Prediction Guard
Self-hosted AI control plane that applies runtime policy controls to models, agents and MCP tools inside the customer's own infrastructure
Last checked October 28, 2026Quick facts: Prediction Guard is an AI compliance vendor founded in 2023 and headquartered in Lafayette, United States. The vendor publicly documents coverage for NIST AI RMF, EU AI Act, and ISO/IEC 42001. Pricing is tiered. Profile last checked October 28, 2026; sources are cited at the bottom of this page.
About Prediction Guard
Prediction Guard sells a self-hosted "sovereign AI control plane" for regulated organizations. It sits between applications and the models, agents and MCP tools they call, and applies access, safety and policy controls to each request inside the customer's own infrastructure. The company was founded in 2023 in Lafayette, Indiana; CEO Daniel Whitenack is a co-founder. It raised a $3.7 million seed round led by Sovereign Capital in May 2025, and Noblis Ventures made a strategic investment of undisclosed size in July 2025. At the time of the seed round it had eight employees and said it had 18 customers in health care, defense, transportation and manufacturing. The platform registers each agent with an identity and role and limits it to the models and tools that role needs. Runtime controls include PII masking, prompt injection detection, toxicity filtering, input sanitization, MCP tool scoping and human-in-the-loop approval; requests that fail a policy are blocked and logged. It produces an exportable AI bill of materials (CycloneDX) of models, guardrails and MCP servers, runs model security scans and streams security events to SIEM tools such as Splunk, Datadog and Grafana. Applications call it through OpenAI- and Anthropic-compatible APIs. It deploys into a customer VPC on AWS, Azure or Google Cloud, on Kubernetes on premises, or air-gapped. Prediction Guard says its control overhead is under 200 ms per request. **AI compliance coverage** Prediction Guard says its policies align with the NIST AI RMF, NIST AI 600-1, the OWASP Top 10 for LLM and agentic applications, OMB M-26-04 and AIUC-1, with one-click baselines and an exportable "proof of alignment" for the NIST AI RMF. A company blog post (August 2026) adds the EU AI Act and ISO/IEC 42001. The same post labels its OWASP mapping "self-published" and not independently verified, and notes that the audit log records enforcement events rather than full replayable traces. We found no published control-by-control mapping. The homepage shows badges for a SOC 2 Type 2 report by Sensiba and HIPAA compliance; there is no public trust center. **Who it's for** Regulated enterprises and public sector bodies that want AI guardrails and agent controls to run inside their own network, including disconnected environments. It is sold to government through Carahsoft. Its homepage quotes executives at Noblis (also an investor), SimWerx and iBase-t.
Frameworks supported
Regulations, voluntary standards and compliance programs Prediction Guard documents support for on their own materials. Chip shading reflects the strength of the claim, not an independent audit.
Voluntary standard · US · voluntary
Regulation · EU · Phased — in force; high-risk rules deferred
Voluntary standard · Global · voluntary
Prediction Guard features
Capabilities Prediction Guard markets publicly. Inclusion means the feature is documented on the vendor's site — not that it's best-in-class. Last checked October 28, 2026.
Runtime Enforcement
Inline policy enforcement on AI inputs and outputs at request time.
Policy Enforcement
Mechanism for applying organizational and regulatory policies to AI usage at runtime.
LLM Guardrails
Runtime input/output filtering for LLMs — PII redaction, toxicity blocking, prompt injection defense, policy enforcement.
Prompt Injection Defense
Detection and mitigation of malicious prompts intended to bypass system instructions or exfiltrate data.
Data Leakage Prevention
Inspection of LLM outputs to block PII, secrets, and regulated data from leaving the boundary.
LLM Guardrails & Content Filtering
Runtime guardrails that block or redact unsafe prompts and responses in production LLM applications.
Audit Logging
Tamper-evident logging of governance events (approvals, model changes, policy decisions) required by EU AI Act Article 12 and similar regulations.
AI Bill of Materials
Structured inventory of components, datasets, and models that make up an AI system.
AI Supply Chain Risk
Assessment of risks from third-party models, datasets, and AI service providers.
Human-in-the-Loop
Workflows that require human review or approval for AI decisions.
Integrations
Documented by Prediction Guard in public product materials.
- Splunk
- Datadog
- Kubernetes
- LangChain
- AWS Marketplace
- Microsoft Azure Marketplace
Prediction Guard pricing
No prices on predictionguard.com. Its AWS Marketplace listing gives a 12-month bundled license at $165,000 covering two AI systems, 25 requests per second and unlimited monthly requests, with up to 10% off for 24 months. The Azure Marketplace listing describes a fixed-price offering that is never seat-based, with private offers on request.
Pros and cons of Prediction Guard
Pros
- Runs entirely in the customer's environment (cloud VPC, on-premises Kubernetes or air-gapped), so prompts and logs need not leave the network.
- Covers agents as well as models: per-agent identity and roles, MCP tool scoping, sandboxed code execution and human-in-the-loop approval.
- Public list price on AWS Marketplace ($165,000 a year for two AI systems), and the license is not priced per seat.
- Exports a CycloneDX AI bill of materials and streams enforcement events to Splunk, Datadog or Grafana for audit and security teams.
Cons
- Framework mappings (NIST AI RMF, EU AI Act, ISO 42001, AIUC-1) are vendor-stated, and no control-by-control mapping is published.
- High entry price for smaller teams: the only published price is $165,000 a year.
- SOC 2 Type 2 and HIPAA appear only as homepage badges; there is no trust center or published report summary.
- Small, young company: founded in 2023, eight employees at its May 2025 seed round, and customer evidence limited to three homepage testimonials.
Alternatives to Prediction Guard
F5 AI Guardrails
Compare Prediction Guard vs F5 AI Guardrails
Lakera
Compare Prediction Guard vs Lakera
Prisma AIRS
Compare Prediction Guard vs Prisma AIRS
Cisco AI Defense
Compare Prediction Guard vs Cisco AI Defense
WitnessAI
Compare Prediction Guard vs WitnessAI
Noma Security
Compare Prediction Guard vs Noma Security
Frequently asked
What is Prediction Guard?+
Prediction Guard is a self-hosted AI control plane from Lafayette, Indiana, founded in 2023. It sits between applications and the models, agents and MCP tools they use, and applies access controls, PII masking, prompt injection detection, toxicity filtering and logging inside the customer's own infrastructure.
How much does Prediction Guard cost?+
Prediction Guard doesn't publish prices on its website. Its AWS Marketplace listing offers a 12-month license at $165,000 for two AI systems, 25 requests per second and unlimited monthly requests. Its Azure Marketplace listing describes fixed, non-seat pricing with private offers on request.
Can Prediction Guard run air-gapped or on premises?+
Yes. Prediction Guard says it deploys into a customer VPC on AWS, Azure or Google Cloud, on an on-premises Kubernetes cluster, as a standalone binary or in a fully air-gapped environment with offline packages.
Does Prediction Guard support the NIST AI RMF or the EU AI Act?+
Partly. Prediction Guard says its runtime policies align with the NIST AI RMF, NIST AI 600-1, the OWASP LLM and agentic Top 10 lists and AIUC-1, and that it can export proof of alignment to the NIST AI RMF. A company blog post also names the EU AI Act and ISO/IEC 42001, without a published control-by-control mapping. It provides runtime controls and logs, not risk assessments or regulatory documentation.
What are the alternatives to Prediction Guard?+
The closest alternatives in this directory are F5 AI Guardrails, Lakera, Palo Alto Networks Prisma AIRS, Cisco AI Defense, WitnessAI and Noma Security. Each applies runtime guardrails or policy controls to AI applications and agents.
Sources
- Prediction Guard: Home (testimonials from Noblis, SimWerx and iBase-t; SOC 2 Type 2 and HIPAA badges)
- Prediction Guard: Product overview (supply chain, runtime controls, AI operations, agent orchestration)
- Prediction Guard: Govern (NIST AI RMF, NIST 600-1, OWASP LLM Top 10, OMB M-26-04, AIUC-1 baselines; AIBOM; SIEM)
- Prediction Guard: Deploy (cloud VPC, on-premises Kubernetes, air-gapped, hybrid)
- Prediction Guard blog: Runtime AI governance platforms 2026 (framework claims)
- Prediction Guard: Launches sovereign AI control plane on Microsoft Azure Marketplace (March 2026)
- AWS Marketplace: Prediction Guard - A Sovereign AI Control Plane (12-month price)
- Microsoft Marketplace: Prediction Guard (fixed-price offering, NIST AI RMF and OWASP alignment claims)
- LangChain docs: ChatPredictionGuard integration (langchain-predictionguard package)
- Carahsoft: Prediction Guard for Government
- Inside Indiana Business: Prediction Guard secures $3.7M seed funding (May 2025; founded 2023, eight employees, 18 customers)
- GlobeNewswire: Noblis invests in Prediction Guard (July 22, 2025)
- M25: Why we invested in Prediction Guard (Lafayette, Indiana; CEO and co-founder Daniel Whitenack)
Keep reading
See an error or outdated detail?
Profiles carry a last-checked date. If something is out of date or wrong, send a correction and we will review it.
Work at Prediction Guard?
Claim this profile for free to propose corrections; every change is still checked by our editors. Paid options add your own material or a labeled spot higher in directory lists. They never change the facts, coverage or review on this page.