Agent spend is set at runtime by the agent's own decisions. See how per-task records in Diagrid Catalyst let a finance team model and bound the cost of running agents.

Find, inspect, and safely intervene on workflows that are running, failed, or waiting on an external event, using workflow operations in Diagrid Catalyst.

When an agent crashes at step forty-seven, it should resume there, not start over. Why durable execution belongs in the runtime, where framework checkpointers fall short, and how Catalyst on Dapr recovers crashed agent workflows automatically.

Why durable execution isn't enough for agent systems, and how Diagrid Catalyst, built on Dapr, brings verifiable execution lineage to agent workflows.

Catalyst now shows where workflows fail and reruns them in bulk from the failed step, finds workflows waiting on human input so you can unblock them, and runs in Catalyst Cloud or inside your own Kubernetes cluster.

Dapr 1.18 brings workflow history propagation, tamper detection, access policies, global concurrency limits, and secure MCP, so production workflows and agents stay trustworthy, controlled, and auditable. All these features are available in Catalyst today.

Latest release enables organizations to prove the authenticity, integrity, and lineage of workflow and AI agent execution. New capabilities digitally sign execution history, propagate trusted provenance across services, and generate attestations that allow auditors to verify exactly how work was performed.

Most agent projects stall before production, and the model is rarely the cause. The four failure modes behind it: durability, security, cost, and observability.

A recap of the May 20, 2026 Diagrid webinar on why agents need cryptographically attestable identity, why MCP gateways are not enough, and how SPIFFE, Dapr, and Catalyst bring zero trust to agentic systems.

AI agents lack verifiable identity, enforceable authorization, and tamper-proof audit trails. See how Diagrid Catalyst delivers cryptographic identity, zero-trust policies, and a signed chain of custody for production AI agents.

Once your agents are in production, the hard part begins. See how Catalyst makes it possible to trace failures, surface bottlenecks, and operate multi-agent systems reliably — using a multi-agent orchestration quickstart as a running example.

Learn how to safely evolve .NET Dapr Workflows in Aspire using version patches for small additive changes and named workflow versions for larger structural changes, without breaking in-flight instances.

Catalyst now offers durable execution for 10 agent frameworks, an agent runtime registry, a live application graph, and workflow rerun, versioning, and search controls for production AI agents.

MCP gateways solve routing. They don't solve agent identity, authorization, or proof. Here's what enterprise AI agents actually need for the zero trust security that is required to trust agents with your data.

The 2026 State of Dapr report is now available. See how the Dapr community is using workflows, AI agents, and MCP servers, and what it takes to move from prototype to production.

Available in 9 languages

Dapr Agents 1.0 brings durable execution, persistent memory, and multi-agent orchestration to build production-ready AI agents you can trust.

Diagrid introduces durable workflow support for leading AI agent frameworks, allowing agents to automatically recover from failures and complete long-running workflows in production.

Aspire is an easy way to build distributed applications locally for both .NET and Python. We're excited to introduce the Catalyst Aspire integration, which makes it dramatically easier to develop, test, and run Dapr applications using Aspire.