Customer APIs
Connect internal services through authenticated HTTP calls or import a supported OpenAPI specification. Select the operations that become agent tools and assign the credentials needed to use them.
Integrations & APIs
AI agent integrations.
Build with APIs and MCP.
Give agents access to the tools your organization already uses. Extend what they can do, then bring agent work into your own applications through DT’s API, webhooks, and MCP endpoint.
Internal APIs and MCP tools
Custom capabilities and skills
DT APIs for your applications
01 / Connect your tools
Work with the systems you already run.
Choose the connection that fits the system and the responsibility. Access follows the assigned credentials, available operations, and deployment’s network rules.
MCP connections
Connect external Model Context Protocol servers that expose tools and data. Administrators can manage shared connections and assign access to agents. Configured self-hosted deployments can also use local-command servers.
Business applications
Use supported native connections such as Google Sheets, HubSpot, and Apollo, or configured connected-app flows. Available operations depend on the provider, account permissions, and agent assignment.
Custom tools and skills
Package reusable instructions as skills. Where sandbox compute is enabled, add code-backed tools for organization-specific transformations and actions within the configured execution boundaries.
03 / Build with the platform
Use DT from your own applications.
An internal portal, existing service, or another AI client can work with DT’s persistent agents. Your application can initiate work, follow its progress, and use the results.
Persistent Agent REST API
Create and configure agents, send assignments, manage files, and inspect ongoing work. Scoped API keys keep requests within the acting user’s or organization’s permissions.
Events, results, and decisions
Follow timeline events or receive supported events through signed webhooks. Retrieve files and resolve supported pending actions, including human input and approval requests, from an integration you build.
DT’s MCP endpoint
Let an external AI client create, message, and inspect accessible DT agents through the platform’s own MCP endpoint. The client can use DT as a persistent work environment with files, history, and linked responsibilities.
Connecting an MCP server gives DT tools to use. Connecting a client to DT’s MCP endpoint lets that client work with DT agents.
04 / An example integration
From an internal request to a reviewed result.
A program portal could use DT to prepare a recurring status brief. This illustrative flow shows how the platform connects to an internal application and its supporting systems.
Start from the portal
The application sends an assignment through DT’s Agent API. The agent uses its standing instructions, earlier work, and assigned source material to prepare the next update.
Work with approved systems
The agent reads status records through an authorized internal API. If the workflow includes an approval-required OpenAPI write, the proposed operation becomes a specific review request.
Bring the result back
The portal follows progress, presents supported decisions to the reviewer, and retrieves the resulting brief and tracker. The agent retains its working context for the next cycle.
This is an integration pattern, not a prebuilt portal or a customer case study. The application, connections, and review experience are implemented for the intended workflow.
05 / Your implementation path
Build with your team. Bring in ours when you need us.
Your team can develop integrations using platform documentation and the interfaces above. DT can provide additional expertise for the parts you want help with.
Customer-led development
Use the platform’s API, MCP, administration, and deployment documentation to plan connections, establish access, and build a solution your team can operate. Bring your intended workflow to a technical discussion.
Discuss the developer interfacesOptional forward-deployed engineers
Engage DT for initial solution consulting or hands-on implementation. We can work alongside your team on integration, deployment, testing, documentation, and handoff.
Explore engineering supportPlan ownership, documentation, and handoff with our guide to forward-deployed engineering for defense AI. It explains what customer teams should retain after an implementation engagement.
Next steps
What will your team build?
Explore the platform around your systems and workflows. Your team can lead implementation, with DT engineering or independent consulting available when you need it.