AI assistants and agents
SaaSy includes an AI chat assistant and a set of background agents that watch your workspace data. The assistant answers questions about your data, agents flag things they noticed, and anything an agent wants to send or change on your behalf waits for you to approve it first.
Ask questions in Ask SaaSy
Ask SaaSy is a chat page that answers questions using your workspace data. It requires the Growth plan or above (feature key ask_saasy); on Starter the page shows an upgrade prompt instead.
- Open Ask SaaSy from the sidebar.
- Type your question in the composer — the placeholder reads "Ask SaaSy anything..." — and send it. Answers stream in as they are generated.
- Read the Sources list under an answer. Each cited source carries a confidence percentage. If an answer has no sources, treat it as unverified.
- Watch for the amber advisory notes. Answers touching legal, tax, or compliance topics carry a statement that the response is not professional advice, plus a link to the data-handling statement.
- Use the controls under an answer to copy it, regenerate it, rate it, or delete it.
- Past conversations are listed in the sidebar; New chat starts a fresh one.
A meter under the composer shows messages used against your daily cap and when it resets, when your plan sets one.
Answers can be wrong, incomplete, or out of date. Check the cited sources before acting on anything consequential, and never treat an answer as legal, tax, or accounting advice.
Review what agents noticed
Proactive Actions is a read-only log: "What your agents detected and acted on, newest first." Typical triggers are a churn risk, an SLA breach, or a disconnected integration.
- Open Proactive Actions.
- Each row shows the action type, the agent that raised it, a priority badge (Critical, High, Medium, or Low), a status badge, and a timestamp. Statuses include Dispatched, Duplicate, Rate Limited, and Failed — so the log also records when the system declined to act.
- Click Resolve once you have handled the underlying issue, or Dismiss if it needs no action.
- Refresh pulls the latest entries.
This page is a record, not a queue you must clear. Nothing on it is waiting on your approval.
Approve or reject pending agent actions
Pending Agent Actions is the approval gate. Its subtitle states the contract: "Review and approve or reject actions your agents want to take." When an agent proposes a tool call that needs your sign-off, it appears here and does not run until you approve it.
- Open Pending Agent Actions. Only actions awaiting review are listed.
- Read the proposal. For a drafted email you see the To, Subject, and Body exactly as the agent composed them. For other action types you see the tool name and its parameters as a labelled list.
- Check the Agent that proposed it and the time it was proposed.
- Click Approve to let the action run, or Reject to stop it. Approving records the action against your workspace's outcome ledger, so approved agent work leaves an auditable trail.
Review the content, not just the intent. An approved email is sent as written, and this queue is the last point at which you can change your mind.
If you want to control who can approve agent actions, set that up under Team & roles rather than assuming a default.
Run a playbook
Playbooks are named routines that turn your data into a written summary. The page notes that Friday Brief is live today and more playbooks roll out as the platform grows; others in the catalog are shown as Coming soon.
Friday Brief is a week-in-review of customer health, MRR, and open alerts, delivered every Friday afternoon through the channels you configure.
- Open Playbooks, then Friday Brief.
- Click Run now to generate a brief immediately instead of waiting for the schedule. Generation takes a few seconds; if it times out, the brief may still arrive — refresh shortly after.
- Read the brief. It is organised into Summary, Highlights, Risks, Opportunities, and Action items.
- Browse Recent history to reopen an earlier brief.
- Click Settings to change the schedule. You can enable or disable the schedule, set the day and Timezone (an IANA name such as
America/Los_Angeles), and choose delivery channels: in-app is always available, plus a Slack DM to the owner or a post to a Slack channel. Click Save changes.
A brief summarises the data in your workspace at the time it runs. Its action items are suggestions for you to weigh, not instructions that were carried out.
For the underlying signals these features read from, see Health scoring, Churn prediction, and Alerts. Plan availability is covered in Modules and plans.