Agentic Trust : IAM for Agents and MCP's

The Identity & Access Control (IAM) backbone for AI Agents {Authenticate -> Authorize -> Audit every AI agent & MCP tool} Helping devs & enterprises secure Agentic Workflows

Comment

AgenticTrust — Enterprise IAM for AI Agents and MCP

Inspiration

The recent incident involving a rogue REPLIT agent deleting a production database really resonated with me, especially given my background in identity security

It exposed three core risks for enterprises:

  • No Verifiable Agent Identity: Unable to track or control which agent acted, risking impersonation and hidden threats.
  • Weak Permissions: Agents often have more access than necessary, increasing risk of data leaks or unintended destructive actions.
  • Incomplete Audit Trails: Lack of thorough logs makes incident investigation, recovery, and compliance impossible.

I wanted to build a solution that closes these gaps specifically for AI agents and MCP workflows.

What it does

AgenticTrust provides tailored Identity and Access Management (IAM) for AI agents and MCPs to help enterprises and developers securely scale and govern agentic workflows by offering:

  • Verifiable agent identities with cryptographic security.
  • Dynamic, least-privilege, task- and tool-scoped permissions.
  • Centralized real-time MCP registry for lifecycle and permission management.
  • Tamper-proof audit trails across any orchestration platform and cloud environment for accountability.

How we built it

  • Validated hypothesis & idea with industry experts (Nathan Dehmuth, Coalfire) and AI-native vendors (Workspan.com, Shopos.com).
  • Developed a secure agent registry and dynamic permission engine in under 6 hours using AWS Bedrock and integrated with partner tools (Claude , Dynatrace, Smithery)
  • Prioritized developer-friendly onboarding so teams focus on agent logic, not IAM complexities.

Challenges we ran into

  • Honestly, one of our first big hurdles was figuring out if what we were building was actually needed or just a cool idea. Talking to Nathan Dehmuth (VP of Cloud Service at Coalfire and a hackathon judge) made a huge difference—he totally got the problem, validated our approach, and encouraged us to take it seriously even after the hackathon. That gave us a ton of confidence to keep going.

  • Managing scope and feature prioritization under the hackathon’s tight timeline to deliver a clear, impactful MVP.

Accomplishments that we're proud of

  • Within a short time, we validated our idea and secured soft commitments from two AI-native enterprise design partners—Workspan.com (Series D, ~$600M valuation) and Shopos.com ($20M raised, building an agentic e-commerce framework)—who confirmed this is a real challenge they face and are eager to pilot our solution.

  • Delivered a working prototype, positioning AgenticTrust as the first IAM platform purpose-built for AI agents and MCP.

What we learned

  • Enterprises won’t scale AI agent deployment without strong, tailor-made IAM; Gartner says ~40% delay rollouts due to missing controls.
  • Developers need simple, plug-and-play IAM solutions to move fast without becoming identity experts.
  • Security must be built in from day one, not added later.

What's next for Agentic Trust

  • Collaborate with design partners to refine product-market fit through real-world deployments.
  • Launch beta and convert partners into paying customers within one month.
  • Simplify integrations so developers focus on agents while we handle identity/access, essential for security reviews and fast adoption.
  • Raise funding to scale and expand integrations further.

What we'd have liked to do with more time (and plan to do post-hackathon)

  • Integrate real-time threat intelligence from Wiz MCP servers.
  • Integration with AgentCore for seamless identity management.

Built With

Share this project:

Updates