Skip to content

Latest commit

 

History

History
212 lines (170 loc) · 11.2 KB

File metadata and controls

212 lines (170 loc) · 11.2 KB

Use Case 1 — M&A Due Diligence Research (Private Equity)

The three GTM questions this use case answers:

  • Is a human required in order for us to perform this transaction?
  • Where can we realistically provide agentic transactions as a customer segment?
  • How can we empower our existing customers with this workflow?

Narrative Overview

The Scenario

A mid-market private equity firm runs a lean deal team. Each managing director covers 15–20 active processes simultaneously, requiring preliminary due diligence across financial performance, litigation history, regulatory filings, and competitive positioning. A junior analyst currently spends two to three days per target assembling a first-pass research package before the deal team decides whether to proceed to full diligence.

The firm has deployed an AI research agent to automate the first-pass package — querying three proprietary data sources and producing a structured memo within 90 minutes of a new target entering the pipeline.

The Seller Perspective

Each data provider sells annual enterprise subscriptions priced for human analyst teams. Their data is highly relevant to this workflow, but the agent cannot navigate a pricing page, request a trial, or execute a contract. Publishing an ATS makes each provider immediately discoverable and transactable.

Per-query pricing is structurally better aligned to agent consumption than flat annual seat licensing: intensive bursts during active deal processes, quiet periods between them.

Provider Data Category Pricing Settlement
Capital Intel Financial company data $0.05 / query Prepaid credit / x402
LegalVault Litigation records $0.12 / record Prepaid credit
NewsMatrix Media intelligence $0.03 / news package Prepaid credit

The Buyer Perspective

The firm's AAP encodes its data access policy: spend limits per transaction and per day, permitted data categories, required certifications, and an escalation rule for first-time provider access above $0.10 per transaction.

  • Capital Intel ($0.05) → APPROVED immediately
  • NewsMatrix ($0.03) → APPROVED immediately
  • LegalVault ($0.12) → ESCALATED on first access (above $0.10 threshold) → human approves once → autonomous thereafter

Answering the Three GTM Questions

Is a human required? No, for the vast majority of transactions. The LegalVault first-access escalation requires one-time human approval. After that, the complete three-source research package runs without human involvement per transaction. ~97% fully autonomous.

Where can we serve agents as a customer segment? Any PE firm, hedge fund, corporate development team, or strategic acquirer running agentic research workflows. This segment does not appear in current CRM pipelines because they have no way to transact today.

How do we empower existing customers? PE firms and financial institutions already on API contracts can activate agent workflows immediately once an ATS is published. Existing credentials become agent-accessible without a new sales cycle.


Structured Flow

Provider ATS — Capital Intel

Published at: https://capitalintel.com/.well-known/agent-terms.json

{
  "ats_version": "0.1.0",
  "provider": {
    "id": "com.capitalintel.data",
    "name": "Capital Intel",
    "jurisdiction": "US",
    "certifications": ["SOC2_TYPE2", "ISO27001"]
  },
  "product": {
    "category": "financial_company_data",
    "description": "Company financials, filings, and performance data for 50K+ US firms."
  },
  "pricing": {
    "currency": "USD",
    "models": [{ "model_id": "per-query", "type": "per_query", "unit_price": 0.05 }],
    "settlement_methods": [
      { "method": "prepaid_credit", "min_purchase_usd": 5.00, "topup_via": ["stripe"] },
      { "method": "x402", "use_case": "per_query" }
    ]
  },
  "usage_policy": {
    "permitted_uses": ["internal_analysis", "due_diligence"],
    "prohibited_uses": ["model_training", "redistribution"]
  },
  "legal": {
    "governing_law": "State of Delaware, United States",
    "auto_renew": false,
    "agent_binding": "agent_acts_as_agent_of_buyer"
  }
}

Agent AAP — PE Firm Research Agent

{
  "aap_version": "0.1.0",
  "policy_id": "com.acme-pe.agents.research.v1",
  "spend_controls": {
    "per_transaction_limit": 0.10,
    "per_day_limit": 50.00,
    "preferred_pricing_models": ["per_query"],
    "preferred_settlement_methods": [
      { "method": "prepaid_credit", "use_case": "per_query" },
      { "method": "x402", "use_case": "per_query" }
    ]
  },
  "data_access": {
    "permitted_categories": ["financial_company_data", "legal_case_data", "news_media"],
    "required_provider_certifications": ["SOC2_TYPE2"]
  },
  "legal_authority": {
    "binding_authority": "per_transaction_only",
    "auto_renew_prohibited": true,
    "governing_law_whitelist": ["US", "GB", "EU"]
  },
  "escalation_rules": {
    "escalate_on": ["new_provider_first_access_above_0.10"]
  }
}

Authorization Decision Flow

┌─────────────────────────────────────────────────────────────────────┐
│  Stage      Actor            Action                      Schema     │
├─────────────────────────────────────────────────────────────────────┤
│  DISC       Agent            Fetches /.well-known/       ATS        │
│                              agent-terms.json                       │
├─────────────────────────────────────────────────────────────────────┤
│  PARSE      Agent            Extracts: per_query $0.05,  ATS        │
│                              SOC2_TYPE2, US jurisdiction,           │
│                              prepaid_credit settlement              │
├─────────────────────────────────────────────────────────────────────┤
│  EVAL       ACPEvaluator     14-point policy check:      AAP        │
│                              price $0.05 < $0.10 ✓                  │
│                              category permitted ✓                   │
│                              SOC2_TYPE2 present ✓                   │
│                              jurisdiction US ✓                      │
│                              use case permitted ✓                   │
│                              (all 14 checks pass)                   │
├─────────────────────────────────────────────────────────────────────┤
│  AUTH       ACPEvaluator     APPROVED — no escalation    AAP → ATR  │
│                              triggers fired                         │
├─────────────────────────────────────────────────────────────────────┤
│  PAY        Agent            Draws $0.05 from prepaid    Settlement │
│                              credit wallet                          │
│                              (balance: $4.95 remaining)             │
├─────────────────────────────────────────────────────────────────────┤
│  ATR        Agent            Posts signed ATR:           ATR        │
│                              - SHA-256 hash of ATS                  │
│                              - wallet draw reference                │
│                              - agent DID signature                  │
│                              - deal file ID                         │
└─────────────────────────────────────────────────────────────────────┘

Mermaid Flow Diagram

flowchart LR
    A[Agent\nReceives Target] --> B[Fetch ATS\ncapitalintel.com]
    B --> C{ACPEvaluator\n14-point check}
    C -->|All pass| D[APPROVED]
    C -->|First access\n> $0.10| E[ESCALATED\nHuman review]
    C -->|Policy fail| F[REJECTED]
    D --> G[Draw $0.05\nPrepaid wallet]
    G --> H[Retrieve data]
    H --> I[Post signed ATR\nDeal file]
    E --> J[Human approves\none time]
    J --> D
Loading

Outcome Distribution — 100 target queries / month

Outcome Count % of Total Human Required?
✅ APPROVED 97 97% No — fully autonomous
⚠️ ESCALATED 2 2% Yes — first-access review
❌ REJECTED 1 1% No — auto-rejected

After first-access approval of LegalVault: 100% of subsequent queries run autonomously.


GTM Implications

For Data Providers

  • New customer segment: Agent-deployed PE and corporate development workflows represent demand that is currently invisible — no MQL, no trial request, no SDR touchpoint. The agent hit your paywall and moved on.
  • Pricing model shift: Per-query pricing captures deal-cycle consumption patterns that per-seat licensing structurally underprices. A firm running 20 active processes simultaneously may query your product 500 times in a week, then go quiet for three weeks.
  • First-access escalation is a trust signal: The AAP's first-access escalation rule means human deal team members are still in the approval chain for new providers. That's a relationship touch point and a credibility checkpoint — not friction.

For Agent Builders

  • AAP design tip: Set per_transaction_limit at the meaningful threshold for your org's risk tolerance, not at the lowest possible value. Over-restrictive limits create unnecessary escalation volume and reduce the agent's utility.
  • Known providers list: Pre-populating known_providers in the ACPAgentClient for established data sources eliminates first-access escalation for providers your org has already vetted.
  • ATR deal file integration: The transaction_id in the ATR can be mapped to a deal CRM record, creating an automatic evidence trail per target that links data sourcing to deal outcomes.

Related Files


Agent Commerce Protocol v0.1.0 — github.com/fxops-ai/agent-commerce