Computer Interaction
Analyze CSV and Excel files directly in chat
Computer can now take CSV and Excel files as input for analytical questions. You attach a spreadsheet and ask questions about the data, and the agent reasons over the contents in memory to answer. If a file is too large to fit, the agent tells you up front rather than failing silently partway through.
🔗 For more information about analyzing files in chat, refer to Core capabilities.
A faster, more capable server-side runtime
Computer's web experience now runs on a new server-side runtime that brings canvas, human-in-the-loop approvals, stop-streaming, and skills together in one place. Responses stream more reliably and long-running actions can pause for your input mid-task instead of running to completion blindly.
Bulk testing for agents in Agent Studio
You can now run a batch of test queries against an agent and review the results together, so you can validate performance and catch regressions before deploying a change. Results include response text, latency, tool calls, and errors, and you can run these tests yourself without engineering help.
🔗 For more information about Agent Studio, refer to Agent Studio.
Build App
Team Spaces
Every EPD team gets their own workspace - views, data, preferences scoped to what matters to them. Issues, Sprints and Roadmap are scoped to each team. Computer now has team memory. It knows exactly what you work on as a team, not just as individuals.
🔗 For more information about Teams, refer to Teams for Build.
Milestones for enhancements
Enhancements now support named milestones as a first-class entity, so you can break a project into dated checkpoints and track delivery against them. Create and edit milestones directly from an enhancement and assign issues to them. Issue lists can now group and filter by milestone, and with the right tools added, Computer can answer questions like "which milestones are at risk of slipping" or "what's the progress on milestone X" and create or populate milestones for you.
🔗 For more information about Milestones, refer to Milestones.
Guided Day 1 setup for admins
Org admins get a guided setup page that helps you set up your workspace for onboarding your teams. The wizard walks through essential setup tasks like setting up Parts, Teams, data sync via connectors, object customization and automations.
Links block upgrades on the record view
Links on the record view are now clearer and faster to manage. Dependency links use directional Blocked by and Blocking types (available for Enhancements too), and Parent/Child labels now read from the source item's perspective: pick Parent of to add a child, or Child of to add a parent. Links appear as compact chips in the Attributes panel and in a dedicated full-page Links tab grouped by type, link types are surfaced upfront when creating a link, and a new Replace option swaps a linked item in place.
🔗 For more information about dependency links, refer to Links and dependency management.
Skill and Workflow Builder
Copy and paste nodes on the workflow canvas
Workflow Builder now supports standard clipboard actions. You can copy a single node or select several at once and paste them, keeping all conditions, actions, and field mappings intact. Copy and paste works within a workflow and across two workflows open in different tabs, and references inside a copied group are remapped automatically so the shape of your logic is preserved.
🔗 For more information about Workflow Builder, refer to Workflows overview.
Use connection keyrings inside code and HTTP nodes
Code and HTTP nodes can now reference stored connection keyrings instead of hardcoding API keys and tokens in your script. A single node can attach more than one keyring, so an integration that talks to two systems no longer has to be split across nodes. Credentials stay out of workflow exports and move cleanly across environments.
🔗 For more information about the code node, refer to Code node.
Quality-of-life improvements across the builder
A batch of usability fixes makes the workflow canvas more predictable and consistent. Node behavior, search, filtering, and data handling all got attention, smoothing out the small friction points that add up when you build and maintain many workflows.
Mobility
Discussions on iOS and Android
You can now follow and take part in record discussions directly from the mobile apps. Comment threads that were previously desktop-only are now available on both iOS and Android, so you can stay in the loop and respond on the go.
Computer on mobile with a unified Today experience
The mobile app now offers Computer as a personal-led experience, combining search and agent chat into one primary Today screen. You start typing to see search results and send to chat with the agent, with support for stopping streaming, human-in-the-loop actions, and profile and subscription management.
Unread badges across the app
The mobile app now surfaces unread counts where they matter. The inbox shows a count in the header and defaults to New, and Updates carries a badge with an unread count, so you can see what needs attention at a glance.
Markdown and text file previews
Mobile now renders previews for .md and .txt files inline, so you can read attached documents without leaving the app or downloading them first.
Support App
A setup agent for faster onboarding
Support App now has an agent that guides setup, turning what used to be a manual configuration effort into a series of simple approvals. The goal is to shorten the time it takes to get a new customer live, with the agent frontloading the work so teams reach value sooner.
Plan next steps from the ticket side panel
Agents can now trigger a "Plan next steps" action directly from the ticket side panel, bringing agent assistance into the flow of working a ticket instead of a separate surface.
Knowledge Graph
Agentic knowledge graph search for everyone
Agentic search over the knowledge graph is now enabled for all users. The agent can traverse the graph and combine structured and syntactic search to answer questions grounded in your data, with improved citation handling for clearer, more trustworthy answers.
Merge duplicate meetings into one canonical record
Meetings that arrive as duplicates through multiple imports or integrations can now be merged into a single canonical record. The merge consolidates metadata, participants, links, and timeline events onto the primary meeting and archives the duplicate, which keeps analytics accurate and prevents sync conflicts across connected systems.
AirSync Native Integrations
Jira to DevRev sprint one-way syncing
You can now import and synchronize Jira sprints directly within DevRev. Select "Sprint" as an object type during the Jira Airdrop import to bring sprints with attributes like sprint name, state, start and end dates, and goal displayed on each sprint object. Sprint-to-issue relationships are preserved as they are in Jira, with links from Jira issues to their associated sprint objects, and you can navigate from an issue to its sprint view and manage sprint assignments using dropdowns or drag-and-drop. Sprint-issue relationships, sprints, and sprint boards sync one way from Jira to DevRev.
🔗 For more information about Jira sync, refer to Jira Software AirSync.
App customization
Smarter dependent fields on forms
Dependent field rules got a round of fit and finish. Fields can now become immutable based on another field's value, and rules support the && operator for combining conditions, giving you more precise control over how forms adapt as users fill them in.
🔗 For more information about dependent fields, refer to Object customization.