Economico for AI startups
Know what it costs to serve your customers
Credit packs and subscriptions bring in revenue. Model and cloud usage create costs. Put both sides on the books so the margin question has a real starting point.
Your API bill does not say which customer caused it
Suppose a Scan action calls a model, while every Discover tap calls a paid maps API. A customer who opens Discover daily may cost more to serve than one who scans occasionally, even on the same plan. Vendor bills tell you what you spent, but not which action or customer caused it. Economico can record the usage measurements and vendor costs you supply; it does not trace your product's API calls.
Inspect the economics, not just the top line
A customer may buy credits before using them, while model and other API bills arrive later. Your agent can show what was sold, consumed and recognized, what a vendor charged, and which credit allowance was used. To estimate margin by action, plan or customer, you still need to link the relevant delivery costs to that usage.
Show the usage and prepaid capacity consumed by this customer, the revenue recognized and the vendor costs attributed to the period.
Bring the usage data you trust
Economico does not meter your product, find the paid services it calls, or fetch provider invoices automatically. Your agent or application supplies measurements and costs; you review their classification and attribution before relying on a margin figure.
An AI startup's first quarter · in journal entries
Credits in, provider bills out — see the period result.
Scroll through Brightforge AI's first quarter — plans, credit packs and vendor bills recorded in one ledger. Margin by plan still needs those costs attributed to the plans they served.
Incorporate and raise
Selene and Matias found Brightforge and close the YC standard deal. Half a million in — none of it revenue.
7 tool calls
Economicocontracts.create✓
{ "ledgerBindings": { "money": "led_00000000000000000000000840" }, "parties": { "founder-a": "selene-brooks", "founder-b": "matias-ortega" }, "templateDocumentId": "doc_01KBBKB1006DMSKYKM0Q9M4ZPM", "terms": { "authorize": { "shares": "12000000" }, "buy-founder-a": { "par": "50000", "premium": "1200000", "shares": "5000000" }, "buy-founder-b": { "par": "50000", "premium": "1200000", "shares": "5000000" } } }Economicocontracts.record✓
{ "activityKey": "buy-founder-a", "contractId": "ctr_01KBBKB100D1C7TST6ZKM5Z9D4", "effectiveAt": "2025-12-01T00:00:00Z", "facts": { "bank": "mercury-checking" }, "occurrenceKey": "buy-founder-a", "sourceFactId": "founder-stock:selene-brooks" }Economicocontracts.record✓
{ "activityKey": "buy-founder-b", "contractId": "ctr_01KBBKB100D1C7TST6ZKM5Z9D4", "effectiveAt": "2025-12-01T00:00:00Z", "facts": { "bank": "mercury-checking" }, "occurrenceKey": "buy-founder-b", "sourceFactId": "founder-stock:matias-ortega" }Economicocontracts.create✓
{ "ledgerBindings": { "money": "led_00000000000000000000000840" }, "parties": { "investor": "y-combinator" }, "templateDocumentId": "doc_01KCFMX300HPV6ZJEJBR2WG6ZZ", "terms": { "fund": { "amount": "12500000" } } }Economicocontracts.record✓
{ "activityKey": "fund", "contractId": "ctr_01KCFMX300NP5WPGQW5D8211D7", "effectiveAt": "2025-12-15T00:00:00Z", "facts": { "bank": "mercury-checking" }, "occurrenceKey": "fund", "sourceFactId": "mercury:wire:yc-post-money-safe" }Economicocontracts.create✓
{ "ledgerBindings": { "money": "led_00000000000000000000000840" }, "parties": { "investor": "y-combinator" }, "templateDocumentId": "doc_01KCFMX300HPV6ZJEJBR2WG6ZZ", "terms": { "fund": { "amount": "37500000" } } }Economicocontracts.record✓
{ "activityKey": "fund", "contractId": "ctr_01KCFMX300BJDZS3VAD0PKTRKH", "effectiveAt": "2025-12-15T00:00:00Z", "facts": { "bank": "mercury-checking" }, "occurrenceKey": "fund", "sourceFactId": "mercury:wire:yc-mfn-safe" }
The cost side arrives first
OpenAI for inference, LangSmith for observability — the vendor contracts are already in the inbox.
4 tool calls
Gmailsearch_gmail_messages✓
{ "q": "from:(openai.com OR langchain.com) (agreement OR terms OR welcome)" }Gmailread_gmail_message✓
{ "messageId": "19b4f1c2e8a07d3b" }Economicoparties.create✓
{ "categories": [ "vendor" ], "name": "OpenAI", "partyId": "openai", "url": "https://openai.com" }Economicoparties.create✓
{ "categories": [ "vendor" ], "name": "LangSmith", "partyId": "langsmith", "url": "https://smith.langchain.com" }
Plan plus credits, billed and paid
Fern Logistics takes the AI Workflow plan with a credit pack it used up in January — $3,200, collected within two weeks.
2 tool calls
Economicocontracts.record✓
{ "activityKey": "invoice", "contractId": "ctr_01KDY02100TJBWGSH7DW3YA286", "effectiveAt": "2026-01-05T00:00:00Z", "facts": { "credit_pack": "70000", "credits_used": "70000" }, "occurrenceKey": "2026-01", "sourceFactId": "BFA-INV-001" }Economicocontracts.record✓
{ "activityKey": "pay", "contractId": "ctr_01KDY02100TJBWGSH7DW3YA286", "effectiveAt": "2026-01-16T00:00:00Z", "evidence": { "payment": [ { "kind": "other", "ref": "mercury:ach:fern-2026-01-16" } ] }, "facts": { "bank": "mercury-checking", "invoice": "BFA-INV-001" }, "occurrenceKey": "BFA-INV-001", "sourceFactId": "mercury:ach:fern-2026-01-16" }
The token bills land
OpenAI $1,450 and LangSmith $399, paid from the Mercury checking account. The books show the period's recorded revenue and costs, but the bills are not yet attributed to a customer or plan.
4 tool calls
Gmailsearch_gmail_messages✓
{ "q": "(receipt OR invoice) after:2026/01/01 before:2026/02/01" }Gmailread_gmail_message✓
{ "messageId": "19c0a7d51e3b9f42" }Economicocontracts.record✓
{ "activityKey": "bill", "contractId": "ctr_01KDY0210034VDXQ3QN559GJ0R", "effectiveAt": "2026-01-31T00:00:00Z", "evidence": { "service": [ { "kind": "other", "ref": "doc_01KG8NJW00BMPMYTBVVD8GT0A7" } ] }, "facts": { "amount": "145000", "bank": "mercury-checking" }, "occurrenceKey": "2026-01", "sourceFactId": "openai:receipt:2026-01" }Economicocontracts.record✓
{ "activityKey": "bill", "contractId": "ctr_01KDY02100T15ZR9JRFVBQAZR1", "effectiveAt": "2026-01-31T00:00:00Z", "evidence": { "service": [ { "kind": "other", "ref": "doc_01KG8NJW00GPRFSMVEC2W8YKRA" } ] }, "facts": { "amount": "39900", "bank": "mercury-checking" }, "occurrenceKey": "2026-01", "sourceFactId": "langsmith:receipt:2026-01" }
February: growth goes red
A second customer, a token overage — and a $5,500 prompt-engineering bill that outruns them both.
2 tool calls
Economicoreports.run✓
{ "from": "2026-01-01", "kind": "income_statement", "until": "2026-03-01" }Economicoreports.run✓
{ "as_of": "2026-03-01", "kind": "balance_sheet" }
The quarter closes in the black
Three customers and token prices that moved twice — the quarter's result is visible, while per-plan cost attribution remains to be done.
2 tool calls
Economicoreports.run✓
{ "kind": "revenue_summary", "period_end": "2026-03-31T23:59:59.999Z", "period_start": "2026-01-01" }Economicoreports.run✓
{ "from": "2026-01-01", "kind": "income_statement", "until": "2026-04-01" }
Ask whether each plan still clears its cost
Put Fern Logistics on the AI Workflow plan — $2,500 a month plus a credit pack.
Record OpenAI's $1,450 January inference bill as cost of revenue.
Show this quarter's recorded revenue and delivery costs. Which costs still need attribution before we can estimate margin by plan?
What you'll use
The money loop, tuned for AI startups.
Get started
Books your agent runs. Numbers you can explain.
Hand your agent the setup guide, approve its connection, and start with one real transaction. Inspect the books as you go.