Guild provides a vendor‑neutral control plane that abstracts enterprise AI agents from specific LLM models and enforces SSO/SAML/OIDC identity with fine‑grained role‑based access controls. It records every agent execution in an immutable, exportable audit log and manages agents as versioned, permission‑scoped artifacts with centralized cost and usage tracking. The platform also includes an Agent Hub for discovering, sharing, and reusing trusted agents across teams.
Funding
Funding not disclosed
Founders
Product
Problem
Enterprise teams deploying AI agents face fragmented tooling: each agent is tied to a specific model or vendor, lacks centralized identity and permission management, and provides no immutable audit trail. This “shadow AI” creates security, compliance, and operational risks when agents act on production systems without governance.
Solution
Guild delivers a vendor‑neutral control plane that abstracts AI agents from underlying models and frameworks. It enforces enterprise‑grade identity (SSO/SAML/OIDC) and role‑based access controls, applying least‑privilege policies at the moment of tool invocation. Every agent execution is captured in an immutable, exportable log for observability and compliance. Agents are packaged as versioned, permission‑scoped artifacts that can be promoted, rolled back, or shared across teams via the Agent Hub. The platform also provides centralized cost and usage tracking, enabling organizations to monitor spend across workspaces, models, and triggers while maintaining full auditability.
Target Audience
The primary customers are enterprise engineering and AI‑Ops teams that build, deploy, and manage production AI agents across multiple models and vendors, as well as platform engineering groups needing centralized governance and observability for autonomous software.
Features
- Model‑agnostic runtime that supports OpenAI, Anthropic, Google, and open‑source LLMs without locking in a vendor.
- Integrated enterprise identity provider (SAML/OIDC) with SCIM provisioning and fine‑grained RBAC for agents and service accounts.
- Least‑privilege policy engine that mediates all external tool calls and isolates credentials from agent code.
- Immutable, tamper‑proof execution logs with full traceability of prompts, tool invocations, and outcomes, exportable for audit and compliance.
- Versioned agent artifacts with lifecycle management (publish, promote, rollback) and dependency tracking.
- Agent Hub marketplace for discovering, installing, and reusing trusted agents across organizations, supporting semantic search and access controls.
- Centralized cost attribution and usage dashboards per workspace, model, and trigger, with programmable alerts for budget enforcement.
- API‑first design with support for webhook triggers, A2A protocol integration, and CI/CD pipelines.