Enkrypt AI Guardrails

Enkrypt AI Guardrails is a runtime layer that can approve, change or block risky behavior across AI agents, tools, retrieval-augmented generation and MCP. It applies controls at prompt, retrieval, tool and output boundaries, including filtering, rewriting, blocking, escalation and approval or denial of tool calls. The risks it addresses include prompt injection, unsafe tool use, boundary violations, sensitive-data exposure, jailbreaks, toxic content and compliance concerns. Policies also cover text, image and audio inputs. The product supports API-first use with existing model stacks, agent hooks or middleware, MCP Gateway, SIEM and ticketing routes, and identity claims such as Okta, Azure AD and custom JWT claims. Logs include policy ID, version and reason code and can be exported to SIEM. The product page claims latency under 15 milliseconds. Explore is free, with 50 monthly credits and seven-day data retention; Launch costs $149 monthly, while paid pricing is listed from $134/mo billed annually. Enterprise offers VPC or on-premises deployment, custom retention and custom SLAs.

Enkrypt AI Guardrails combines enforcement controls, multiple input modalities and auditable decisions. The free plan has limited monthly credits and retention, so review plan limits against the intended evaluation or deployment.

It is intended for enterprises securing AI deployments, AI safety researchers and people building safer AI systems. The listed controls and audit logs may be relevant to teams managing agent and model risks.

Enkrypt AI Guardrails is a runtime security layer for teams operating AI agents and connected tools. It is best suited to organizations that need enforcement across an existing AI workflow, rather than a filter for model output alone. Its broad controls and auditable decisions are a strong combination, but credit allowances and retention make plan choice consequential.

Overview

The product can approve, modify, or block risky behavior at prompt, retrieval, tool, and output boundaries. That matters in agent workflows, where exposure can come from a tool action or retrieved material as well as a generated response. Its coverage includes prompt injection, jailbreaks, unsafe tool actions, privilege and tenant boundary violations, sensitive data exfiltration, toxic content, and compliance risks.

Policies also cover image and audio inputs, including injection defenses. This makes Enkrypt more relevant to multimodal deployments than a text-only control layer, though teams should choose it for the enforcement workflow they need rather than modality breadth alone.

Key features

Controls within an existing stack

API-first use, agent hooks or middleware, and an MCP Gateway offer ways to place controls around existing models and tools. Identity context can include Okta, Azure AD, and custom JWT claims, which is useful when policy decisions depend on who is making a request. Python is the listed SDK language; teams building in other languages may need to rely on the API or their own integration work.

Auditable enforcement

Logs include the policy ID, version, and reason code, and can be exported to SIEM. Those records make it easier to support internal control evidence and review why a decision occurred. The product page claims latency under 15 milliseconds and stable behavior under load; those claims are relevant to teams weighing runtime controls against responsiveness.

Security and support

The pricing page lists SOC 2 Type II, ISO 27001, GDPR Ready, HIPAA Ready, and NIST AI RMF Aligned. Support ranges from community and email to dedicated Slack or Teams, with 24x7 support availability for Enterprise. The broader certifications and support options help make Enkrypt worth considering for organizational deployments, but the support tier depends on the plan.

Pricing

Enkrypt uses a freemium model. Explore is free forever at no cost, with 500 credits to start, 50 credits per month, seven-day data retention, and community support. It is suited to spot checks and evaluation, not a comprehensive red team assessment, which requires 5,000 credits.

Launch costs $149 a month, billed monthly, and includes 5,000 starting credits, 250 credits per month, 30-day retention, and email support. It is the lower-cost paid option, but its recurring allowance is modest relative to its starting credits. Scale costs $1,499 a month, billed monthly, with 10,000 starting credits, 1,000 per month, 30-day retention, and dedicated Slack or Teams support. Its larger allowance and direct support suit heavier evaluation or use, at a substantial monthly premium over Launch.

Enterprise has custom pricing and includes custom allocation, custom retention, custom SLAs, VPC or on-premises deployment, and unlimited usage for VPC. It is the fit for organizations needing deployment control or tailored terms. The published pricing note says paid plans start at $134/mo (annual), while Launch and Scale are shown with monthly billing; buyers should distinguish that annual starting figure from the monthly plan prices.

Lower tiers trade shorter retention and less support for lower cost, while Explore's 50-credit monthly cap limits sustained evaluation. Confirm that a plan's recurring quota, not just its starting credits, matches the intended workload.

Platforms

Enkrypt is available through API, self-hosted deployment, and web. Enterprise adds VPC or on-premises deployment and unlimited VPC usage, making it the clearest option for teams that require that deployment model.

Who it's for

Enkrypt is aimed at enterprises securing AI deployments, AI safety researchers, and people building safer AI. It is strongest for teams with agents, tool calls, retrieval, or multimodal inputs to govern and a need to inspect enforcement decisions. A small team seeking a broad, sustained assessment on a free tier may find Explore's 50 monthly credits too restrictive.

Pros and cons

• Pros: Controls span prompts, retrieval, tools, and outputs, addressing risks that a final-response filter can miss.
• Pros: Text, image, and audio policies extend coverage to multimodal inputs.
• Pros: Policy-versioned logs with reason codes and SIEM export support audit review.
• Cons: Explore's 50 monthly credits and seven-day retention constrain recurring evaluation and review.
• Cons: A comprehensive red team assessment requires 5,000 credits, beyond Explore's monthly allowance.
• Cons: Scale costs $1,499 a month, and Enterprise requires custom pricing for VPC or on-premises deployment.

Alternatives

Browse AI Guardrail Software for a broader category view. Consider OpenSecureAI Prompt Firewall instead if its free plan's 2,000 API requests per month and in-browser tools or npm packages fit the workload. Fiddler Guardrails is another freemium option, with its free plan describing real-time protection against harmful exposure, hallucinations, toxicity, PII or PHI, prompt injection, and jailbreak attempts.

Openlayer Guardrails may suit a reader whose needs align with its free Basic plan's one member, five projects, one inference pipeline per project, and 20,000 monthly inferences. PromptGuard offers a free plan with 20,000 scans a month, one API key, one project, and 24-hour log retention, which may fit a scan-volume-led evaluation. PII Firewall is a free, self-hosted Linux option distributed as an Apache 2.0 Python package with a regex-only install that needs no ML dependencies.

Amazon Bedrock Guardrails is a paid web option with separately priced text content filters and denied topics at $0.15 a month for Standard and Classic tiers, and image content filters at no cost. HiddenLayer AI Runtime Security is another paid option, available through API, Linux, self-hosted, and web. Oracle Mobile Authenticator is a free option for Android, iOS, and Windows.

Verdict

Choose Enkrypt AI Guardrails if your team needs runtime controls across agents, tools, retrieval, and multimodal inputs, with records that support audit review. Its broad enforcement points are the main reason to choose it; the recurring credit caps and limited Explore retention are the main reasons to compare alternatives or budget for a higher tier.