See the scoped proof surfaces running. Receipts record submitted inputs, policy results, and hashes for integrity review; illustrative surfaces are labeled, and a receipt does not independently prove what an agent did.
Agents block a supplier, flag a transaction, or ship a change — and weeks later the evidence, the rule, and the reasoning are gone.
Submitted evidence, policy evaluation, action metadata, and verdict are recorded with hashes. The receipt makes that recorded scope reviewable; it is not proof of unrecorded agent behavior.
Auditors, regulators, and customers are asking what record exists. A receipt can show the submitted inputs, policy verdict, and integrity scope — no more.
The category shift
What you actually get
See the artifact — click any field
Illustrative Decision Receipt with synthetic values — not a customer result, certification, or performance claim. Flip to Live for the current scoped audit, or run the input-driven live proof. Integrity results apply only to the records and checks shown.
Why this matters today
AI blocks a supplier. Three weeks later, no one can explain why — no evidence, no rule, no record to reopen.
The same block leaves a Decision Receipt behind it:
The flagship Summit explainer — narrated, illustrative and synthetic throughout. One clean film for now; more on the way.
Field guides — the ideas on one page each
The Policy Gate · New
The Tamper Test · New
From Call to Receipts · New
Deterministic Replay · New
The Proof Estate · Map
The Verification Path
Receipt Anatomy
Category Landscape
Accountability Gap
Proof Ladder
Pilot Economics · Interactive
Action Categories
Persona Map
Regulatory Timeline
Commercial Ladder
All 21 guides →
You do not need to understand Summit's product map first. Tell us your mission, the first outcome that matters, and the data boundary. We will route you to the right demo and produce a bounded evaluation brief.
Choose the closest fit. You can change it later.
Pick one primary outcome so the evaluation stays decisive.
This sets the evaluation's initial data and deployment assumptions.
Illustrative evaluation design—not a performance promise, customer result, certification, or authorization claim. Final scope, measures, security controls, and acceptance criteria require buyer agreement.
This page presents recorded claims with stated verification scopes. Summit Pass is an OAuth 2.1 gate in front of configured MCP servers and agent tools; for calls actually routed through that gate, it can issue a receipt recording submitted inputs, policy evaluation, and hash material. Start bounded and expand only on measured evidence.
Hosted private beta. Commercial terms are confirmed in a written order form. Receipts record submitted inputs, policy verdicts, and integrity material; they do not independently prove that an agent performed an asserted action or certify that an AI output is correct, safe, or lawful.
Not demos — working products with real cryptography behind them. Each card states its status and how to check it yourself; availability last checked July 20, 2026.
For configured workflows, Proof Mode can pause an action for human review and record the submitted action metadata, approval event, policy verdict, and integrity hashes in a scoped receipt.
A receipt ledger for recorded inputs, policy checks, verdicts, and hashes. The current public audit reports valid chain continuity but failed artifact integrity with seven errors; historical signatures are not included.
Win the pilot before it becomes theater. Fixed-scope, fixed-price pilots with verifiable outcomes — decision receipts and deterministic replay on every deliverable.
Compare a published hash and stated provenance with the public transparency record. No account is required; conclusions remain limited to the available record, algorithms, and key material.
Quickstarts, API reference, and integration guides for the Decision Receipt API — admissibility evaluation and signed receipts, from your first call.
Selected claims on this page link to checkable records. Start with the input-driven live proof, then inspect sample hashes and the current degraded production audit. These checks establish recorded-input, policy, and hash integrity properties only.
Provide a unique claim, observe real read-only HTTP tool boundaries, submit the exact input to the non-production evaluation route, retrieve the receipt, and verify it. Run the live proof →
Sealed demo Decision Receipts, including a deliberate failure case — inspect the fields and run the verification path. Open the gallery →
An agent takes four consequential actions through a deny-by-default policy gate; each seals into a real WebCrypto SHA-256 hash chain in your browser. Break it and watch it fail. Run the simulation →
Paste any Summit output hash into the public verification portal — integrity, timestamp, provenance, no account. Trust Portal →
The public audit currently reports chain continuity true, artifact integrity false, seven errors, and historical signatures not included. Current audit JSON → · API docs →
Ishtar, the glass-box console where receipts, policies, and replay meet review — as a working demo on synthetic data. Open Ishtar → · console walkthrough →
LIVE = running now, checkable. BETA / PREVIEW = working, bounded access. ILLUSTRATIVE / SYNTHETIC = honest demo material, never customer evidence. Full legend →
Working and preview environments that can record selected inputs, policy results, and receipt hashes. Coverage is limited to instrumented paths and stated scopes.
A threat-intelligence investigation graph — APT actors, CVEs and ATT&CK techniques resolved into one graph, with selected source records, policy results, and receipt hashes preserved for review. The receipt does not independently prove the analyst or agent's reasoning.
A local-first recorder and live conversation guide — explicit consent, on-device capture, and decisions, commitments and risks extracted into a clean Markdown/JSON export. Nothing leaves the room without you.
Connect an identity and run mock evaluations — watch policy checks, evidence paths, and replay-grade certificates generate in real time.
Submit a unique non-production claim, observe real read-only HTTP calls with measured timings, retrieve the server-issued receipt, and run its verification path. The page states exactly what the workflow does and does not prove.
End-to-end autonomous media operations — production, editorial, rights and distribution, orchestrated and auditable at scale.
Forensic audio intelligence — fingerprinting, metadata extraction, and tamper-evident receipts on sound itself. Machine narration and long-form audio, the omnimedia pipeline out loud.
Interactive walkthroughs of the core ideas — built to be understood in one sitting.
Compare one customer-support agent workflow with and without assurance: evidence captured, handoffs linked, policy checked, action recorded, and one reviewable Decision Receipt created.
Scripted agents call configured demo tools against generated traffic; the demo records submitted inputs, nine-rule policy results, receipt hashes, tamper checks, and an illustrative month-six recall. It does not show a customer system of record.
An illustrative customer-call workflow: selected action inputs pass through a configured policy gate and are recorded into a synthetic hash chain for inspection, tamper testing, and deterministic replay.
The main surfaces as presentable decks — take them into any room: VOX Paid Pilot · Proof Mode · Decision Receipt · IntelGraph · one-pager (PDF)
A bounded 30-day paid pilot: instrument one agreed AI code-assistant path so observed actions can emit scoped Decision Receipts. Illustrative material uses synthetic repositories and events; no customer engagement is implied.
Instrument one flow — a pipeline fix, guarded merge, or controlled deploy — to record selected inputs, policy verdicts, and integrity hashes mapped to agreed control objectives. The receipt covers the instrumented record; it does not independently prove the agent's complete behavior. Pilot scope, schedule, pricing, and terms are confirmed in writing.
A coding agent diagnoses and fixes a failing CI/CD pipeline on a compliance-bound repo. Flip between Without Summit and With Summit, open any Decision Receipt, then tamper with the record and watch verification fail.
Where Summit sits beside your DevOps platform, how the thirty days run by owner, what a Decision Receipt actually contains, and the three ways to emit receipts — architecture, implementation, and a 30-day timeline.
The pilot on a single page — offer, scope, 30-day arc, and what the buyer holds at day 30. Built to forward to a compliance lead or hand across a table.
Real-world-shaped examples that make buyer conversations concrete — workflow, receipt moments, acceptance tests, artifacts, and decision gate. Illustrative scenarios only: not customer deployments, endorsements, performance claims, or compliance certifications.
Five demo scenarios for sales calls, technical discovery, and pilot scoping — AI code remediation, controlled document answers, control-evidence response, SOC escalation, and delegated personal-agent action. Each carries its own claims boundary.
Compose a buyer-specific pilot brief from any of the five scenarios — set the workflow, owners, decision date, and excluded data, then copy, download, or print. Runs entirely in your browser; nothing is submitted.
One diagram per pilot workflow: the buyer's action path left to right, the moment a signed Decision Receipt is sealed, and the independent verification an assessor can run afterward. Screen-share a single row.
Put in your own assumptions — reconstruction hours, loaded cost, actions in scope, dispute rate — and see the monthly cost of unprovable actions against a fixed-scope pilot. Runs entirely in your browser; illustrative estimate, not a savings promise.
The end of assumed truth — the book on legitimacy and accountability in machine-mediated decisions. The thesis everything on this page implements.
The flagship narrative podcast & microsite — the thesis, dramatized.
The Summit audio slate — long-form conversations on evidence, justification and the machinery of accountable decisions.
Essays from the front line of cognitive infrastructure — 27 and counting.
The working-paper series — decision admissibility, replay as an accountability primitive, provenance before inference, standing to contest, and the NIST AI RMF crosswalk. First paper published; the series is rolling out.
Cognitive supply chains — a bill of materials for thought. Supply-chain risk intelligence, applied.
Fact sheet, boilerplate, founder bio and press releases — everything a journalist needs, verifiable.
The Summit design bundle — the on-canon visual system behind every surface above: Cabin and Space Mono, the Ink / Verified Green / Paper palette, the summit mark, and the Decision Receipt as the recurring proof artifact. Design references only — sample data, receipts, and figures are illustrative and are not customer results, certifications, or performance claims.
Twenty-one public-safe, responsive field guides — including the Policy Gate, the Tamper Test, Deterministic Replay, the self-service Summit Pass path, the Proof Estate map, the Verification Path, receipt anatomy, pilot economics, and the regulatory timeline.
The visual catalog of the design system — every artifact, mark, and layout in one on-canon index. The map to the collateral set.
Terms, the status legend, color roles, and diagram conventions — the shared vocabulary behind every Summit artifact.
The summit mark component — the canonical monogram used across the system. A brand asset, not an illustrative claim.
The flagship scrollable explainer — how a decision becomes a reviewable Decision Receipt, walked end to end.
The core pitch as a presentable 16:9 deck — the problem, the Decision Receipt, and the workflow, room-ready.
The narrative for investors — thesis, category, and the Decision Receipt as the central artifact. Illustrative figures only.
Six reusable on-canon slide layouts — the template set behind the decks.
The single-page overview — what Summit records and the Decision Receipt an evaluator can open and replay.
The partner-facing single page — how Summit fits beside a partner offering, on one sheet.
The technical reference — receipt fields, policy outcomes, and verification, for a technical reviewer.
How Summit integrates — the emit paths and the Decision Receipt API surface, at a glance.
The cover sheet for the buyer trust pack — the on-canon front for a leave-behind set.
An illustrative sanctions-escalation template — a worked example shape, not a customer deployment or endorsement.
The 18×24 anatomy of a Decision Receipt — every field of the recurring proof artifact, annotated.
Square and banner frames for social — on-canon, with illustrative sample copy.
Outreach and follow-up email templates plus a signature block — the sales text kit, on canon.
