Process Street adds 105 API endpoints to let AI agents run workflows headlessly

Process Street launched 105 new public API endpoints and mirrored them as tools on its hosted MCP server, allowing AI agents and software to create, update and run workflows without using the interface. The rollout is available now to all customers and broadens the platform’s automation while keeping existing permission controls in place.

Why it matters: - Process Street is moving its compliance operations platform into an agent-driven model, where software and AI agents can build and run workflows directly. - The change expands automation beyond workflow execution to workflow creation, permissions and data handling. - The rollout could reduce manual clicks for teams managing SOPs, audit trails and compliance processes.

What happened: - Process Street released 105 new public API endpoints and exposed each one as a tool on its hosted Model Context Protocol server at mcp.process.st. - The platform can now be driven entirely by software and AI agents from any surface, without users clicking through the interface. - The endpoints and matching MCP tools are live now for all Process Street customers on all plans.

The details: - The new API coverage spans workflow authoring end to end, including creating, updating, duplicating and publishing workflows. - The endpoints also cover tasks, logic rules, assignment rules, due date rules and widgets. - Pages support revisions, publishing and markdown export. - Data Sets support incoming webhooks. - A paginated workflow run listing removes the old 200-result cap. - The API also covers one-off tasks, My Work, scheduled workflows, outgoing webhooks, attachments, folders and permissions. - Permissions can be managed at the folder, workflow, page and individual task level. - The MCP server mirrors the public API, so each of the 105 endpoints also appears as an agent tool. - An assistant connected to Process Street can build a workflow, create a compliance review, assign it, publish a revised procedure, load a vendor list into a data set, set folder permissions and schedule repeat runs. - The server works with Claude, ChatGPT on Pro, Team, Business, Enterprise and Edu plans, Microsoft Copilot Studio and any application that supports the Model Context Protocol. - One-click connectors are available for Claude, ChatGPT and Copilot Studio. - Other clients connect to the remote server with a Process Street API key using bearer token authentication. - Access follows existing Process Street permissions. - Admins, Builders and Users can use the API and MCP server, and each role can only interact with resources allowed by its permissions. - The company introduced no new permission model and moved no data out of the platform to make the tools available. - Organizations with SAML SSO enabled are not supported at this time. - Process Street maintains ISO 27001 and SOC 2 Type II certification and is HIPAA and GDPR compliant. - API documentation is available at developer documentation. - MCP server documentation, including connection instructions for each supported client, is available at MCP server documentation. - Process Street is headquartered in San Francisco and says more than 3,000 companies use the platform, including Salesforce, Slack, the Government of Canada, Colliers and Betterment.

Between the lines: - The launch signals a shift from a UI-first product to a platform designed for autonomous agents and external tools. - By exposing workflow creation, permissions and scheduled execution through both API and MCP, Process Street is trying to make compliance operations callable from wherever teams already work. - Keeping the existing permission model intact is likely a key selling point for regulated customers.

What's next: - Customers can start using the new endpoints and MCP tools immediately. - Teams using supported agents and clients can connect through the hosted MCP server or via API key authentication. - Organizations using SAML SSO will need to wait for future support before adopting the feature set.