Veto
veto.so
- Free plan
- Web
- API
- Self-hosted
What Veto is
Veto checks AI-agent tool calls before they execute and returns an allow, deny, or approval-required verdict. Policies can permit safe actions, route risky actions for approval, and block forbidden actions. Decision records can capture the actor, tenant, tool, policy, and approver. Integrations listed by Veto include OpenAI, Anthropic, Gemini, Amazon Bedrock, Cohere, Groq, Mistral, Together AI, LangChain, LangGraph, MCP, and Playwright, among others. The production SDK path is TypeScript, while Python is a preview package for local checks. Veto says its Apache-2.0 SDK and policy engine run locally, and cached policy can keep critical enforcement operating when Cloud is unreachable. Customer payloads, policies, and decision records are processed to provide the service and are not used for model training. The Developer plan is free with 5,000 monthly checks; Hosted costs $299 per month; Enterprise pricing is custom.
Who it suits
Veto is aimed at startups and vendors selling agents into workflows where customers require policy controls, approvals, and evidence before production. It can fit teams wanting hosted or locally run enforcement.
Good to know, and what to watch
- Allow, deny, and approval decisions
- Auditable actor and approver records
- Apache-2.0 local engine
- Cached offline enforcement
- Broad AI integrations
- Python SDK remains preview
- Developer plan has a hard usage cap
- Hosted plan charges for overages
Plans
Every plan and priceEnterprise
Developer
Hosted
Free plan
The verdict
Veto combines policy checks, approval workflows, and decision records for AI-agent tool calls. Its free Developer tier offers a way to begin, while hosted and enterprise deployments add capacity and governance options.
What the maker publishes
- Data use
- Veto says customer tool-call payloads, policies, and decision records are processed to deliver the service and are not used to train models. source
- Decision records
- Veto can record actor, tenant, tool, policy, and approver for governed decisions. source
- Integrations
- The integrations page lists OpenAI, Anthropic, Gemini, Amazon Bedrock, Cohere, Groq, Mistral, Together AI, LangChain, LangGraph, MCP, and Playwright, among others. source
- Maker
- The site identifies Veto as a product of Plaw, Inc. source
- Offline enforcement
- The FAQ says the critical enforcement path can run locally from cached policy if Veto Cloud is unreachable. source
- Open source
- Veto says its SDK and policy engine are Apache-2.0 and can be run locally. source
- Policy rules
- Policies can allow safe actions, hold risky actions for approval, and block forbidden actions. source
- Purpose
- Veto checks AI agent tool calls before execution and returns an allow, deny, or approval-required verdict. source
- SDKs
- The production SDK path is TypeScript, and Python is described as a preview package for local tool-call checks. source
- Security
- The security page says production endpoints require HTTPS/TLS, database storage and backups use provider-managed encryption keys, and API keys are stored as one-way hashes. source
- SOC 2
- Veto says its security program is mapped to the SOC 2 Trust Services Criteria and that current control evidence is shared with qualified customer security teams during review. source
- Support
- Veto says it responds to security questionnaires, vendor assessments, or compliance questions within 24 hours. source
- Target users
- Veto describes its strongest commercial fit as startups and vendors selling agents into workflows where buyers require policy, approval, and evidence before production. source
Checked against the maker's pages on 7 Oct 2026.