From ticket to deployment
knowvation connects to the Jira project your Salesforce team works from. When a ticket is ready, a team of AI agents picks it up and moves it through a delivery loop:
- Analyze. The Business Analyst agent reads the ticket, asks clarifying questions in Jira and writes acceptance criteria.
- Design. The Architect agent designs the solution and prototypes it in a sandbox.
- Build. The Developer agent builds it, no-code, low-code or code, in a sandbox.
- Test. The QA Engineer agent writes unit tests and tests the change in a sandbox.
- Deploy. Finally, knowvation ships a Salesforce deployment and pull request in GitHub, Azure DevOps, GitLab or Bitbucket.
The BA agent asks in Jira
Before anything is designed, the Business Analyst agent asks the questions a good BA would, as comments on the Jira ticket, such as whether "manager" means the opportunity owner's manager or the sales director. Back and forth is typical. Work stays paused until the BA agent is satisfied it has the answers and understands the work as an expert BA would be expected to.
The Architect agent designs
The Architect agent reviews the acceptance criteria, the ticket itself, and anything else from the BA at handoff. The Architect agent restates the problem concisely, explores the Salesforce org, and ultimately decides on a solution. The Architect packet includes a technical solution recommendation, the alternatives it considered and why it chose its particular recommendation. The BA and Architect agents usually talk back and forth; you can watch their conversation live, or after the fact, on the ticket.
The Developer agent builds
The Developer agent picks up the Architect's design and builds it in a sandbox through the native Salesforce APIs: configuration, low-code such as flows, validation rules and approval processes, or code. Before it builds, it looks up the skills that apply, written rules for doing one job well, such as how to write a validation rule Salesforce accepts. Its work summary packet records what was built, the choices it made along the way, and anything left open.
The QA Engineer agent tests
The QA Engineer agent tests the change against the acceptance criteria the BA agent wrote. It writes unit tests and runs functional tests in the sandbox. When something fails, the QA and Developer agents work it out together. Its work summary packet records what was tested and the results, so reviewers can see the evidence before the pull request is merged.
Back on the ticket
Every agent writes a work summary packet: context, decisions, what was recommended or built, and open questions. The packets are attached to the Jira ticket, so nobody has to chase documentation.
You stay in control
- Choose how much the agents do on their own, from fully agentic delivery to a human reviewing and approving solutions before build.
- Step in any time with an @mention. The working agent reads it on its next turn.
- Set spend limits per ticket and per agent. See spend controls.
Not on Jira?
- knowvation also picks up work from Linear and from Salesforce.