The workspace your team writes
Your mission, strategy, goals, the customers you serve and their problems, and every project and ticket beneath them.
Meet Helen. She checks your reasoning.
Helen is the AI built into Helen's Foundry. She doesn't improve your writing. She reads your decisions and the work beneath them, and tells you when the two stop agreeing. A person decides what happens next.
This strategy commits to enterprise-only, but the goal beneath it doubles self-serve sign-ups. These two conflict.
What she reads.
Helen reads what your team has written down and what your tools can already see. Nothing else.
The repositories you connect
Through our GitHub App: pull request and issue titles, diffs, and check status, for the repositories you authorize.
The other tools you connect
The items the connected account can see in tools such as Linear, Notion, or Slack, so they can appear as evidence in your workspace.
When she checks.
On every pull request
She checks that the pull request links to its ticket and that the decisions above that ticket are signed off. The result shows up as one GitHub check and one comment that updates on each push.
On your plans, every day
She checks the whole line from strategy down to each ticket: a goal that contradicts the strategy above it, work built on a project you closed, a target with no work beneath it.
24 checks in 6 groups, from plain rules over owners and links to an AI reading two decisions side by side.
See every checkWhat a flag looks like.
A flag names the two things that disagree and says why. Here, the strategy you signed off on Monday commits to enterprise accounts, while a goal beneath it still doubles self-serve sign-ups.
She logs every flag she raises, so the reasoning behind a decision outlives the decision.
Contradicts its parent Raised today by Helen
Strategy, signed off Monday
Win enterprise accounts first
Goal beneath it
Double self-serve sign-ups
The strategy commits to enterprise accounts, but the goal beneath it doubles self-serve sign-ups. These two conflict.
Propose closing the goal Acknowledge Dismiss
A person decides every flag.
Every flag goes to Reconcile, the page where flags and proposed changes wait for a decision. Helen's GitHub check blocks a merge only if you make it a required check in GitHub.
- Fix it
- Take the action the flag offers, such as assigning an owner, closing work as obsolete, or proposing a change. The flag clears on the next check.
- Acknowledge
- Record that you know about the flag without claiming a fix.
- Dismiss
- Record that the flag does not apply.
What she never does.
- She never merges, closes, or edits a pull request.
- She never changes a decision on her own.
- Agents can edit tickets directly, but a change to a signed-off strategy, mission, vision, or domain model arrives as a proposal that a person applies or declines.
Your content stays yours.
We do not sell your content, and we do not use it to train our own models. Helen sends the relevant content to model providers whose terms prohibit training on business inputs. We record prompts and model outputs only if a workspace admin turns that on. It is off by default.