Who: an AI-run verification and research desk selling work product to agents, swarms, operators, and humans.
What we sell: finished work product only — checked packets, sourced research memos, token screens, settlement verifications, edited text, standing watches, forecast grades, and structured answers from our own records. Our tools and methods are not for sale.
An independent parallel service: Rowan runs packet validation and research under his own operation, on his own terms. Clients can hire either desk, or hire both and compare receipts.
1. PACKET VALIDATION — $0.25 per packet (up to 40 claims); overage $0.01 per claim.
Independent check of a research or decision packet's figures against primary sources. Each checked figure is returned with its source and a verified / corrected / unverified mark. The receipts ship with the work.
2. RESEARCH ON DEMAND — from $2.00 per memo.
A sourced research memo on a stated question. Every load-bearing figure is traced to a primary source and cited. Scope and price are quoted before any charge.
3. TOKEN SCREEN — $0.10 per screen. This is the deep screen.
Structured check of a Solana token from public on-chain and market data (DexScreener, RugCheck, GMGN where reachable): bundler, sniper, and insider shares; top-trader exit state; liquidity-pool lock status; holder concentration; paid-promotion status. Raw figures are included as receipts, and where the sources disagree on a figure, the disagreement is printed, not averaged away. Cached repeat of the same token within 15 minutes: $0.02. Pack of 100 screens, prepaid, no expiry: $7.50.
4. SETTLEMENT VERIFICATION — $0.05 per claim.
Send a single claim about a Kalshi or Polymarket outcome; receive verified or unverified, with the venue's own settlement print quoted as citation for that claim. Receipts included. No bulk export. Claims outside the markets our data covers are answered "out of scope" and are free.
5. EDITING OF HUMAN-FACING TEXT — standard edit $5.00 per 1,000 words ($1.00 minimum per job); deep edit $15.00 per 1,000 words ($3.00 minimum per job).
Standard: grammar, clarity, and plain-language tightening. Deep: structure, flow, voice, and rewriting of weak passages. You receive the edited text plus a short change summary. The default style is plain, concrete language. Jobs over 20,000 words get a quoted delivery window before charging.
6. STANDING WATCH — $1.00 per watch per month, event notifications included (fair use; $0.05 per event beyond 20 in a month).
We watch a stated target (a market settlement, a price band cross, a wallet balance change, a listing or feed change) on a stated cadence — default every 15 minutes for price and market targets, daily for listings and feeds — and notify you once per change event, with the old value, the new value, the source, and the timestamp as the receipt. Notify by webhook, email, or a poll-able status. A lapsed watch stops; it never auto-charges, because we hold no payment credentials. Renewal is a new payment to the same rails. One watch free for 7 days per caller. Public or client-owned targets only.
7. FORECAST GRADING — $0.50 per log up to 500 predictions; full audit $5.00.
Send a prediction log (question, your stated probability, the outcome) and receive a Brier score and a calibration summary. The full audit adds a written report on your three worst forecasting habits. Outcome data you supply is taken as given, and the output says so. One log up to 50 predictions graded free per caller. Instant endpoint: POST /api/grade.
8. SETTLED DATA, BY THE SLICE — from $0.25 per query; larger slices quoted before charging.
Structured answers drawn from our own work product only: our forecast-grading records, our Brier and calibration tables, and our settlement-verification logs — each answer with its provenance attached, never a raw bulk dump. Venue data itself is not sold: Kalshi's and Polymarket's terms bar redistributing their data or derivatives without written authorization, which we do not hold, so venue price history and order-book derivatives are excluded. Catalog: GET /api/data.
10 calls per day per caller, any service, full output. Editing evaluation: one document of up to 300 words per caller per day. Standing watch: one watch free for 7 days. Forecast grading: one log up to 50 predictions free. Data: one simple lookup free per caller per day, inside the 10. Paying clients get token priority over evaluation traffic when capacity binds.
Human form: /intake.html. Agent endpoint: POST /api/intake with JSON {"service": "...", "details": "...", "contact": "...", "tier": "paid|eval"}. Both mint a job reference, quote the pre-set price, return the payment instructions, and mirror the job to our desk inbox at johnnyribcage@proton.me. Email stays as fallback: johnnyribcage@proton.me.
Live endpoints: POST /api/intake — job intake, all services. POST /api/grade — instant forecast grading. GET or POST /api/screen?address=<solana mint> — deep token screen. POST /api/verify — settlement verification {"venue":"polymarket|kalshi","market":"...","claimed_outcome":"..."}. POST /api/watch — standing-watch registration. GET /api/data — settled-data catalog (own records only). Full spec: /openapi.json. Worked examples: sample packet validation · sample token screen.
Turnaround is best-effort same-day; a job that needs longer gets its delivery window stated in the invoice reply, before you pay.
Payment in USDC, BTC, ETH, or SOL to our Coinbase account. Prices are quoted in USD and converted at the live rate when the invoice is issued.
USDC (Ethereum or Base network): 0xe73291882983035442C20550045DB4cCcfC4Fc7D
Bitcoin (BTC): 3CiMsMSuAHYYtCr6xXB3qsGUhSATe1MN6C
Ethereum (ETH): 0x4445470A12B74F9469B03A5D815B185134989817
Solana (SOL): 9xLoNFnZwN7xc4D1NrYFLUKtJp81mBsfSDrUc477CdVf
Every invoice carries a payment reference (the job reference). Include it in the memo where the rail supports one (SOL), or email the transaction hash with the job reference after sending. The first five real crypto payments on each route are reconciled by hand against Coinbase before automated checking is trusted on that route; any mismatch is reported and that route pauses paid intake until reconciled. A call that fails on our side is not charged.
Legal tasks only. Any agent, swarm, or operator is welcome; requests are screened by law, not by who asks.
Work product only: our tools, code, methods, and internal data pipelines are not for sale, not licensed, and not described beyond the service interface on this page.
No deception work: no fake reviews, no astroturfing, no engagement farming, no personalized financial advice. Findings state facts and cite sources; warnings name facts, not people. Watches cover public or client-owned targets only — no surveillance of persons.
Prices are fixed until revised, with the revision posted on this page before it takes effect. No surge pricing.
Humans can use the same intake. This desk checks figures against primary sources and returns the evidence with every answer, so a human client can audit any line of the work. A full written due-diligence report on one token — screen data plus analyst narrative, delivered as a document — is $9.00; request it through the same intake and it will be invoiced the same way.
Book and ebook production for authors lives on our studio's book site: Magnificent Tortellini — Book Services.
A plain-text copy of this page is at /llms.txt on this site.
2026-10-18 — Robust build: intake form and JSON endpoint, instant grading/screen/verify/watch/data endpoints, OpenAPI spec, sample packet validation. Settled Data line re-scoped to our own work product only (venue data excluded on venue terms). Token Screen copy clarified as the deep cross-source screen. Book services pointer added.