November 26th, 2025
We’ve added support for 5 new Native Integrations:
November 24th, 2025
Workflows can now output documents, images, videos, and audio. The simplest way to build Workflows with multi-modal outputs is to use Agent Builder—describe what you want to create, and Agent Builder will generate Custom Nodes that integrate with the relevant generation APIs. Multi-modal outputs are displayed in the console, monitoring layers, and AI Apps.
November 24th, 2025
We’ve updated the Workflow Sandbox layout. The Agent Builder panel is now top-level, we’ve removed breadcrumbs, and simplified the navigation tabs to Sandbox/Evaluations/Deployments in the center, giving you more space for the canvas and console.
November 22nd, 2025
You can now click the “Compare” button when publishing a Workflow to see a code diff between your Sandbox and the latest deployed Release.
November 20th, 2025
We’ve added support for 27 new Native Integrations:
November 19th, 2025
We’ve added support for Kimi K2 Thinking via Fireworks AI. This model provides advanced reasoning capabilities for complex problem-solving tasks.
November 19th, 2025
You can now add a Set State Node to your Workflows to update state variables during execution. This allows you to maintain and modify state across your Workflow runs, such as tracking counters, accumulating chat history, or storing intermediate values.
To use the Set State Node, add it to your Workflow and configure one or more state operations. Each operation specifies:
For example, you can increment a counter (counter = counter + 1) or append to chat history (chat_history = chat_history.concat(Agent.ChatHistory)).
November 19th, 2025
You can now check the status of a Workflow execution using the Check Workflow Execution Status endpoint. The endpoint returns the current execution state (PENDING, FULFILLED, REJECTED, etc.), along with outputs and an execution detail URL once the Workflow completes. This makes it easy to poll for completion when using async execution patterns.
November 19th, 2025
We’ve added support for asynchronous Workflow execution, allowing you to initiate long-running Workflows without waiting for completion. Use the new Execute Workflow Async endpoint to start a Workflow execution and receive an execution_id immediately. Async executions automatically queue when you exceed your concurrency limit, making this endpoint ideal for batch jobs where you don’t need everything to complete at once. You can initiate many executions quickly and they’ll process as capacity becomes available.
Track execution completion and access outputs using the Check Workflow Execution Status endpoint. See our Batching Executions guide for polling patterns and webhook alternatives.
November 19th, 2025
You can now configure Workflows to execute automatically based on schedules or external events. Previously, Workflows were only ever executed when explicitly invoked (e.g. via API or through an AI App). Now you can configure Workflow Triggers directly in the Workflow Sandbox by adding either a Scheduled Trigger or an Integration Trigger.
You’ll find Triggers as new node types at the bottom of the “Add” panel.
Scheduled Triggers run Workflows on a recurring schedule using cron expressions. You can describe your schedule in plain English (like “every day at 9am”) and Vellum will convert to a cron expression, or provide the cron expression directly.
Integration Triggers invoke your Workflow automatically in response to webhook events that come from Native Integrations (e.g., Slack, Gmail, GitHub, etc). After authenticating with the Native Integration, you can configure the specific event type and related settings. For example, with a Linear Integration Trigger, you can specify a team ID to only trigger on issues created within that team.
When testing your Workflow with Triggers, add a new Scenario from the Inputs panel and select the option that corresponds to your Trigger. This automatically populates the Scenario with the event payload attributes from the integration, letting you specify custom values and test with simulated Trigger data.
When you deploy a Workflow Sandbox containing a Trigger, the Workflow Deployment executes automatically based on that configuration. Scheduled Triggers run according to the cron schedule, and Integration Triggers fire when the specified event occurs. You can view and enable/disable Triggers on the Workflow Deployment overview page.
Triggers mark a big step towards using Vellum workflows to build Agents that operate autonomously in response to real-world events. We can’t wait to see what you build with them!
November 19th, 2025
Personal API Keys now respect your user permissions and roles. If you don’t have the necessary permissions for an action, API requests will return a 403 error. For example, you’ll need the Deployment Editor role to deploy Workflows via API using your Personal API Key.
November 18th, 2025
We’ve added support for Gemini 3 Pro Preview via the Gemini and Vertex AI APIs, along with the new thinkingLevel and mediaResolution API features.
November 18, 2025
Agent Builder will now create visual diagrams to help convey its plan while you build.
You can click on the diagram to expand it:
November 18, 2025
You can now upload CSV, TXT, and MD files to Agent Builder to streamline building new Workflows and Agents.
Below are just a few ways you could use this:
November 16, 2025
You can now upload PDF and Image files to Agent Builder to streamline building new Workflows and Agents. Below are just a few ways you could use this:
November 13th, 2025
We now support OpenAI’s GPT 5.1 model via Chat Completions and Responses endpoints.
November 13th, 2025
We’ve redesigned Nodes in Workflows with a cleaner, more condensed look that helps you see more of your Workflow at a glance.
November 13th, 2025
You can now configure generic private code package repositories for Python that support basic authentication. This allows you to pull dependencies from custom PyPI-compatible repositories. You can use these private repositories when using Code Execution Nodes in Workflows and also in Code Metrics.
To add a private repository you can navigate to the Private Package Repository page here. You can also add a repository by clicking the “Add Private Repository button” in the dropdown of the new Repository field when adding a package.
November 12th, 2025
You can now enable the use of Web Search Tools for Anthropic Claude models.
November 11th, 2025
Evaluation Metrics can now reference Environment Variables as inputs. This is useful for custom Metrics that need secret values like API Keys. Previously, you’d reference workspace-level secrets directly.
November 10th, 2025
You can now add Web Search Node to your Workflows from the Workflow Sandbox UI. Pass in a query and it’ll return the top 5 search results from the internet, including the page title and URL for each result.
This node used the Serp API Integration under-the-hood to actually perform the web search.
November 10th, 2025
We’ve made several improvements to the console for viewing and debugging Workflow runs.
November 8th, 2025
Agent Builder now plays an audio chime when it’s done building your Workflow if you switched to another browser tab or application while waiting.
November 8th, 2025
You can now run evaluations for Workflows that use third-party integrations like Notion or Slack. Previously, these evaluations would fail when they tried to invoke an integration. Now they’ll succeed as long as you’ve authenticated with the integrations used in the Workflow.
November 7th, 2025
We’ve simplified how you manage Environment Variables, Secrets, and API Keys in Workspace Settings.
November 5th, 2025
You can now customize the color and icon of your Workflows. Just click the more menu, then Edit Details, and click on the icon shown. We’ll then show the icon on the Workflows homepage, in your AI App, and more.
November 1st, 2025
We’ve added support for 14 new native integrations:
You can now connect these integrations directly to your Workflows in Agent Builder, Agent Nodes, or Custom Nodes.