Pricing
Start free on Developer. Every plan includes agent identity, rule-based guardrails and the proof below. Team and Enterprise are set up with us directly.
In every plan, free
- AI System graphAgents, models, tools and data sources recorded in your AI systems, with their relationships.
- Signed AIBOMA versioned bill of materials, signed by default, for each AI system, comparable release to release.
- Trust passportA verifiable trust page you can publish for each AI system.
- Signed audit trailThe decisions Praesidia makes, on a tamper-evident record signed under the default configuration.
Secure your first agents — no credit card required.
- Up to 3 connections
- 1 communicating agent · 100 agent-to-agent tasks / mo
- 1 application
- Rule-based guardrails
- Decisions, tool policies and step-up approvals
- Audit log, webhooks & API keys
- GDPR data-subject erasure requests
The Individual and Advanced limit tiers, set up with us directly.
- Everything in Developer, plus:
- Higher connection, agent and request limits
- ML guardrails; LLM guardrails on Advanced
- Task policies, cost monitoring and budget policies
- Teams, roles and governance dashboard (Advanced)
- Jira and Linear issues from a finding (Individual and up)
- EU AI Act risk classification, impact assessment and AI-literacy register (Advanced)
- SIEM forwarding (Advanced)
Enterprise access, approval and compliance controls for agents at scale.
- Everything in Team, plus:
- High capacity limits; no monthly agent-to-agent task cap
- SSO (SAML/OIDC) and SCIM 2.0
- Multi-step approval workflows
- Custom roles and access reviews
- Framework readiness exports, EU AI Act technical documentation and post-market monitoring
- Dedicated account manager
Developer is self-serve. Enterprise is scoped and priced per engagement, and we set the plan up with you — there is no checkout to click through. Team is set up the same way.
What each plan includes
Core identity, rule-based guardrails, model configuration and audit evidence are included from Developer. Team uses the Individual or Advanced tier; compare them separately below.
| Limit or feature | Developer | Team Individual | Team Advanced | Enterprise |
|---|---|---|---|---|
| Price and limits | ||||
| How to get it | Self-serve, free | Talk to us | Talk to us | Talk to us |
| Connections (active, pending or idle) | 3 | 15 | 50 | Custom |
| Team members | 1 (owner only) | 1 | Up to 9,999 | Custom |
| Team nesting depth | No teams | No teams | 2 levels | Custom |
| Communicating agents | 1 | 5 | 25 | Custom |
| Agent-to-agent requests per month | 100 | 5,000 | 100,000 | No monthly cap |
| Active applications | 1 | 3 | 20 | Custom |
| Cross-organization agent shares | No | No | Up to 10 | Custom |
| Support | Community | Priority email | Dedicated account manager | |
| Core evidence, evaluation and routed actions | ||||
| Rule-based guardrail creation and editing | Yes | Yes | Yes | Yes |
| MCP server management, webhooks and API keys | Yes | Yes | Yes | Yes |
| Bring-your-own-key model configuration | Yes | Yes | Yes | Yes |
| Signed audit bundles and audit packages | Yes | Yes | Yes | Yes |
| AI System inventory, AIBOM and trust passports | Yes | Yes | Yes | Yes |
| Evaluations, datasets and release-gate verdicts | Yes | Yes | Yes | Yes |
| GitHub / GitLab repository discovery | Eligible; where enabled | Eligible; where enabled | Eligible; where enabled | Eligible; where enabled |
| Synthetic red-team campaigns | No | No | Yes | Yes |
| Proof timeline, grades and coverage | No | No | Yes; where enabled | Yes; where enabled |
| Custom OpenAI-compatible gateway routing | No | No | Yes; where enabled | Yes; where enabled |
| Managed protected HTTP execution | No | No | No | Yes; where enabled |
| On-demand executive reports | Yes | Yes | Yes | Yes |
| Scheduled executive report delivery | Eligible; where enabled | Eligible; where enabled | Eligible; where enabled | Eligible; where enabled |
| Guardrails, cost and operations | ||||
| ML-powered guardrails (content moderation) | No | Yes | Yes | Yes |
| LLM-powered guardrails (prompt injection, hallucination) | No | No | Yes | Yes |
| Connection-level guardrail editing | No | Yes | Yes | Yes |
| Cost monitoring and usage tracking | No | Yes | Yes | Yes |
| Budget policies and spending limits | No | Yes | Yes | Yes |
| In-app notifications | No | Yes | Yes | Yes |
| Decisions, tool policies and step-up approvals | Yes | Yes | Yes | Yes |
| Task policies (rate limits and spend caps for groups of agents) | No | Yes | Yes | Yes |
| Teams, access and security | ||||
| Team hierarchy and group management | No | No | Yes | Yes |
| Role-based access control | No | No | Yes | Yes |
| Agent governance dashboard and compliance controls | No | No | Yes | Yes |
| SIEM forwarding (Splunk, Datadog, Microsoft Sentinel) | No | No | Yes | Yes |
| EU AI Act risk classification, fundamental-rights impact assessment and AI-literacy register | No | No | Yes | Yes |
| GDPR data-subject erasure requests | Yes | Yes | Yes | Yes |
| Jira and Linear issues from a finding | No | Yes | Yes | Yes |
| ServiceNow incidents, approvals and CMDB links | No | No | Yes | Yes |
| Custom security policies | No | No | Yes | Yes |
| IP allowlist and blocklist restrictions | No | No | Yes | Yes |
| Custom domain verification (DNS) | No | No | Yes | Yes |
| Enterprise controls | ||||
| Single sign-on (SAML/OIDC) | No | No | No | Yes |
| SCIM 2.0 provisioning | No | No | No | Yes |
| Service accounts for machine-to-machine agent ownership | No | No | No | Yes |
| Custom role definitions | No | No | No | Yes |
| Multi-step approval workflows | No | No | No | Yes |
| Temporary role elevation with expiration | No | No | No | Yes |
| Automated access reviews | No | No | No | Yes |
| Per-team resource quotas | No | No | No | Yes |
| Advanced analytics and reporting | No | No | No | Yes |
| Framework readiness and compliance exports | No | No | No | Yes |
| EU AI Act technical documentation and post-market monitoring | No | No | No | Yes |
A plan entitlement does not establish deployment availability. Features marked “where enabled” need a separate rollout setting; confirm the capabilities enabled for your workspace with us. Single-step approvals are in every plan. Advanced proof records do not include managed protected HTTP execution, which also requires Enterprise approval workflows. Cross-organization sharing is separate from federated execution, which must be enabled independently. Release gates stop a deployment only when your pipeline acts on their verdict.
The table shows default limits per organization. Enterprise has no monthly agent-to-agent request cap; other listed capacities have finite defaults. Team and Enterprise delivery terms are scoped with us. Prefer email? Contact Sales by email at hello@praesidia.ai.
Common questions
Can I change plans later?
Yes. Changes take effect as soon as they are applied. Self-serve payment is not open yet, so beyond Developer we set your plan up with you directly — talk to us about a demo.
What counts as a connection?
A connection is a governed link between two entities — an app, an agent, or an MCP server. Controls apply on supported routed paths; registering a connection does not intercept direct traffic, and content checks differ by path.
What counts toward the monthly request limit?
Each agent-to-agent task counts once; a retry of the same task does not. Guardrail evaluations and audit records never count. The count covers the whole organization and resets each calendar month (UTC).
At the limit, further submissions are declined until the next month. Enterprise has no monthly agent-to-agent request cap.
What is a design partner?
A team running agents in production that works directly with our engineers: we work on your estate, and what you run into shapes what we build next. It is a working relationship, not a plan on this page. How design partnerships work.
Do you support open source?
Praesidia is proprietary, commercially licensed software. The audit verifier and the client SDKs carry permissive open-source licences (Apache-2.0 and MIT). The TypeScript SDK and audit verifier source are available in public Praesidia repositories; contact us for packaged distribution. Paid plans add higher limits and advanced governance and enterprise capabilities.
Start securing your AI agents today.
Create your workspace and connect one agent. No credit card required.