Back to all work
Revenue IntelligenceIn production~$280K/yr saved

Qlari

AI-native revenue intelligence — board-grade briefs, ideal-customer targeting, and self-building dashboards.

Role Architected & led end-to-end

Qlari is an internal revenue intelligence product built to replace Clari and Revic, giving Sales and leadership a real-time pipeline, account, deal-health, and forecasting experience tailored to how OPSWAT actually sells — with an AI layer that writes, sources, and visualizes intelligence itself.

What it does

  • Live CRM & CPQ integration — real-time pipeline health, renewal risk, and expansion signals.
  • AI account research briefs and a natural-language revenue assistant for reps.
  • AI-generated dashboards & charts — executives ask; validated visuals appear.
  • Deal risk and expansion insights from nightly health scoring.
  • MCP server exposing revenue intelligence to other agents and tools.

The engine, in three parts

01 · AI Sales Briefs

A 20-page account brief a strategist would take a week to write. Generated while you grab coffee.

Type a company name. Qlari resolves who they really are, reads everything OPSWAT knows about them, scans the open web, and returns a fully-cited, sales-ready battle plan — structured the way top reps actually sell.

  • The Gambit — a 10-second thesis and an opening line built to earn the first meeting.
  • Full ForceManagement 9-box + MEDDPICC — grounded in real named contacts, never invented.
  • Lead-with recommendations — the products with the strongest buying signal right now, each with the signal that justifies it.
  • Integrated solution plays — multi-product architecture stories, not single-tool pitches.
  • For renewals — deal health, churn signals mined from support tickets, and expansion whitespace.
Why it's differentEvery fact is cited or cut. A cheap model finds, a grounded model extracts, a flagship model writes — quality up, cost in cents. Living briefs quietly re-check hot accounts for a fresh breach, a new CISO, or a churn signal — and tell you first.
Grounded on
SugarCRMOSCAR knowledge baseZoomInfoLive webDNS fingerprintingATS job boardsSupport history
02 · Prospect Radar — ICP scan

Stop guessing who to chase. Your best deals define your next ones.

Prospect Radar learns the Ideal Customer Profile from deals already won, then continuously sources and scores net-new companies against it from public registries — pipeline generation that starts from evidence, not a hunch.

  • ICP from your own wins — clusters closed-won by industry, region, and product; proposes weighted verticals, size bands, regulatory drivers, disqualifiers, and matched SIC/NAICS codes.
  • Sources candidates automatically — SEC EDGAR, the FDIC banking directory, SAM.gov, and the existing 6sense pipeline.
  • Two-axis scoring — ICP Fit × Readiness, plus a one-line best opening angle. A perfect-fit account that isn't ready surfaces as a nurture, not a miss.
  • Quadrant view — the whole prospect universe at a glance; filter, override, or exclude.
Why it's differentOne click promotes any candidate straight into a full research brief. Radar finds them; the Intelligence Engine arms you for them — both halves of pipeline generation, wired end to end.
03 · Self-building analytics

Describe the chart you want. Watch it appear — correct, on-brand, and safe to ship.

"Closed Won by quarter, top 8 industries." That sentence is the whole workflow. The AI writes the query, picks the visualization, styles it to the palette, and hands you a live preview before anything goes live.

  • 14 chart types, one sentence each — from waterfalls and quadrant scatters to funnels with auto-conversion and KPI grids.
  • Restyle by asking — "make the target line a dashed grey" and it happens. No CSS, ever.
  • Propose → preview → publish — every change is a diffable proposal; every save is a version that can roll back.
  • Sandbox redesigns — clone a whole dashboard, redesign it privately, share a review link, publish atomically.
  • Trust by construction — read-only database role, layered SQL allowlist, strictly-typed chart specs, allowlisted fonts and validated colors. Nothing an AI emits can inject anything.
The MCP angleThe same authoring workflow is exposed over MCP — Claude or any connected agent can build and validate charts against live data, hand a human a review link, and publish a fully-versioned update. Same permissions, same guardrails.
Why it mattersIt proves internal products can beat generic SaaS when company-specific data, workflow, and intelligence are deeply integrated.
ImpactReplaced two commercial SaaS platforms and saves approximately $280K per year — while creating a product Sales actively prefers to use.
Built with
FastAPIPostgreSQLReactRechartsMCPPlaywrightAzure