TL;DR: AI tools can only help with the context they can see. If story intent is scattered across Jira, Slack, docs, and memory, humans and AI both have to rebuild it. I wrote about using the backlog as shared working memory so teams can resume work without replaying everything: https://imdone.io/articles/shared-working-memory-for-humans-and-ai

Hey 👋,

The most expensive part of a handoff is not finding the ticket.

It’s rebuilding intent.

What were we trying to learn?

Why did the acceptance criteria change?

Which assumption was still open?

What did the last person discover before they had to stop?

That context usually lives across Jira, Slack, meetings, screenshots, docs, browser tabs, and memory.

AI coding tools run into the same problem.

Claude Code, Copilot, Cursor, and similar tools can only help with the context they can see. If the real story is scattered, the AI gets a thin slice of the work and the team still has to replay the rest.

That’s the shared-context problem.

I wrote about why I think the backlog can become shared working memory for both humans and AI.

Not more documentation for its own sake.

More like:

- the hypothesis stays with the story

- the success signal stays visible

- assumptions are written down

- design notes and diagrams stay close to the work

- implementation plans can be resumed

- the next person, human or AI, does not have to reconstruct intent from scratch

The question I keep coming back to is:

If a developer or AI assistant picked up this story tomorrow, would they know the outcome, current decision path, and evidence gate without asking someone to replay it?

If the answer is no, the team has a shared-working-memory problem.

Here’s the article:

https://imdone.io/articles/shared-working-memory-for-humans-and-ai

If this resonates, reply and tell me where your team loses the most story context.

I’m especially interested in where AI tools help in the moment but still make you rebuild context later.

— Jesse