Continuous improvement and conversion

Map a buyer job to the paid endpoint an agent should call next.

Move from seeded/indexed endpoints to non-seed paid usage by making the strongest buyer jobs obvious, especially lead briefs, runtime route preflight, workflow audit, and receipt verification. Use the buyer language already winning in-market: no API keys, no subscriptions, pay per request.

Why buyers pay for Lead Brief Lite

Outcome

  • Pay $0.03 when the buyer wants a decision-ready brief and the agent already has public/caller-supplied lead facts (not raw fields).
  • One call returns: brief (headline + fit + useful facts + next actions + caveats), source_pack, missing_fields (explicit gaps), proof_boundary, receipt (hashes + id), and human_summary.
  • After x402 settlement clears, treat the receipt + proof_boundary + missing_fields as the required guardrail layer: inspect them before the agent does outreach, CRM write, or chains to another paid tool.
  • Best fit: public hints exist; buyer wants usable artifact + auditable "what was known / what was not claimed" evidence instead of another interpretation step.

Preview first

Inspect the response shape before payment when a human or agent needs to see the exact contract.

curl -X POST https://boundary-guard.vercel.app/lead-brief-lite -H 'content-type: application/json' --data '{"entity_type":"company","query":"Acme Robotics sales lead: https://example.com/about, sales@example.com, Raleigh, makes warehouse automation software.","known_fields":{"name":"Acme Robotics","location":"Raleigh, NC"},"brief_goal":"sales","policy":{"no_secret_mode":true,"require_source_pack":true,"max_bullets":5}}'

Paid host trigger

Use the paid Render host when the buyer wants x402 settlement, receipt evidence, and the exact paid route that shows up in marketplace discovery.

The Render host returns an x402 payment challenge until the caller supplies a valid Base-mainnet USDC payment.

Low-friction promise

  • No customer account, API key, or subscription is required to discover the route, inspect the schema, or read the buyer guide.
  • Use the Vercel preview first to inspect the exact response shape without payment.
  • Pay the Render host only when the next useful step is a decision-ready brief with source-pack and receipt evidence.
  • If normalized fields are enough, route to /data-enrich-lite instead of paying for Lead Brief Lite.

Pressure test vs winning low-friction competitors

Question

Can Boundary Guard match the low-friction promise buyers already reward in data/search utilities without pretending to be a raw upstream access bundle?

Winning language

  • StableEnrich wins with pay-per-request access and no auth or subscriptions.
  • AgentMail wins by naming one clear agent job instead of a generic toolbox.

Boundary Guard answer

Preview free on Vercel, then pay $0.03 only when the next useful step is a source-linked lead brief with missing-field flags, proof boundary, and receipt evidence.

Agentic service-level provider/category enrichment is still thin, so Boundary Guard has to carry the buyer job clearly in llms.txt, the buyer guide, and marketplace-facing copy.

Agent-native discover, preview, pay, chain playbook

Use Lead Brief Lite ($0.03) when the buyer has public lead facts and wants a usable brief + source pack + missing fields + proof_boundary + receipt. Preview the response shape first on Vercel, pay on the x402 host when settlement + receipt matter. After payment clears, the post-payment guardrail is mandatory: read receipt + proof_boundary + missing_fields before the agent acts or chains to outreach/CRM/other paid tools. This is the trust layer for agents that have already spent heavily elsewhere.

Discover

Agent action: Read /llms.txt, /.well-known/x402, /openapi.json, and /buyer-conversion-guide.json.

Success signal: The router sees POST /lead-brief-lite, price $0.03, buyer queries, expected fields, and proof boundary.

Preview

Agent action: Call the Vercel preview host with public lead facts to inspect response shape without payment.

Success signal: Response includes brief.headline, useful_facts, source_summary, missing_fields, proof_boundary, and human_summary.

Pay

Agent action: Call the Render paid host when the buyer needs x402 settlement and receipt evidence.

Success signal: The caller receives the x402 payment challenge, settles on Base USDC, and keeps the Boundary Receipt.

Chain

Agent action: Route the brief into outreach, CRM prep, AgentMail-style communication, or workflow audit only after checking missing fields.

Success signal: The downstream agent can act from the brief without inventing private CRM, Apollo, Exa, or Firecrawl evidence.

Measure

Agent action: Classify the paid call as seed/internal, marketplace quality signal, or non-seed buyer usage before counting revenue.

Success signal: Revenue reporting remains proof-safe and the next factory run can learn from the result.

Router rules

  • Prefer /lead-brief-lite over /data-enrich-lite when the next agent needs an outreach-ready or research-ready artifact.
  • Prefer /data-enrich-lite first when the input is still mostly messy fields, URLs, domains, contacts, or source snippets.
  • Prefer /runtime-route-preflight before expensive model/search/API spend when the buyer is choosing where to route budget.
  • Prefer /workflow-audit before multi-step paid workflows, especially when a lead brief will trigger communication or spending.
  • Prefer /receipt-verify when a downstream agent needs to check receipt evidence from a previous paid Boundary Guard call.

AgentCash / MCP style notes

  • Discovery should not require a customer account, API key, or subscription; the x402 payment challenge is the paid-call gate.
  • Expose the route through llms.txt, OpenAPI, root JSON, /.well-known/x402, and marketplace metadata so tool routers can find it.
  • Keep examples small enough for an agent to copy into a single trial call.
  • Do not claim StableEnrich, Apollo, Exa, Firecrawl, Serper, or AgentMail parity; position this as the brief/decision/receipt layer around those flows.

Copy-paste preview command

curl -X POST https://boundary-guard.vercel.app/lead-brief-lite -H 'content-type: application/json' --data '{"entity_type":"company","query":"Acme Robotics sales lead: https://example.com/about, sales@example.com, Raleigh, makes warehouse automation software.","known_fields":{"name":"Acme Robotics","location":"Raleigh, NC"},"brief_goal":"sales","policy":{"no_secret_mode":true,"require_source_pack":true,"max_bullets":5}}'

Lead Brief Lite vs raw enrichment

What the buyer is paying for

Choose /lead-brief-lite: A brief the agent can use immediately for outreach prep, qualification, or research handoff.

Choose raw enrichment first: Fields, source rows, or upstream search output that still needs another decision step.

Best starting input

Choose /lead-brief-lite: A URL, company/person name, email, location, or short public lead note.

Choose raw enrichment first: Messy records that still need entity cleanup, source extraction, or field normalization.

Why the price can be higher

Choose /lead-brief-lite: The buyer is paying for decision-ready packaging, proof boundary, and receipt evidence in one call.

Choose raw enrichment first: Use the cheaper raw step when another tool still has to interpret the result before action.

Call first by market signal

AI runtimes and model access

Choose a model/API route without overspending or violating data policy.

Call first: /runtime-route-preflight

High observed transaction volume makes pre-spend route checks a natural buyer-side wedge.

Data enrichment, search, and extraction

Turn caller-supplied public lead facts into a usable sales or research brief with post-payment guardrails.

Call first: /lead-brief-lite

Agents need a usable brief + source pack + missing-field flags + proof_boundary + receipt evidence in one $0.03 paid call. After settlement, the guardrail (receipt + proof_boundary + missing_fields) must be inspected before the agent acts on the lead or chains paid steps. No API keys or subscriptions.

Data enrichment, search, and extraction

Normalize messy public company, person, URL, or lead hints before CRM or outreach work.

Call first: /data-enrich-lite

This is the lower-cost raw enrichment step when the caller wants normalized fields instead of a brief.

Commerce, payments, and fulfillment

Check a purchase, payment, or fulfillment action before the agent spends.

Call first: /guard

Commerce services show meaningful dollars per transaction; a cheap preflight reduces bad agent spend.

Agent observability and receipts

Audit a multi-step workflow or verify supplied receipt/payment evidence before trust.

Call first: /workflow-audit

Receipt and workflow evidence is Boundary Guard's strongest differentiated lane against simple x402 gateways.

Lead Brief Lite quickstart

Paid call

POST https://boundary-guard-x402.onrender.com/lead-brief-lite

Price: $0.03

The agent is ready for x402 settlement on Base USDC + Boundary Receipt so downstream steps (or auditors) can verify the exact public facts, proof boundary, and guardrails that were paid for.

The Render host returns an x402 payment challenge until the caller supplies a valid Base-mainnet USDC payment.

Plain preview

POST https://boundary-guard.vercel.app/lead-brief-lite

A human or agent wants to inspect the exact response shape, sample brief, missing fields, and proof_boundary language without payment.

Expected fields

  • brief.headline
  • brief.buyer_fit_signal
  • brief.useful_facts
  • brief.source_summary
  • brief.next_agent_actions
  • brief.qualification_questions
  • brief.missing_fields
  • brief.proof_boundary
  • receipt.policy_version
  • receipt.request_hash
  • receipt.response_hash
  • human_summary

Sample payload

{
  "entity_type": "company",
  "query": "Acme Robotics sales lead: https://example.com/about, sales@example.com, Raleigh, makes warehouse automation software.",
  "known_fields": {
    "name": "Acme Robotics",
    "location": "Raleigh, NC"
  },
  "brief_goal": "sales",
  "policy": {
    "no_secret_mode": true,
    "require_source_pack": true,
    "max_bullets": 5
  }
}

What the buyer gets back

Decision-ready output

The buyer should get an outreach-ready or research-ready brief, not just normalized fields.

Look for brief.headline, brief.useful_facts, brief.missing_fields, brief.proof_boundary, and receipt hashes.

Trust-after-payment answer

Current x402 chatter is more worried about unusable output than about micro-pricing. Lead Brief Lite answers that by returning source-bound facts, explicit gaps, and a receipt-backed summary.

Sample response excerpt

{
  "decision": "allow",
  "brief": {
    "headline": "Acme Robotics - sales brief",
    "fit_summary": "Warehouse automation vendor with a clear Raleigh footprint and a public sales contact.",
    "buyer_fit_signal": {
      "status": "ready_for_outreach_prep",
      "score": 90,
      "reasons": [
        "Public web/domain context is present.",
        "At least one contact email is present."
      ],
      "blocking_gaps": [
        "buyer_tech_stack",
        "budget_timing"
      ]
    },
    "useful_facts": [
      "Warehouse automation software appears in the caller-supplied company description.",
      "Contact emails: sales@example.com."
    ],
    "source_summary": [
      "url: https://example.com/about (confidence 0.75)"
    ],
    "next_agent_actions": [
      {
        "step": "qualify",
        "action": "Decide whether Acme Robotics fits the target account profile."
      },
      {
        "step": "fill_gaps",
        "action": "Resolve missing buyer_tech_stack and budget_timing before outreach."
      }
    ],
    "qualification_questions": [
      "What workflow, budget, or pain signal makes this account worth outreach now?"
    ],
    "missing_fields": [
      "buyer_tech_stack",
      "budget_timing"
    ],
    "proof_boundary": "No-secret brief generated from caller-supplied/public inputs only; no private CRM, paid upstream API, or external enrichment lookup was performed."
  },
  "receipt": {
    "policy_version": "lead-brief-lite-v1",
    "receipt_id": "example-receipt-id",
    "input_sha256": "input-hash",
    "output_sha256": "output-hash"
  },
  "human_summary": {
    "headline": "Allowed: lead brief is ready.",
    "plain_english": "Boundary Guard used the caller-supplied facts, returned a compact brief, and marked what is still missing before outreach."
  }
}

Choose Lead Brief Lite vs Data Enrich Lite

What does the caller want back?

Choose /lead-brief-lite: The agent needs an outreach-ready or research-ready brief with useful facts, missing-field flags, and proof boundaries.

Choose /data-enrich-lite: The agent needs normalized fields and source extraction before another tool decides what to do next.

How complete is the caller input?

Choose /lead-brief-lite: There is already enough public lead context to summarize fit, caveats, and the next follow-up question.

Choose /data-enrich-lite: The caller mostly needs entity cleanup, URL/domain normalization, or contact/source extraction.

What should the buyer compare against?

Choose /lead-brief-lite: Compare it with StableEnrich-style raw enrichment when the buyer wants a usable brief rather than upstream-style field access.

Choose /data-enrich-lite: Use it as the lower-friction step before a brief, CRM write, or outreach workflow.

First paid call checklist

  • Use the Vercel preview first if a human or agent wants to inspect the response shape without payment.
  • Switch to the Render paid host when the caller is ready to settle an x402 payment and keep Boundary Receipt evidence.
  • Send only caller-supplied or public lead facts; do not send secrets, private CRM exports, or credentials.
  • Expect a brief plus missing-field flags, source summary, proof boundary, and receipt hashes.
  • Track whether the paid call was internal seed proof or a non-seed buyer action before counting it as revenue.

Job to endpoint map

Prepare a proof-safe lead brief from public/caller-supplied facts

POST /lead-brief-lite at $0.03

brief.headline, buyer_fit_signal, useful_facts, source_summary, next_agent_actions, qualification_questions, missing_fields, proof_boundary, and receipt.

{
  "entity_type": "company",
  "query": "Acme Robotics sales lead: https://example.com/about, sales@example.com, Raleigh, makes warehouse automation software.",
  "known_fields": {
    "name": "Acme Robotics",
    "location": "Raleigh, NC"
  },
  "brief_goal": "sales",
  "policy": {
    "no_secret_mode": true,
    "require_source_pack": true,
    "max_bullets": 5
  }
}

Pick an AI runtime/model/API route before spending

POST /runtime-route-preflight at $0.02

recommended_route, budget_summary, risk, findings, recommended_action, and receipt.

{
  "task": "Summarize a support transcript and produce JSON action items.",
  "max_budget_usd": "0.08",
  "data_sensitivity": "internal",
  "candidate_routes": [
    {
      "route_id": "fast-json-small",
      "provider": "Example Runtime",
      "price_usd": "0.01",
      "capabilities": [
        "json-mode"
      ]
    },
    {
      "route_id": "deep-private",
      "provider": "Example Runtime",
      "price_usd": "0.06",
      "capabilities": [
        "private",
        "long-context"
      ]
    }
  ]
}

Verify a receipt package or payment evidence before trusting an API result

POST /receipt-verify at $0.03

decision, findings, verified fields, recommended_action, and receipt.

{
  "receipt": {
    "policy_version": "guard-v1",
    "receipt_id": "example"
  },
  "allowed_policy_versions": [
    "guard-v1",
    "workflow-audit-v1"
  ],
  "payment": {
    "network": "eip155:8453",
    "currency": "USDC",
    "amount_atomic": 10000
  }
}

Audit a multi-step agent workflow before it runs

POST /workflow-audit at $0.03

decision, findings, recommended_action, workflow receipt, and human_summary.

{
  "workflow_name": "lead outreach enrichment",
  "steps": [
    {
      "name": "enrich lead",
      "tool": "paid x402 enrichment API"
    },
    {
      "name": "send outreach",
      "tool": "email sender"
    }
  ],
  "budget": {
    "max_usd": "0.25"
  }
}

When to use Lead Brief Lite

Use it when

  • Caller already holds public lead facts (URL/email/name/location/note) and the immediate job is "produce a usable brief for the next agent step".
  • The buyer will pay $0.03 for the brief + source_pack + missing_fields + proof_boundary + receipt, then enforce the post-settlement guardrail: read receipt + proof_boundary + missing_fields before outreach/CRM/handoff.
  • The workflow needs auditable "what public facts + what guardrails" evidence that can be re-verified later via /receipt-verify or /workflow-audit.

Do not use it when

  • Do not use it as a replacement for private CRM access, Apollo, Exa, Firecrawl, Serper, or other upstream search/enrichment APIs.
  • Do not send secrets, private customer records, credentials, wallet keys, or confidential notes.
  • Do not call it before you have decided the route or when you only need raw normalized fields (use the cheaper /data-enrich-lite instead).
  • Do not treat output as third-party verified customer revenue, adoption, or endorsement evidence.

Competitor positioning

StableEnrich-style raw enrichment

Observed strength: High observed data-enrichment volume and buyer count: 93591 tx, 640 buyers.

Boundary Guard response: Lead Brief Lite packages caller-supplied/public facts into an immediately usable brief with missing-field flags, source summary, proof boundary, and receipt evidence.

Keep improving: Tighten examples, fields, marketplace copy, and proof-safe lead workflows until non-seed usage appears while repeating the low-friction promise: pay per request, no API keys, no subscriptions.

AgentMail-style communication utility

Observed strength: Clear agent job and repeatable workflow surface: 1657 tx, 41 buyers.

Boundary Guard response: Position Lead Brief Lite as the preparation step before outreach, with no-secret handling and source-bound claims.

Keep improving: Add routing copy for sales agents, CRM agents, and research agents when revenue or usage data shows which buyer segment responds.

BlockRun/YOPO-style outcome pricing

Observed strength: Outcome-based model and tool routing is pulling major runtime volume: 347181 tx, 378 buyers.

Boundary Guard response: Keep Boundary Guard focused on the decision-support layer around spend: what to call, when to call it, and what proof the buyer keeps after the call.

Keep improving: Borrow the low-friction language, but avoid implying broad model access, upstream search access, or outcome routing that Lead Brief Lite does not provide.

Buyer objections from current x402 chatter

What do I get after payment besides another blob of enrichment fields?

After x402 payment clears you receive a compact, decision-ready brief + source summary + explicit missing_fields + proof_boundary + receipt (with input/output hashes) + human_summary. The guardrail contract: inspect the receipt and proof_boundary before any agent acts on the lead.

How is this different from StableEnrich-style raw enrichment?

StableEnrich-style sells broad upstream field access. Lead Brief Lite sells the $0.03 brief packaging + missing-field flags + proof boundary + receipt evidence layer on top of caller-supplied public facts. Use it when the next step is "use this for outreach prep" and the buyer wants a usable artifact rather than "give me more raw rows to decide later".

What if the brief is incomplete or the agent should not trust it yet?

The design makes incompleteness and trust limits explicit: missing_fields, proof_boundary ("no-secret... no private lookup was performed"), retry/block decision, and the receipt you keep after settlement. Post-payment guardrail: read those before acting or chaining.

Learning loop

Cadence

Every factory cycle and automation run

Inputs

  • x402scan service rows, category volume, transactions, and buyer counts
  • Agentic Market calls, endpoint detail, provider/category enrichment, and listing copy
  • CDP merchant discovery resources and seed propagation state
  • Render/Vercel endpoint health, x402 challenges, docs, OpenAPI, llms.txt, and well-known inventory
  • Hermes VPS seed ledger and non-secret transaction/receipt proof
  • state/last-run.json and state/improvement-history.jsonl from the LarryBuildsAI x402 factory
  • product bugs, stale claims, pricing friction, successful changes, failed assumptions, and customer-facing gaps

Improvement dimensions

  • product usefulness and output quality
  • agent discovery through OpenAPI, llms.txt, /.well-known/x402, CDP, Agentic, and x402scan
  • buyer-facing copy, examples, and quickstarts
  • pricing, packaging, and margin safety
  • proof boundaries, receipts, and trust evidence
  • visual polish, page scanability, and mobile readability
  • deployment reliability, paid challenge health, and monitoring
  • revenue signals, repeat buyers, and non-seed transaction evidence
  • automation handoffs, goal quality, and coder/operator follow-through

History feedback rule

Each run must reuse the strongest successful pattern, avoid at least one known failure pattern, and improve or pressure-test one dimension. If it cannot act, it must record the blocker and next proof needed.

Required output

Record what changed, what worked, what failed, what proof moved, and the next measurable improvement. If no metric moved, improve conversion, distribution, product usefulness, docs, metadata, or verification coverage.

Measurement

  • non-seed paid transactions and unique buyers
  • x402scan rows and observed external volume
  • Agentic l30d calls and estimated revenue
  • CDP and Agentic endpoint propagation
  • paid route 402 challenge health
  • buyer-guide and marketplace-copy freshness

Current hypotheses

  • Lead Brief Lite should convert better than raw enrichment because it gives sales and research agents an immediately usable brief instead of normalized fields only.
  • Runtime Route Preflight should attach to high-volume model/API spend if listing copy names budget, latency, and policy-routing jobs directly.
  • Receipt Verify and Workflow Audit should be marketed to x402 API sellers and marketplaces because they answer trust-after-payment questions that simple payment gateways do not solve.
  • Do not raise prices until non-seed paid usage appears; reduce friction, improve listing copy, and add proof-safe examples first.

Proof boundary

All current paid endpoints are live, internally seeded, and visible in source/CDP/Agentic discovery. This guide is a conversion and routing aid, not a claim of customer revenue, customer adoption, marketplace ranking, or third-party endorsement.