Voice AI Prospect Radar
An independent prospecting tool for voice-AI vendors, combining UK business discovery, a map and ranked shortlist, explainable scoring, on-demand AI briefs and a persistent review pipeline.
Product walkthrough published June 2026. Public business data and illustrative review states; counts and fit scores reflect the recorded demo snapshot.
Finding a business is only the first step. A useful prospecting tool also needs to explain its fit and preserve the next steps.
I developed this independent MVP around the work of finding and reviewing prospects for voice-AI vendors. The application brings business data, a map, a ranked shortlist, a readable explanation and saved review tickets into one interface. It is a prospecting product; voice calling is outside its scope.
From requirements to working code.
- Built UK business discovery, map filters and a ranked prospect shortlist using Google Places data.
- Implemented a deterministic scoring rubric with visible reasons and handling for missing signals.
- Added on-demand AI briefs to help a reviewer understand an individual prospect.
- Built persistent review tickets with normalized pipeline states and saved prospect context.
- Added access controls, caching and reusable enrichment results to manage provider usage.
Explain the ranking
Use a transparent scoring rubric so a reviewer can inspect the signals behind a prospect’s position. The rubric is a screening aid, with missing-data penalties and explicit limitations.
Separate discovery from the map
Keep business discovery and filtering independent of map rendering, so the shortlist remains a useful review surface.
Generate briefs when needed
Request AI enrichment for a chosen prospect and reuse the result, keeping the discovery flow responsive and provider usage controlled.
Preserve the next step
Save review tickets and prospect context so the work continues beyond browsing pins and opening a brief.
What the work supports.
- The public repository and project write-up document the implementation and its development decisions.
- The repository includes Vitest coverage for score boundaries, missing signals, malformed enrichment responses, filtering and ticket-state normalization.
- A recorded walkthrough shows the dashboard, saved map, prospect brief, business search and review pipeline.
Previously deployed as a gated portfolio MVP. Current end-to-end availability has not been verified. The hand-designed scoring rubric has not been calibrated against sales outcomes; review signals are proxies. Commercial accounts and full CRM integration remain future work.