A clean-sheet pricing architecture for a signals-producing platform with one human application, one API, one MCP surface, and one closed intelligence layer.
The prior four-meter idea was elegant taxonomy, but weak pricing. It tied price to four different nouns and invited arbitrage across app, API, and MCP.
Seats punish collaboration and do not exist for machine callers. Identity remains essential for audit, but should not be the core meter.
Kinetic cannot defensibly attribute world outcomes dollar for dollar. It also corrupts neutral infrastructure and breaks API economics.
Credits normalize heterogeneous work but obscure forecastability. Use a dollar balance with simple published work classes.
Bill only when one Subject × one Verb version produces a new canonical Result version committed to the Ledger. A failed, degraded, unchanged, or below-coverage Result is free.
The same work costs the same whether initiated by a human, API, or MCP. Source rights may make a Verb app-only. Channel entitlements can differ, but the work rate does not.
Publish the machine rate architecture later. Keep committed application and government pricing private until real WTP exists.
The account establishes the relationship. The shared balance makes spend predictable. But the customer chooses Kinetic because each Verb deploys a specialist investigative unit and returns deterministic, provenance-bound Components.
Build the canonical path across wallets, hops, endpoints, and source Records.
RETURNS · TRACE MAP · TIMELINE · ENDPOINT SIGNALSCheck a wallet, party, or counterparty against sanctions, risk, and control evidence.
RETURNS · MATCH TABLE · RISK SIGNALS · SOURCE RECORDSFind hostile domains, leaked content, impersonation, and other external traces.
RETURNS · FINDING SET · ENTITY COMPONENTS · EVIDENCECanonicalize aliases, ownership, control paths, infrastructure, and linked Subjects.
RETURNS · ENTITY PROFILE · CONTROL PATH · CONFLICTSSynthesize a rights-safe operator profile from Records already permitted to meet.
RETURNS · ACTOR PROFILE · LINK MAP · SCORED INFERENCESRe-run a named Verb on an approved cadence and mint only new Result versions.
RETURNS · DELTA COMPONENTS · NEW SIGNALS · FRESH FILE| HERO | CANONICAL VERBS IN THE ROSTER | SOFTWARE CLASS | PRICE SURFACE |
|---|---|---|---|
| TRACE | crypto.trace · judgment.locate | Synthesis | each registered Verb has its own exact rate |
| SCREEN | crypto.screen · sanctions.screen · entity.screen · crypto.expose | Deterministic or Synthesis by manifest | rate shown before Run |
| SCAN | impersonation.scan · leak.scan | Deterministic | rate shown before Run |
| RESOLVE | entity.resolve · crypto.attribute | Synthesis | rate shown before Run |
| PROFILE | actor.profile | Synthesis | app/API entitlement follows source rights |
| WATCH | crypto.monitor · impersonation.monitor · sanctions.watch · leak.monitor · judgment.monitor · actor.watch | scheduled Deterministic or Synthesis | standing budget + cadence + hard cap |
| FILE TRANSFORM | evidence.assemble · evidence.seal | Builder/export function | included; assurance quoted only when ordered |
| WORLD ACTION | matter.freeze · matter.demand · matter.enforce | outside the universal signals contract | operator-gated service, never self-serve or MCP |
The catalog is not six generic price buckets. TRACE is the storefront family. `crypto.trace` and `judgment.locate` are separately purchasable canonical Verbs inside it, callable directly by API/MCP or selected in the app. A later TRACE bundle is a prepaid allotment of those same Verbs, never a second charge. Work classes remain internal economics and procurement shorthand.
There is no single direct competitor. Kinetic sits between four established price systems. The evidence below separates vendor-published facts, observed contracts, directional benchmarks, and our hypotheses.
| MARKET | WHAT IS SOLD | PUBLIC / OBSERVED CORRIDOR | DOMINANT UNIT | WHAT KINETIC LEARNS | EVIDENCE |
|---|---|---|---|---|---|
| Machine data rail | Deterministic lookup, search, screen, extraction | €0.10/query · $1.50/verificationdistinct vendor units, no FX normalization | successful request | AE Bill the canonical result, not compute or attempts. | VP 01–02 |
| Agent research rail | Search, extraction, multi-step research | $0.001–$2.40/requestvendor-native depth tiers | request or native work tier | AE Depth classes can be legible without exposing sources. | VP 03–04 |
| OSINT investigator | Analyst graph/search capability | €3K–€7.5K/user/yr | seat + pooled credits | AE Operator-tool floor, not Kinetic's ceiling. | VP 05 |
| Focused graph app | Graph visualization / investigation | €490/user/mo | seat | AE Public price may shorten procurement below enterprise. | VP 06 |
| Digital risk / brand defense | Detection, monitoring, takedown allowance | $25K–$75K ACV | protected noun + allowance | AE Sell scope, include work. Do not expose analyst hours. | DB 07 |
| Enterprise intelligence | Data, modules, graph, workflow | $50K–$500K+/yr | platform + modules + volume | AE Committed pricing is commonly private and annual. | DB 08 |
| Per-case forensics | Investigation report / case pack | $4K–$5K+/case | case or File | AE Directional operator WTP ceiling. | VP 09–10 |
| Evidence workflow | Review, storage, case collaboration | $3.50/GB archived → custom$3.50 is a GSA suspended-case item, not standard hosting | evidence volume, unlimited users | AE Unlimited-user models are collaboration-compatible. | OC 11–12 |
| Federal intelligence program | Software, data, training, support | $1.7M–$11M+ observedaward value or obligation; not annual list price | fixed-price program | AE Different procurement product, not a discount tier. | OC 13–14 |
| MCP marketplace | Thin tool access | No reliable paid corridor observed | subscription or tool call | AE Protocol adoption is real; paid demand is unproven. | DB 15 |
VP vendor-published · OC observed contract/spend · DB directional benchmark · AE Kinetic hypothesis. EUR is shown as published; no FX normalization. Contract ceilings are never treated as list price.
Seizure360 plus live MCP is closest. Pricing is not public. Public materials show an intelligence suite, not one deterministic contract from Verb to File.
Claims vendor-neutral orchestration across tracing tools and court-ready reporting. No public machine rail was found as of 2026-08-04.
Plaid supplies the two-door plan. Everlaw supplies unlimited-user workflow. Kinetic needs both, joined by one work balance.
The buyer landscape collapses into three economic jobs. This is the minimum segmentation that produces different budgets, procurement paths, and churn mechanics.
| ECONOMIC JOB | PRIMARY BUYERS | TRIGGER | REFERENCE ALTERNATIVE | VALUE EVIDENCE | RIGHT METRIC | FIRST HYPOTHESIS |
|---|---|---|---|---|---|---|
| Work a live record | Law firm, receiver, insurer, independent operator, brand team | A matter, incident, claim, judgment, or watch requires verified work now | Analyst hours, forensic report, digital-risk subscription, disconnected tools | Time-to-File, reproducibility, coverage, review throughput, audit completeness | account commitment + included work balance | Private pilot offer |
| Embed verified work | Developer, exchange, fintech, platform, agent builder | A product or agent needs a result it can cite and replay | Raw APIs + internal orchestration | successful canonical results, latency, provenance completeness, schema stability | successful Verb class | Pay as you go → annual commit |
| Deploy an institution | Government, large exchange, enterprise intelligence program | Program-wide workflow, sovereignty, accreditation, or fixed-budget need | Incumbent suite + integration/services contract | case throughput, auditability, deployment control, source rights, response service-level agreement | fixed annual program + capacity schedule | Custom firm-fixed-price order |
Jobs completed, successful Results, Signals included, sources cited, time-to-File, monitoring coverage, review time, reproduction, uptime, and audit completeness.
Dollars recovered, value secured, loss prevented, case won, funds frozen, or outcome caused by Kinetic. World outcomes remain outside the price equation.
One economic spine, two doors, two software work classes, and a separate human-services line. This preserves the one-platform thesis without pretending every buyer buys the same way.
| WORK CLASS | CUSTOMER PROMISE | EXAMPLES | BILLABLE EVENT | DISCLOSURE | PRICE STATUS |
|---|---|---|---|---|---|
| Deterministic | Bounded lookup, screen, or multi-source canonicalization | sanctions.screen, wallet screen, entity.resolve | new Result version above the coverage floor | typed input/output, provenance, freshness, limitations | Research band only |
| Synthesis | Rights-safe cross-record computation into richer Components | crypto.trace, actor.profile, pattern result | new File-ready Result version above the coverage floor | method version + complete citations; protected method available to counsel or auditor under controlled review | Research band only |
| Human assurance | Named technical validation or specialist source work | validation, custom source acquisition, specialist execution | accepted work order | separate services line; never legal advice, representation, or case-taking | Quoted |
The work classes are hypotheses, not rates. Start with two software classes; split them only after the cost ledger proves distinct distributions. One billable Result equals one Subject × one Verb version above its published coverage floor. Cross-record means the customer's tenant plus licensed sources. Tenant records never improve another tenant's Result unless both contribution and derivative rights are separately, explicitly licensed. File creation is not a toll. Components are typed units placed in a File; Signals are verified factual Components minted by synthesis. File export is included. Bulk raw-Result export is a rights-aware entitlement, never a surprise fee.
No Kinetic number should be approved before the internal cost ledger exists. Each Verb class must survive provider repricing, cache loss, review-rate shock, and low utilization.
Price to p90 job COGS, not only the mean. Fixed source licenses are amortized against actual volume. A zero-retention source has a cache rate of zero by contract. Human assurance stays in its separate service P&L.
Recommended planning floors: app software 80%+, machine work 65%+, human assurance reported separately. These are hypotheses to validate.
Billed cash is commitment + approved prepaid top-ups. Consumption cannot create revenue or exceed the purchased balance. Unused balance is not treated as recognized breakage. The color gate is machine-work basis: green ≥65%, amber 55–65%, red <55%. App software planning target remains 80%+.
| STRESS TEST | RUN IT AT | FAIL CONDITION | RESPONSE |
|---|---|---|---|
| Fan-out shock | top two source costs +50% | machine margin below 65%; alarm below 55% | reclassify Verb or renegotiate source; never expose source surcharge |
| Cache collapse | cache = 0 for rights-restricted sources | base case no longer clears | price from rights-safe baseline |
| Review shock | human touch rate 2% → 15% | software margin depends on unpaid labor | separate human assurance order |
| Low utilization | 25% of licensed source capacity used | fixed license destroys unit margin | delay license or set minimum commitment |
| Agent runaway | 3× monthly commitment in 24h | any unapproved overage posts | hard stop, scoped budget, explicit top-up |
| Tenant-rights wall | all tenant-derived contribution disabled | economics or quality depends on cross-tenant reuse | default tenant isolation; use only licensed sources and explicitly contributed derivatives |
| Rebuild arbitrage | raw-result exports high, File use low | buyers reconstruct closed layer | fix the Result boundary in each contract; future boundary changes apply only on renewal |
At zero paying customers, a giant survey is not rigor. The valid program begins with observed jobs and economic alternatives. Large-sample methods come only after a credible offer exists.
Interview user, economic buyer, and procurement gate separately. Do not use company size as a substitute for need. Recruit to saturation.
Reference alternative + measurable differentiation. Use analyst hours, vendor spend, File throughput, reproducibility, and audit risk. Never use recovery outcome.
Record unit, floor, included amount, overage, term, seats, services, trial, and procurement path. Never compare price points without structure.
Within each segment, hold offer shape and included balance constant and randomize only price. Ask for a signed order, not a hypothetical rating. Record objections verbatim.
Rank which outputs and controls truly earn a package gate. Does not set price. Use aggregate counts only below 200; do not claim segment-level utilities.
Estimate demand at price cells only when the offer is concrete and the sample can support it. Direct-stated willingness to pay carries hypothetical bias. Treat output as a ceiling and calibrate it against paid acceptance.
Test package-price tradeoffs with a None option, 8–15 tasks, and sufficient sample. Never run a 25-person conjoint and call it evidence.
Use quote-to-close, paid activation, balance burn, expansion, second File, and renewal cohorts. The invoice becomes the truth.
| MEASURE | WHY IT MATTERS | DAY-ONE EVENT | DECISION USE |
|---|---|---|---|
| Paid acceptance by offer cell | real WTP, not stated intent | offer.quoted / offer.accepted / offer.lost | base commitment and discount ceiling |
| Work balance burn | forecastability and margin | work.debited by class | included amount and top-up size |
| Second File within 60 days | platform repeat, not one-off report | file.exported | activation and retention |
| Signals per successful Verb | synthesis yield without claiming outcome | signal.minted | class boundaries and product quality |
| % Results touching retained tenant records | rights-safe ledger contribution, not a causal improvement claim | result.enrichment_applied | ledger use and renewal analysis |
| Manual review touches | hidden service COGS | review.opened / completed | engineering priority and assurance pricing |
| Price objection code | diagnose unit vs amount vs procurement | deal.loss_reason | packaging changes, not reflexive discounts |
This is the first behavioral pricing evidence before scale, not a rate card. Within each segment, offer shape and balance stay fixed; only price moves.
| CELL | WHO | FIXED OFFER SHAPE | RANDOMIZED PRICE CELL | N | SUCCESS RULE |
|---|---|---|---|---|---|
| A · Operator | boutique investigator / small firm | month-to-month app, unlimited viewers, $1,000 work balance, one legal entity | $1,250 / $1,500 / $1,750 per month | 4 | paid order; balance forecast understood without explanation |
| B · Firm | law / recovery / receiver team | annual app, unlimited viewers, $18,000 work balance, standard assurance access | $24K / $30K / $38K per year | 4 | paid or signed pilot order; no per-identity objection |
| C · Machine | developer / platform | $100 funded production balance, two work classes, hard cap | three COGS-safe rate-card variants | 2 | funds balance and accepts displayed class rate before production Result |
Three qualified buyers understand the value but reject the amount by more than 40%, with the same budget evidence.
Three of the first ten cannot forecast the work balance, or buyers spend more time disputing class definitions than evaluating the File.
Three independent buyers accept the same upper price cell without concession. Margin headroom alone is not evidence of willingness to pay.
These rulings set the research architecture. They do not approve a public rate card, Stripe change, contract, or launch.
Selected load-bearing sources. Accessed 2026-08-04. Research bands remain directional unless a vendor page or observed contract is explicit.