Pipesignal: Agentic AI for B2B Project Discovery
Industrial sellers often hear about a project after the specification is settled. I built Pipesignal to test whether public signals can bring them into the conversation earlier: it finds the signals, enriches them and ranks what deserves a seller's time.
Evidence and scope
Founder-built product concept
Product strategy, concept and build
A product concept with a recorded demo. It has not yet been tested with a sales team, so no pipeline or revenue figures are claimed.
Outcome and evidence
Industrial sellers often hear about a project after the specification is settled. I built Pipesignal to test whether public signals can bring them into the conversation earlier: it finds the signals, enriches them and ranks what deserves a seller's time.
My role
I wrote the product brief, shaped the commercial concept and built the prototype. This is my own product work, separate from my employment.
The problem
By the time a tender is public, the specification is usually settled. Earlier signals sit in permits, investment announcements and planning records, but they are scattered and slow to research.
The approach
- Find public project signals and keep the source.
- Sort them by project stage, region and the people involved.
- Use agents to enrich each signal and propose priorities for a seller.
Constraints and governance
Public signals can be incomplete, out of date or wrong, so value and timing need checking against the source, and contact research needs the right permissions. A working demo does not yet prove reliable data in continuous use.
Outcome and evidence
A working concept and a recorded demo. The next test is whether a real sales team finds relevant projects earlier, with acceptable accuracy and effort.
Outcome and evidence
- Discovery objective: Pre-tender signals
- Design pattern: Agent-assisted research
- Evidence status: Personal concept / prototype