Evidence Fabric
Captures tenant- and case-scoped records with source context, append-only Evidence Events, governed Tool Invocations, artifact integrity, lifecycle lineage, and minimized provenance.
Design-partner stage · Deployed on AWS · limited onboarding
Make consequential AI actions provable.
DigiTrust is the evidence and authority system of record for enterprise AI.
DigiTrust preserves what agents and systems did, verifies what supported the result, enforces accountable human approval, and produces independently reproducible assurance across models and clouds.
Reference platform through durable approval and remediation. AWS Marketplace publication is pending. The public ChatGPT app remains synthetic and read-only.
Deterministic verification
Durable human approval
Exact-use authorization
Deployed on AWS
Platform model
Evidence Fabric → Verification Engine → Human Control Plane → Assurance & Audit
DigiTrust separates probabilistic AI execution from the deterministic evidence, authority, verification, approval, authorization, and audit controls required for consequential enterprise use.
Verification Engine
Applies explicit policy and claim criteria to assess eight evidence-quality dimensions, support, authority, separation of duties, contradictions, disclosure risk, and signed input integrity.
Human Control Plane
Binds accountable approval to the exact verified subject, preserves failed history through remediation, and blocks self-approval, authority drift, expiration, revocation, and target substitution.
Assurance & Audit
Produces customer-safe outputs and signed audit exports whose identity, evidence, verification, decisions, limitations, and current authorization state remain independently reproducible.
The Evidence Packet QuickStart remains DigiTrust's recommended commercial entry point. The current reference platform implements the evidence, verification, approval, remediation, authorization, and audit foundations. Durable packet-candidate issuance remains a Stage 4 production milestone.
Evidence quality and structured trust
DigiTrust verifies the quality, authority, and permitted use of evidence.
Eight quality dimensions use declared metadata, durable identity, lifecycle state, provenance, source information, and explicit policy criteria. DigiTrust does not infer truth from free-form prose: it compares typed, hash-bound assertions under explicit scope, time, authority, lifecycle, and classification rules.
Contradiction and disclosure controls are explicit and fail closed. Critical unresolved structured contradictions remain hard failures. Secrets, credentials, private keys, system prompts, and cross-tenant content cannot be downgraded by redaction metadata.
Trust lifecycle
Capture. Preserve. Verify. Resolve. Approve. Authorize. Reproduce.
A transparent, deterministic path turns authorized evidence into accountable enterprise use without hiding hard findings behind a score or allowing an agent to become the system of record.
Enterprise human-control plane
AI can perform the work. Accountable people still control consequential use.
DigiTrust separates generation, verification, and approval. Approval cannot be created by an agent, inferred from free-form text, transferred to another output, or used after expiration or revocation.
Exact approval
A human decision applies only to the exact tenant, case, subject, version, verification result, scope, conditions, authority, and time window that were reviewed.
Durable remediation
Failed verification remains visible. Correction creates new same-scope evidence and a later independently verified run rather than rewriting the original failure.
Replay-safe authorization
Changed content, substituted evidence, authority drift, expired approval, revoked approval, or a mismatched verification report fails closed at the exact-use boundary.
Agents cannot approve their own consequential output. Requestors, bound drafters, verifiers, unauthorized people, and non-human actors are rejected by the durable approval boundary.
AWS deployment and Marketplace progress
Deployed on AWS
DigiTrust completed AWS partner onboarding, validated its reference architecture through the AWS engagement, and deployed the control plane on AWS through an automated GitHub Actions delivery pipeline.
The AWS Marketplace Foundational Technical Review process is complete, with final results and listing publication pending. The draft Marketplace listing displays the exact designation “Deployed on AWS.” This does not claim AWS certification, endorsement, or that the listing is already public.
01 · DeliverGitHub Actions to AWSOIDC-based access, CodeBuild, Amazon ECR, and Amazon ECS Fargate provide an automated deployment path.
02 · GovernApproved Amazon Bedrock model pathsModel routing is explicit, inspectable, evidence-producing, and fail-closed rather than a hard-coded runtime shortcut.
03 · PersistAWS data and trust servicesAmazon DynamoDB, Amazon S3, AWS IAM, AWS KMS, and AWS Step Functions support the deployed reference pattern.
04 · OperateObservability and recoveryAmazon CloudWatch, AWS CloudTrail, EventBridge, SQS, SNS, and AWS Backup support operational visibility and resilience.
Ecosystem progress
One independent trust layer across cloud and AI ecosystems
DigiTrust's canonical evidence and authority contracts are designed to remain portable even when the cloud, model provider, agent runtime, or enterprise workflow changes.
AWS
Partner onboarding completed, control plane deployed, Marketplace FTR process completed, and listing publication pending.
OpenAI
Participating in the OpenAI Partner Network with partner training underway and a provable-enterprise-agent integration path in development.
Cloud-neutral architecture
Reference contracts are designed for extension across AWS, Microsoft Azure, Google Cloud, OpenAI, and other model or agent ecosystems.
Public trust surface
Explore a bounded evidence workflow inside ChatGPT
DigiTrust Evidence Assistant is a public trust surface where prospects, partners, security reviewers, and procurement teams can explore synthetic Evidence Packets and customer-safe summaries.
Demo boundary: synthetic evidence only, read-only tools, no paid workspace records, no billing, and no raw customer data.
Safe prompts to try
- List the synthetic demo Evidence Packets.
- Summarize a packet for a security reviewer.
- Generate a customer-safe assurance summary.
- Compare two synthetic readiness snapshots.
- Explain the DigiTrust evidence workflow.
Recommended first paid engagement
Start with one proof-ready AI workflow
The Evidence Packet QuickStart is DigiTrust's commercial entry point: one workflow, one governed evidence path, one deterministic verification boundary, one accountable approval path, and one customer-safe assurance deliverable.
The current reference platform implements the underlying trust controls. Durable packet-candidate issuance remains a Stage 4 production milestone.
- One scoped enterprise AI workflow
- Evidence, provenance, and authority planning path
- Verification, contradiction, and disclosure orientation
- Human approval and remediation workflow design
- One governed customer-safe assurance deliverable
Trust, validation, and boundaries
Current capability status is explicit
DigiTrust distinguishes implemented reference behavior, deployed AWS patterns, design-partner activity, and future production targets so reviewers can evaluate the platform without inflated claims.
Trust-control platform
Evidence, provenance, deterministic verification, structured contradiction and disclosure controls, durable approval, remediation, exact-use authorization, and signed audit reconstruction.
Cloud runtime
Automated AWS delivery, governed Bedrock routing, AWS data and security services, operational telemetry, and a pending Marketplace listing.
Limited onboarding
Founder-led workflow discovery, Evidence Packet QuickStart engagements, enterprise architecture reviews, and scoped validation.
Stage 4 controls
Durable Evidence Packets, packet-candidate approval, production identity and signing, cloud-native persistence, customer connectors, and notification workflows.
Patent notice
DigiTrust is patent pending. U.S. Patent Application Publication No. US 2026/0121846 A1. Additional U.S. patent application(s) pending.
This notice does not state that a patent has issued.
Confidential review
Qualified buyers, partners, investors, and reviewers may request architecture, security posture, roadmap, AWS validation, and implementation materials under mutual NDA.
Request NDA accessMake one consequential AI decision provable from action to authorized use
Request a walkthrough, discuss the Evidence Packet QuickStart, or define the right design-partner path for your enterprise AI workflow.