Vellum kicked off 2026 strong with some core upgrades to the platforms usability to make agent building smoother.

Here are a few of my favorite highlights from January:

• Chat Message Triggers that easily give any of your agents a chat experience for building agents, with automatic conversation history management
• Mocking Nodes & Integrations lets you test workflows instantly without waiting on external APIs
• Workflow Log Events bring granular observability directly into your workflow execution

Chat with any agent

Adding a chat function to agents used to be a difficult process manually storing chat history threads, but with the new Chat Message Trigger, Vellum automatically creates and manages a chat_history State variable for you.

The Sandbox gets an interactive chat panel when you run an agent with this trigger, and agents deployed with this trigger are immediately available via the API with no extra configuration needed.

Through an AI app you will get a seamless chat window for you and others to use without any configuration.

Use the chat trigger to build agents for:

• Drafting, revising, and refining content through multi-turn feedback
• Research and question answering over documents or tools
• Sales enablement and pre-call prep
• Customer support and internal help desks
• Internal copilots for product, marketing, or ops teams

💡 Use the LinkedIn content writer agent with a Chat Trigger to give it post ideas and make LinkedIn posts with no AI slop! Just add context document about yourself, role, and writing style to the document index to personalize it to your style!

Learn more about Chat Message Triggers →

Build faster with mocked data

Vellum now automatically mocks data for integrations and APIs (like Slack, Hubspot, Gmail, etc.) to make testing and optimizing agent outputs dramatically faster.

Because Vellum knows what each integration’s output looks like, mock data stays accurate enough to confidently optimize your agent before connecting your real data.

💡 Ask Vellum to mock data when making changes or adding functionality to existing agents, so iterating is just as fast as your initial building.

Learn more about Mocking Nodes & Integrations →

Node simplification

We have made a cleaner canvas and simplified node selection by deprecating Merge, Conditional, and Output Nodes are now deprecated. Here’s the upgrade:

• Merge Node → Merge Strategy: A setting available on all Node types. You no longer need a standalone Node to await multiple parallel branches.
• Conditional Node → Ports: A setting found on all Nodes under the “Routing” tab. You can perform conditional logic to route down one path or another on any Node, instead of only on Conditional Nodes.
• Output Node → Workflow Outputs: You no longer need a standalone Node to signify what your Workflow outputs. Instead, reference the Nodes whose outputs you care about directly as Workflow Outputs.

Your existing agents with these nodes will keep working, but these nodes won’t be available to add.

Learn more about Merge, Conditional, and Output Node Deprecations →

Vellum re-connection

If you navigated away from the Agent Builder mid-run or hit a network hiccup, you used to just lose the connection.

Now the Agent Builder reconnects automatically. Switch tabs, drop Wi-Fi for a moment, come back, and your execution is still right where you left it.

Learn more about Agent Builder Reconnection →

Honorary mentions

• Workflow Favorites: Star workflows to pin them to your sidebar and home page for quick access.
• Workflow Log Events: Workflows can now emit log events directly. View them in the Workflow console while building, or in the Execution Details page after the fact. Drop logs at specific points in your execution to trace flow and pinpoint issues without hunting through node outputs.
• Agent Code Downloads: Download your Agent's code as a zip file directly from the Agent Code panel.
• Custom Nodes Preference: The platform now biases toward generating Custom Nodes over Code Execution Nodes for better performance and flexibility.

That's the best highlights from January! Be sure to check out the full changelog here to see all the improvements we shipped.

These updates tighten the loop between idea, test, and adjustment while you’re building.Excited to see what you build with these.