Connect both In Parallel and Jira to the same AI assistant and you can keep your delivery board in step with your execution layer from a single chat — turn the latest decisions, goals, and drift from In Parallel into Jira Epics and tasks, and reflect what's been delivered in Jira back into your In Parallel plan.

This works with the boundary between the two tools, not against it. In Parallel holds the execution layer — goals, decisions, accountable actions, and what changed. Jira holds the delivery work. The assistant reconciles the two so neither has to become the other. (If you're new to that split, see Integrations.)

With both connectors enabled in one chat, you can:

The basic flow is always the same: connect both tools → ask the assistant to sync → answer a couple of clarifying questions → review what it changed. The rest of this guide expands each step and shows the optional automation.

These steps use Claude Desktop with both connectors enabled, because the scheduling step at the end relies on Claude Desktop's scheduled tasks. The reconcile-and-sync pattern itself works in any AI client that can use two connectors at once. The two connectors can be named anything in your client — what matters is that one points at In Parallel's MCP and the other at Jira.

A one-off chat works, but if you'll sync the same board regularly, give the assistant a stable place to work from. In Claude, create a Project for the pairing (for example "In Parallel ↔ Jira") and add two things to its knowledge:

With that context attached, every run starts from the same ground rules and you can keep your prompt short.

In a chat with both connectors enabled, ask in plain language — for example:

"Can you update the Growth Project in Jira using the latest updates from In Parallel?"

The assistant pulls the current state of the matching In Parallel Workspace — the execution plan, recent decisions, action items, and the drift report — and reconciles it against your Jira project before changing anything. You'll see it lay out a short plan of what it intends to do (create Epics, create tasks, verify) and which connectors and tools it's using.

Because syncing is a judgement call, the assistant checks with you before it writes. Typical questions:

Pick the option that matches how your team uses Jira. This is the human-in-the-loop step — nothing is written to Jira until you answer.

Once you've answered, the assistant uses the Jira connector to create and update issues — new Epics for sub-projects, tasks under them, and status changes on existing issues. It works issue by issue and shows progress as it goes.

When it's done, the assistant verifies the board matches In Parallel and summarises what changed: which Epics were created, which issues were re-parented or updated, and which were moved to Done. Skim it before moving on — this is your confirmation that the board now mirrors your execution layer.

When the assistant creates or updates a Jira issue from an In Parallel record, it notes the origin in the issue — for example a Source: decision, Apr 24 line in the description. That keeps the chain visible: anyone in Jira can see that an issue came from a specific In Parallel decision or action, rather than appearing from nowhere.

Because the assistant read In Parallel's drift report while reconciling, it can flag execution risk that a Jira board alone wouldn't show — overdue action items, stale decisions still sitting unresolved, the program tracking at risk. Treat these as prompts to act in In Parallel (confirm a decision, reassign an action), not as Jira tickets.

The sync runs both ways. When work is completed in Jira, ask the assistant to bring that back:

"Are there any recent updates in the Jira project we should reflect in the In Parallel execution plan?"

It pulls the issue's recent activity and proposes the matching In Parallel change. For example, a Jira issue marked Done maps to approving the In Parallel decision behind it — moving it from proposed to approved, with a rationale noting the Jira completion. As with the push direction, it asks before writing. Reflecting a delivered item back this way also clears the matching drift signal in In Parallel.

Instead of re-asking each time, you can have the assistant build a live status view wired to your connectors — a page that pulls straight from the Jira board (and your In Parallel plan) and refreshes when you open it. It can group issues by Epic, show priorities and due dates, flag overdue items, and headline the counts (To Do / In Progress / Done / Overdue).

To keep both sides aligned without lifting a finger, ask the assistant to schedule the reconcile — for example a daily morning sync that pulls the latest from In Parallel and reconciles it into the Jira board, running end to end without stopping to ask. In Claude Desktop this becomes a scheduled task you can see, run on demand, or turn off.

Note: Claude Desktop scheduled tasks only run while the desktop app is open. If the app is closed at the scheduled time, the run happens the next time you open it. Run it manually once after creating it to confirm it behaves the way you expect.