RevenuePilotAI
A coordinated sales workflow for B2B teams. Specialist AI agents research accounts, ground answers, and prepare CRM actions for human approval.
What does RevenuePilotAI demonstrate?
RevenuePilotAI is a B2B sales assistant that coordinates five specialist AI agents through a Supervisor. Its documented architecture combines RAG-grounded answers, account research, opportunity strategy, and follow-up with human approval before CRM changes. The September 2026 project notes identify the CRM executor as using mock data.
Connect account research, opportunity strategy, and follow-up in a sales workflow with human review before CRM changes.
Inside the build
- A Supervisor coordinates five specialist Claude agents for research, strategy, product knowledge, CRM planning, and follow-up.
- ChromaDB grounds answers in retrieved knowledge, with optional Langfuse tracing.
- A Python backend and Next.js dashboard support an approval workflow; the documented CRM executor currently uses mock data.
Product in active development / documented September 2026 implementation
The problem
B2B sales work spans account research, product knowledge, opportunity planning, and follow-up. Teams need a coordinated workflow with review before changes reach their CRM.
Malik’s engineering focus
Portfolio focus: Supervisor-led agent coordination, ChromaDB retrieval, a Python backend, and a Next.js dashboard with Supabase authentication and row-level security. CRM actions are designed to require human approval.
Validation evidence
The September 5, 2026 code review records frontend lint, type checking, a production build, and a passing Python test suite. These are historical, self-reported checks from the project documentation; they were not rerun for this portfolio entry.
Documented outcome
The documented implementation brings research, strategy, product knowledge, CRM action planning, and follow-up into one sales assistant, with a dashboard approval flow and a read-only guided demo.
Limitations
The September project notes list a real HubSpot backend as future work and identify the CRM executor as using mock data. The documentation does not establish customer adoption, revenue gains, or current production readiness.
Source material reviewed November 1, 2026. Project documentation is self-reported; no customer deployment or commercial result is implied.