Radar

Signal everywhere. Brain nowhere.

The GTM stack knew a lot: HubSpot knew engagement, Salesforce knew pipeline, ZoomInfo knew firmographics, the SEP knew activity. Nobody knew everything at once — and each tool that promised to be the brain wanted per-seat pricing, opaque scoring, and another integration to babysit.

Meanwhile inbound leads waited on manual triage, account prioritization lived in tribal knowledge, and every new "AI SDR" vendor pitch was another $30K line item for something the team couldn't inspect or control.

Why build, and why like this.

The decisions that shaped the system:

One brain, many hands.

Signals flow in from the systems of record; the scoring core normalizes them into account and lead state; agents and rules act on that state through whichever hands the team already uses.

What it's made of. What it talks to.

Next.js + TypeScript app, deck.gl WebGL compass, Claude agents for research and enrichment, direct REST integrations where native connectors didn't exist.

Not a ceiling — this is what Radar was being built against. Gong, review sites, ads platforms, the rest of the rack: anything the team owns with API access can join the mesh.

// The whole scoring model is data — edit the weights, publish, done.
export const DEFAULT_MODEL: ModelConfig = {
  version: "v3",
  xDivide: 5.0, yDivide: 5.0,          // the quadrant cut lines
  fitWeights: { industry: 25, size: 20, revenue: 10, stack: 30, lookalike: 15 },
  signals: [
    // awareness/readiness points, decay windows, half-life on repeats, per-account cap
    { id: "g2-compare",   aw: 4, rd: 9,  awDecay: 14, rdDecay: 7,  halfLife: 14, leadCap: 2 },
    { id: "pricing",      aw: 2, rd: 9,  awDecay: 30, rdDecay: 5,  halfLife: 14, leadCap: 2 },
    { id: "demo-request", aw: 3, rd: 10, awDecay: 60, rdDecay: 10, halfLife: 30, leadCap: 1 },
    { id: "champion",     aw: 8, rd: 6,  awDecay: 90, rdDecay: 60, halfLife: 90, leadCap: 1 },
  ],
  gates: { competitor: true, freeEmail: true, customerRouting: true, dnc: true },
  rollup: { perLeadDivide: 1.5, accountCap: 40, lookbackDays: 90, seniorityWeighting: true },
};

Inside the cockpit.

Captures from the working demo — the compass, an account dossier, the inbound trigger engine. A short walkthrough film lands here next.

RADAR · THE SCOPE2,503 ACCOUNTS · AWARENESS × READINESS

The compass. Every account charted on awareness × readiness — heat shows where the book concentrates, and the movement panel tracks who crossed quadrants this week.

ACCOUNT 360FIT 97 · SALES-READY

The dossier. One account, fully read — the score, the signals behind it, and why it moved this week.

INBOUND TRIGGERSGATE → ENRICH → SCORE → ROUTE

The trigger engine. Inbound never waits on a human to notice it — each play gates, enriches, scores, and routes while the lead is still warm.

The case for one brain.

Speed-to-lead in hours, not days — inbound never waits on a human to notice it. Scoring you can read and edit — every threshold and gate is inspectable, not a vendor's black box. Costs that scale with usage, not seats — agent workloads run on direct API billing. One chart the whole team steers by — awareness × readiness, no tribal knowledge required.