Andi April 2026 Release: From Ticket to Tested, in One Conversation
The hardest part of a Salesforce build is usually everything that happens before the build. A Jira story missing acceptance criteria. A 30-page PRD that needs to be turned into epics. A spec doc full of requirements that have to be checked against what’s already in the org. By the time the actual work starts, half the day is gone.
These are the gaps the April release was built to close.
This month’s update introduces the Atlassian (Jira) Integration in early access, native PDF and DOCX uploads in the Andi chat, meaningful upgrades to the QA Agent, and the full expansion of Supplementary Context across every Andi sub-agent — alongside a set of smaller enhancements that make day-to-day use faster. Together, these changes extend Andi beyond the Salesforce org and into the surrounding systems where delivery actually happens.
Here’s what’s new.
Atlassian (Jira) Integration: Andi expands beyond Salesforce
The Atlassian (Jira) Integration is now in early access. The connection model is the same one teams already know from the Salesforce connection: securely “bring your own Jira login” to authenticate, directly from the upper right-hand corner of the Andi chat window. Once connected, Jira becomes a first-class participant in the Andi conversation. Andi can read the stories assigned to you, pull a story’s full details into chat, identify gaps in the requirements, draft acceptance criteria using your org context, generate a step-by-step development plan, and write all of it back to Jira — without ever leaving the conversation.
This is the first step in extending Andi across the broader delivery toolchain, with deeper Atlassian capabilities (including Confluence) planned in upcoming releases.
Chat File Uploads: Bring documents into the conversation
Salesforce delivery is rarely driven by a clean ticket. It’s driven by PRDs, design docs, technical specs, and standards documents — often dozens of pages long. Until now, getting that content in front of Andi meant copying sections into the chat manually.
That’s no longer the case. Users can now attach PDF and DOCX files directly in the Andi chat — either via the “+” button or by dragging and dropping a file into the conversation. Andi processes the document instantly and can reason over it the same way it reasons over your org context, applying the document’s content directly to whatever you’re working on.
This unlocks a meaningfully faster workflow for early-stage planning and impact assessment. Teams can hand Andi a PRD and have it draft epics and user stories. They can run an impact analysis against the existing org to surface conflicts before any code is written. And they can ask Andi to produce a complete implementation plan straight from the spec.
QA Agent Enhancements
The QA Agent went generally available in March. This month, it gets three upgrades focused on giving teams more control and confidence over what’s being tested.
First, before any test runs, Andi will now present the acceptance criteria it plans to test against and wait for confirmation. This makes the testing process collaborative rather than fully autonomous — teams can review, adjust, or expand the test scope before execution begins.
Second, tests can now run as a specific user via “Login As.” This is critical for validating profile-based logic — permission visibility, ownership rules, role-based access — that can’t be tested under a system user. Andi will authenticate as the specified user and execute the test from their perspective.
Third, after a test is completed, Andi will offer to automatically delete the test records it created. No more manual cleanup, no orphaned data lingering in the org.
Supplementary Context: Now live across every agent
When Supplementary Context entered beta in March, it was scoped to org context queries — a starting point that let teams begin uploading their internal standards and seeing them reflected back in Andi’s responses. This month, that scope expands to every Andi sub-agent.
The Flows, Apex, QA, and Validation Rule agents will now actively search your uploaded guidelines and standards as part of their work. That means the flow Andi builds will follow your Flow Design Standards. The Apex class it writes will match your team’s coding conventions. The test data it generates will use the formats specified in your QA guidelines. The validation rule it deploys will respect your error message standards. Every output, across every agent, reflects your team’s specific way of working.
Additional features and enhancements
A few quality-of-life improvements rounding out the release:
- Quickly Copy Your Prompts: A one-click “Copy Prompt” button has been added next to your own input messages, making it easier to share or reuse prompts that worked well.
- Deploy Flows as Inactive: You can now instruct Andi to “create the flow but do not activate it,” giving you a chance to review the logic in Salesforce Setup before it goes live.
- Smarter Naming Resolution: Andi’s understanding of Global Picklist Labels and plain-English field names has been broadened, so it can more intuitively and automatically resolve them to the correct underlying API names without explicit clarification.
What this release means for teams
February gave Andi the ability to understand your org. March gave it the ability to act on that understanding across the full build-test-modify lifecycle. April starts to extend that intelligence outward — into the systems where Salesforce work gets planned, documented, and tracked.
The Jira integration means a story doesn’t have to be manually translated from ticket to chat to org. Chat file uploads mean a 30-page PRD can become epics, an impact analysis, and a build plan in a single conversation. The QA Agent upgrades make automated testing more collaborative and more thorough. And with Supplementary Context now live across every agent, the work Andi produces — at every step — reflects how your team actually wants things built.
For teams using Andi in production, this release reduces the friction between tools as much as the friction inside them. The delivery workflow is starting to feel like one connected conversation, not a series of handoffs.
