Deal Flow

Pipeline Thesis

DataZoom generates commercial opportunity by eliminating the most expensive bottleneck in M&A, fundraising, and legal due diligence: the manual review of dense document sets. The platform's RAG pipeline — built on pgvector semantic search, sentence-transformers embeddings (all-MiniLM-L6-v2, 384-dimensional vectors), and a Qwen2.5:32B LLM routed through Modal — lets deal teams query hundreds of contracts, equity documents, and financial records in natural language and receive sourced, cited answers in seconds. Every upload processed, every AI query answered, and every cap table extracted is a proof point that shortens the sales cycle for the next account. The product's multi-tenant architecture (Clerk org-based isolation, Supabase RLS) means a single enterprise deployment services an entire firm without data leakage, making DataZoom a platform-level purchase rather than a point-tool evaluation.

---

Target Accounts

---

Qualification Signals

• BR-001 — Document volume pain: Prospect references >50 documents in a single deal or regularly uploads PDFs to shared drives with no queryable index; confirmed by asking about current review workflow.
• BR-002 — Cap table complexity: Organization manages equity across multiple instrument types (SAFEs, options, warrants); evidenced by interest in /api/product/app/api/cap-table/extract and cap-table/review endpoints, or frustration with spreadsheet reconciliation.
• BR-003 — Multi-stakeholder deal room: Two or more parties (counsel, investors, management) need concurrent access to the same document corpus; collaboration token endpoint (/api/product/app/api/collaboration/token) and Clerk multi-tenant model directly address this.
• BR-004 — Compliance or audit trigger: Prospect is preparing for a regulatory audit, litigation discovery, or LP reporting cycle; timeline visualization (schema: timeline_events.impact field with critical/high/medium/low classification) and activity export (/api/product/app/api/activity/export) are immediately valuable.
• BR-005 — AI readiness: IT or legal ops team has already evaluated OpenAI or similar tools but was blocked on data residency; DataZoom's local Ollama path (docker-compose.yml gpu profile, docker/Dockerfile.gpu) and on-premise deployment (fly/orchestrator/Dockerfile, docker/cloud-worker/Dockerfile) resolve the blocker.
• BR-006 — Existing Supabase or Clerk footprint: Engineering team already uses Supabase or Clerk; integration friction is low and supabase start + npm run dev gets a working environment in under 15 minutes (README.md quick-start).
• BR-007 — Mixpanel or analytics maturity: Prospect measures product-led growth metrics; DataZoom's Mixpanel integration (docs/archive/MIXPANEL_TRACKING_PLAN.md, /api/product/app/api/ai/track-interaction) enables usage-based ROI reporting back to the champion.

---

Conversion Loop

1. Activation (Day 0–3) A prospect's first session begins at http://localhost:3000 (self-hosted trial) or a provisioned Clerk org on the cloud deployment. The setup.sh script, supabase start, and npm run dev sequence (README.md) gets the environment live in under 15 minutes. The champion uploads 3–5 documents from an active deal. The ingestion pipeline — Python worker (docker/Dockerfile.worker) → sentence-transformer embedding → document_chunks INSERT with VECTOR(384) column → ivfflat cosine index — processes the corpus silently in the background.

2. First Value Moment (Day 1–5) The champion asks their first natural language question in the AI chat interface (product/app/(app)/context/page.tsx, conversation-panel.tsx). The RAG retrieval layer (product/lib/__tests__/rag-retrieval-enhanced.test.ts) returns a sourced answer with citations pinned to specific document_chunks records. The due diligence checklist auto-generates via /api/product/app/api/business-types/[typeKey]/checklist, pre-populated against the uploaded corpus. The cap-table extraction pipeline fires against any equity documents detected (document_type IN ('equity', 'ip_assignment')), and the review queue at /api/product/app/api/cap-table/review surfaces candidates for human approval.

3. Proof Expansion (Week 2–4) The champion shares the timeline visualization (timeline_events table, event dates, parties_involved[], impact classification) and the risk memo (/api/product/app/api/advisor/risk-memo) with their deal partner or managing director. The advisor batch endpoint (/api/product/app/api/advisor/batch) and strategic options analysis (/api/product/app/api/advisor/strategic-options) generate board-ready outputs that replace billable associate hours. Activity metrics (/api/product/app/api/activity/metrics) give the champion a usage narrative — documents processed, queries answered, events extracted — to take to the budget holder.

4. Commercial Trigger The champion hits an org-level limit, requests a second deal room, or needs e-signature workflow (docs/action_plans/e_sign/09_production_esign_action_plan.md) to close the loop on document execution. This is the upgrade moment. The billing and wallet system (docs/billing_wallet/BILLING_WALLET_IMPLEMENTATION_MASTER.md) converts the trial into a paid subscription. The multi-org scaling architecture (docs/MULTI_ORG_SCALING_PLAN.md) supports firm-wide rollout across practice groups or portfolio companies without re-architecting.

5. Expansion and Referral Each additional Clerk org onboarded (new deal, new portfolio company, new client matter) generates incremental ARR. The activity feed (/api/product/app/api/activity/unified/feed), calendar (/api/product/app/api/activity/unified/calendar), and export (/api/product/app/api/activity/export) create a persistent audit trail that becomes institutional infrastructure — increasing switching cost and driving organic referrals to peer firms who ask how the team is moving so fast on diligence.