AutoM11

• 

  n8n workflow

Inspiration

The drug approval process is slow, manual, and costly, often causing long delays. Regulatory bodies like the FDA and MHRA require multiple revisions that can take months. We set out to automate and speed up the creation of M11 approval forms by mirroring real-world pharma workflows through an AI-driven multi-agent system that collaboratively researches, drafts, reviews, and refines trial documentation.

What it does

AutoM11 automates the drafting of M11 regulatory forms by simulating collaboration between real pharmaceutical departments using AI agents.

1. Evidence Researcher uses Perplexity Sonar to gather trial data and background evidence.
2. Parser structures the research into a standardized format for analysis.
3. Orchestrator manages communication and assigns tasks to departmental agents.
4. Medical Agent handles clinical and endpoint details. 5.Scientific Agent validates study design and objectives.
5. Data Management Agent ensures data accuracy and compliance.

Each department uses Perplexity-powered searches to fact-check and supplement the research. They then verify findings and propose edits collaboratively. The Orchestrator designates the Medical Department as the primary drafter, while the other agents act as critiquers, iterating until consensus is reached.

How we built it

We built AutoM11 in n8n, designing a modular workflow with integrated OpenAI and Perplexity APIs. Each agent uses structured output parsing to ensure reliable communication and consistent reasoning.

Challenges

Designing how agents communicate with each other, integrating APIs in n8n with little prior experience, and ensuring the workflow accurately mirrored real-world pharma processes for future scalability.

What's next for AutoM11

We aim to expand AutoM11 to cover the full M11 form, add human-in-the-loop validation, and evolve it into a scalable automation tool for mid-size pharma companies.

Built With

• n8n
• perplexity
• python
• react
• typescript