PRD v2: Full 4-layer product requirements document - Layer 0: Core Platform (Company Twin, Taxonomy, Channels, Approvals, Trust) - Layer 1: Sales OS (leads, outreach, proposals, pipeline) - Layer 2: Deal Exchange OS (matching, barter, co-sell, deal rooms) - Layer 3: Strategic Growth OS (acquisition, ecosystem, simulator, ROI) Module map: Complete inventory of all 50+ services, 52 API routes, 30 models, 37 frontend components with build status https://claude.ai/code/session_01LsnvBa7HwF5hs99VZbgLGj
4.5 KiB
Dealix PRD — Product Requirements Document
Version: 2.0 | Date: 2026-04-11 | Status: Active
Product Identity
Name: Dealix Type: Commercial Intelligence & Deal Operating System Market: Saudi B2B (primary), GCC (secondary) Position: Not a CRM. A Revenue + Partnership + Strategic Deal OS.
Architecture: 4 Layers
Layer 3: Strategic Growth OS (acquisition, ecosystem, ROI)
Layer 2: Deal Exchange OS (barter, co-sell, reseller, partnerships)
Layer 1: Sales OS (leads, outreach, proposals, pipeline)
Layer 0: Core Platform (Company Twin, Taxonomy, Channels, Approvals, Trust, Memory)
Layer 0 — Core Platform
0.1 Company Twin
Every tenant gets a structured digital twin:
- Identity: name, industry, CR, geography, size
- Capabilities: services, products, white-label capacity, barter assets
- Needs: marketing, delivery, distribution, capital, partners
- Authority Matrix: what AI can commit vs what needs approval
- Red Lines: forbidden claims, blocked sectors, pricing floors
- Deal Preferences: allowed/blocked deal types, min/max values
0.2 Deal Taxonomy (15 types)
sales_lead, referral, co_selling, co_marketing, subcontracting, white_label, reseller, strategic_alliance, channel_partnership, joint_venture, acquisition_scouting, investment_intro, vendor_replacement, capability_gap_fill, tender_consortium
0.3 Channel Engine
- Email: PRIMARY outbound (SPF/DKIM/DMARC, unsubscribe, consent)
- LinkedIn: ASSIST-MODE ONLY (draft, queue, operator review — no bots)
- WhatsApp: WARM ONLY (opt-in, 24h window, approved templates)
0.4 Approval Center
- Class A (auto): summarize, classify, score, internal drafts
- Class B (approval): send outreach, share pricing, propose terms
- Class C (executive): exclusivity, equity, legal, acquisition
0.5 Trust & Verification
Every agent run produces: claim → actual action → evidence → verdict Verdicts: VERIFIED, PARTIAL, UNVERIFIED, CONTRADICTED, BLOCKED
0.6 Shared Memory
Operational + account + market + negotiation + campaign memory DB = source of truth. Memory = assistive recall layer.
0.7 Observability
Cost tracking, performance metrics, channel health, anomaly detection
Layer 1 — Sales OS
Modules
- ICP Engine: customer segment definitions
- Lead Discovery: company sourcing + enrichment
- Lead Intelligence: pain inference, urgency, entry point
- Outreach Engine: email sequences, LinkedIn assist, WhatsApp warm
- Proposal Engine: scoped offers, pricing, pilot options
- Sales Memory: objections, patterns, what worked
- Pipeline: stage management, velocity tracking
KPIs
lead→meeting, meeting→proposal, proposal→close, reply rate, cycle time
Layer 2 — Deal Exchange OS
Modules
- Offer Graph: what we can provide as partner
- Need Graph: what we lack
- Reciprocal Match Engine: mutual value scoring
- Partner Scoring: fit, reciprocity, credibility, risk
- Reciprocal Offer Generator: barter/co-sell/reseller structures
- Pilot Proposal Generator: bounded first-step deals
- Deal Room: workspace with BATNA, concessions, approvals
KPIs
partner response rate, pilot acceptance, pilot→full conversion, reciprocal value
Layer 3 — Strategic Growth OS
Modules
- Acquisition Scouting: target sourcing + scoring + briefs
- Ecosystem Mapper: partner landscape visualization
- Strategic Simulator: scenario modeling (upside/downside/risk)
- ROI Engine: CAC reduction, distribution value, margin impact
- Partner Performance Graph: contribution tracking
- Portfolio Intelligence: vertical wins, pattern detection
KPIs
partner-sourced revenue, acquisition candidates qualified, ecosystem coverage
Non-Goals
- NOT a general-purpose chatbot
- NOT an uncontrolled automation bot
- NOT a financial trading system
- NOT a replacement for legal review on binding terms
Success Criteria
- Product understands each client's business model (Company Twin)
- Discovers and scores strategic counterparties
- Generates structured opportunities, not just raw leads
- Manages multi-channel outreach under policy
- Preserves commercial memory across sessions
- Verifies what agents actually did
- Runs safely with approval gates
- Survives launch simulation
- UI feels premium and operational
- Architecture is coherent and maintainable
Phased Rollout
- Sprint 1: Core Platform + Company Twin + Taxonomy + Approvals
- Sprint 2: Sales OS MVP (lead→outreach→pipeline)
- Sprint 3: Deal Exchange OS MVP (matching→deal room→pilot)
- Sprint 4: Strategic Growth MVP (scouting→simulator→ROI)
- Sprint 5: Hardening + QA + Launch