One loop that runs your customer operation.
Support pressure, pipeline stalls, usage signals, and knowledge gaps all compete for attention. The operator loop turns that noise into one focus and a short list of actions — each with its reasoning attached.
The four steps
- Collect signals — conversations, stage changes, usage events, CSAT, health-signal rules, and knowledge gaps feed a shared picture.
- Pick one focus — HAL weighs support pressure against pipeline momentum and names the single thing that most deserves you today.
- Propose actions — the focus arrives with drafts attached: a follow-up email, an article to write, an outreach note.
- Learn from feedback — approve, edit, dismiss, or snooze. Dismissed patterns stop resurfacing; approvals sharpen what gets proposed.
What HAL optimizes for
- Momentum over activity — the focus is the item most likely to unstick revenue or prevent churn, not the loudest inbox.
- Evidence over vibes — every recommendation shows its work: the conversation, the deal stage, the usage event behind it.
- Small queues — a queue of three you finish beats thirty you hide.
Why approvals matter
The loop's ceiling is deliberately human. Recommendations are drafts — nothing is sent, published, or moved until someone approves it. That's not a limitation; it's the design that lets a small team trust an automated system with customer-facing work. Over time the approval loop is also the training signal: HAL learns which suggestions your team acts on.
Where the loop lives
- Overview — today's focus and top actions, per project.
- HQ — the same loop rolled up across projects, ranked.
- Work — the backlog of proposals, drafts, and initiatives.
Feeding it well
The loop is only as good as its signals. Connect product events via tracking, configure health signals, and keep the operating context current so "important" means what your business means.