The deterministic governance layer to move agentic workflows from pilot to production. Egress DLP, LLM cost budgets, and omnichannel HITL approvals — directly in Teams, Slack, and Telegram.
Four pillars that make agentic AI production-ready
7-category PII scanner with Luhn checksum validation. Redact or block sensitive data before it reaches the LLM or leaves through tool outputs.
Per-org and per-agent daily/monthly spend limits with hard enforcement. Multi-channel alerts via Teams, Slack, and Telegram when thresholds are hit.
Three-tier sandbox that expands agent capabilities as trust is earned. New agents start restricted; proven agents unlock network access and approvals.
Multi-signature human approvals with role-based policies. Approve or deny sensitive actions directly from Teams, Slack, or Telegram with one tap.
When AI agents can send emails, browse the web, and spend money on your behalf, identity alone isn't enough. You need trust.
Agents act on your behalf without earning trust first. No behavioral track record, no accountability.
You hand tasks to agents with no reputation history. Reliable or malicious? No way to tell.
Traditional security kills agents at the first sign of risk. No graduated response, no safe fallback.
When an agent goes wrong, there's no audit trail, no governance tooling, and no way to investigate.
An autonomous personal assistant that earns trust through behavior, not promises. Every action flows through the DAT Framework's trust kernel.
Read, search, and send emails via Microsoft or Gmail. Create calendar events. All through MCP tools, all trust-gated.
Search the web, scrape pages, fill forms, take screenshots. Headless Chromium with SSRF protection and human-in-the-loop approval.
Discover other agents and delegate sub-tasks via A2A protocol. Trust attenuates through the chain — delegatees must earn it.
Set up autonomous watchers: health checks, web scrape diffs, API polling, inbox monitoring. Push alerts via Telegram, Slack, and Teams.
20 built-in deterministic pipelines: apartment search, flight booking, inbox triage, price tracking, trip planning, morning briefs, and more.
Agent creates its own tools at runtime via JSON pipeline definitions. QuickJS WASM sandbox, 6-layer security, HITL approval with full code visibility.
Every tool call flows through this pipeline. Trust drops mid-task? Available tools shrink instantly.
From identity to authorization to behavioral ML — the DAT Framework provides every layer an autonomous agent needs.
Each agent gets a unique, cryptographically verifiable identity using the did:dat method. Built on W3C DID Core standard with full method specification.
Dynamic trust scores with multi-layered anti-gaming protections including rate limiting, behavioral analysis, and fraud detection.
Fine-grained control over what agents can do, with spending limits and time constraints.
Immutable, tamper-evident logs of all agent actions for compliance and debugging.
AI-powered detection of suspicious agent behavior using ensemble learning.
ML-powered behavioral analysis to identify whether activity originates from humans or AI agents.
Traditional security products block everything when risk rises. DAT uses Dynamic Scoping to keep your agents running safely with reduced capabilities.
All tools available. Email, browser interactions, skill building, delegation — all enabled with HITL approval on high-risk actions.
Read-only web, search, shell, memory. High-risk actions blocked (email send, browser clicks, skill building).
Limited tool access. Reduced iteration budget. No network, no shell, no delegation.
Our ML-powered classifier analyzes behavioral patterns to determine whether activity originates from a human or an AI agent with high confidence.
Get your agents verified and trusted in minutes, not months.
Create a unique DID for your agent with cryptographic keys. Define capabilities, compliance status, and metadata.
POST /api/v1/agents
Submit attestations and complete verification levels from basic to maximum security for enterprise use.
POST /api/v1/verify/:did
Your agent's reputation grows with successful interactions. Trust signals are recorded and scores update in real-time.
GET /api/v1/reputation/:did
Users grant scoped permissions to your agent. Actions are validated, logged, and auditable.
POST /api/v1/authorize
A public REST API that lets any platform check an agent's trust score in real-time. Like a credit bureau for AI agents.
Query one agent or up to 100 at once. Get trust score, 5-pillar breakdown, badges, and investigation state.
SHA-256 hashed keys with per-key rate limiting and monthly quotas. Four tiers from free to enterprise.
Boolean trust check with configurable threshold. Instantly decide if an agent is trusted enough for your use case.
# Query an agent's trust score
curl https://www.datops.ai/api/v1/public/trust/did:dat:mainnet:agent_abc123 \
-H "X-TaaS-API-Key: taas_your_key_here"
# Response
{
"agentDid": "did:dat:mainnet:agent_abc123",
"trustScore": 78.5,
"pillars": {
"reliability": 82.0,
"performance": 75.0,
"compliance": 70.0,
"security": 80.0
},
"totalInteractions": 1250
}
# Quick trust check
curl ".../verify?threshold=70" \
-H "X-TaaS-API-Key: taas_your_key_here"
# → { "trusted": true, "trustScore": 78.5 }A microservices architecture designed for high availability and horizontal scaling.
From AI assistants to autonomous trading bots, DAT provides the trust layer.
Your DAT Agent reads email, books flights, monitors the web, and delegates tasks — all governed by earned trust. Try it via Telegram, Slack, Teams, or the dashboard.
Manage spending limits, transaction authorization, and compliance for agents handling financial operations.
Enable secure agent-to-agent communication with verified identities and trust scores.
Deploy internal agents with proper governance, audit trails, and compliance reporting.
Let customers trust agents that shop, negotiate, and transact on their behalf.
Ensure HIPAA compliance and proper authorization for AI agents accessing medical data.
Experience how DAT protects real-world systems with trust-based access control.
Watch AI agent sandbox permissions change dynamically based on real-time trust scores. Create agents, trigger events, observe trust adaptation.
Launch SimulatorChat with a real DAT-secured AI agent running on this page. Watch its trust score change in real-time with every interaction.
Try It Now — Click the bot iconSee how DAT protects patient health records (PHI) by verifying AI agent trust scores before granting access.
Launch DemoWatch DAT enforce risk-based trust requirements for AI trading agents with tiered access control.
Launch DemoExplore enterprise governance: freeze agents, shadow score during investigation, exonerate with patience bonus, manage escrow bonds.
Launch SimulatorWatch the DAT Agent reason, act, and use MCP tools in real-time. See trust scores gate every tool call through the ReAct loop.
Sign In to TryDAT is designed from the ground up to meet the most stringent compliance requirements.
Ready for August 2026 enforcement. Full conformity assessments and documentation.
ReadyEnterprise-grade security controls with continuous monitoring and audit trails.
ReadyHealthcare-ready with PHI protection, access controls, and audit logging.
ReadyPrivacy-first design with data minimization and right to erasure support.
ReadyColorado AI Act and Texas RAIGA compliant with impact assessments.
ReadyImmutable, hash-chain verified logs for legal defensibility.
ReadyLet your AI agents operate with real authority. DAT provides the trust layer that makes it safe for a CEO to let an agent spend $50,000.
Join the growing ecosystem of developers building the future of autonomous AI with verified identities and earned trust.
npm install @datops/dat-sdk