokens, dynamic parameters) and smart instruction management to guide agent behavior based on document context and workflow requirements. End-to-end orchestration tests: Pytest, end-to-end multi-agent validation to ensure robustness, auditability and reliability in production. Performance monitoring: continuous monitoring through advanced metrics (precision@k, recall@k, ROUGE, BLEU), Palantir dashboards for real-time visibility, error analysis and source traceability, and continuous agent adaptation based on observed results. Comprehensive documentation and knowledge transfer to business teams (Marketing/IT) to maximize adoption and autonomy on the solution.

Contexte & Environnement

Pharmaceutical, marketing and regulatory affairs: orchestrate, make reliable and accelerate the creation of regulatory deliverables & market assets through agentic architectures.

Team

Lead Data Scientist / AI Architect, with the Marketing and IT teams

Environment

Paris / remote – Palantir Foundry / Code Workspace, PostgreSQL, OpenAI/HuggingFace embeddings

Objectifs Cles

1Drastically reduce time-to-market.
2Support compliance validation, auditability and traceability.
3Measure end-to-end performance through agentic AI.

Technologies & Infrastructure

OpenAIHuggingFace TransformersLangchainLangGraphPostgreSQLSQLAlchemyAlembicPythonLangfusePytestPalantir Code Workspace

Ready to Achieve Similar Results?

Let's discuss how we can drive measurable impact for your organization.