How to Try or Demo Prediction Guard

Updated

You can evaluate Prediction Guard through two concrete paths: book a demo from the website for a guided walkthrough, or start hands-on with the Agent Control Kit published on Docker Hub. The Docker route is the fastest way to see the control plane in action if you already run containerized workloads, while the demo route fits teams that need to discuss sovereign, self-hosted deployment inside their own security boundary.

Book a Demo

The site's primary call to action is Book a Demo, paired with a See How It Works link. Use the demo when you want to:

  • See how runtime controls are enforced on agent actions inside your security boundary
  • Discuss self-hosted or sovereign deployment in constrained environments
  • Map the platform against NIST and OWASP-aligned governance requirements
  • Get a TCO comparison against stitching together fragmented point solutions

Come prepared with your operating environment, the agent frameworks you already use, and any compliance frameworks you must satisfy. Those three inputs shape which parts of the platform the demo focuses on.

Try the Agent Control Kit on Docker Hub

Prediction Guard was invited into the Docker Verified Publisher (DVP) agentic program and released a verified Agent Control Kit on Docker Hub. This lets you run AI agents at scale inside Docker-verified Prediction Guard sandboxes that operate on your sovereign AI control plane.

The practical flow:

  1. Pull the verified Agent Control Kit from Docker Hub.
  2. Run agents inside the Prediction Guard sandbox rather than directly on your host.
  3. Observe how runtime controls constrain agent behavior without relying on users to make the "right" decisions.

The expected result is contained agent behavior with a limited blast radius — problematic actions are constrained by the control plane, not by convention.

Explore the Platform Components

Two platform areas are worth exploring directly from the site:

Agent Forge — deploy agents on day one. Both interactive and autonomous agents can be created via API or simple natural language prompts. Key systems integrate through MCP, and human plus agent identities are tracked out of the box.

API / SDK — an OpenAI-, Anthropic-, and MCP-compatible API sits on top of the control plane, whether you use self-hosted or cloud endpoint models. Developers keep working with LangGraph, Pydantic AI, Crew AI, Claude Code, Hermes Agent, or OpenAI SDKs while runtime controls are enforced underneath.

If you want to test compatibility before a demo, this is the layer to check against your existing stack.

Contact the Team for Sovereign Deployment

For deployments in challenging or restricted environments, the site points to sovereign, self-hosted options. Reach out through the website's contact or demo channel and specify:

  • Your target environment and network constraints
  • Whether you need fully self-hosted operation
  • Which governance frameworks you must align with

Which Path Fits You

Your situation Start here
Want a guided overview and ROI discussion Book a Demo
Already run containers and want hands-on testing Agent Control Kit on Docker Hub
Need to verify framework compatibility first API / SDK compatibility, then demo
Deploying in a restricted or sovereign environment Contact the team directly

The site does not publish pricing or self-serve signup details, so both the demo and the Docker route are the documented ways to get started.