Agent

An Agent is a recurring, AI-driven automation that runs on a schedule against one or more of your playbooks and delivers the results to the right people, group of people, or Slack/Teams channel automatically.

Instead of opening Alpharun every Monday to check how a team is trending or to hunt for a specific pattern across last week's calls, an Agent does that work on a cadence you set and pushes a summary, with call evidence, straight to inboxes or Slack.

Agents live in the Performance tab and you'll find them under Agents.

Deploy an Agent whenever you're doing repetitive analysis that you'd rather have delivered to you than go fetch. A good rule of thumb: if you catch yourself running the same review hourly, daily, weekly, or monthly - checking score movement, scanning for a compliance miss, watching for a competitor mention, reminding reps of coaching goals - that's an Agent.

Agents are best for recurring, evidence-backed monitoring where the answer changes over time and specific people need to act on it. They're not the right tool for a one-off question you'll never ask again.

Access follows your teammate role:

• Editor - full access. Can create, edit, run previews, and delete Agents.

• Reviewer - view only. Can see Agents and their results but cannot create or edit them.

• Team Member - no access. The Agents section isn't shown to them at all; they only ever receive individual reports addressed to them.

Because Agent setup is Editor-only, the Agents tab is hidden in the navigation for everyone who isn't an Editor. Recipients of a report can always open the report addressed to them, regardless of role.

Templates are prebuilt Agents for the most common needs. Pick one, configure a few settings (schedule and recipients), and it's live. Three run on a schedule and produce run history and saved reports; one is a lightweight configuration over the existing coaching feature.

Weekly coaching goals - Emails each teammate's AI-generated weekly coaching goals to the teammate, their manager, or both. Managers receive a consolidated report of their team's performance, not individual copies. It draws on each person's performance against every playbook criterion, the highest-impact behaviors, what top performers do differently, and week-over-week movement. Delivery is fixed at Monday and Friday at 8am, and you can add coaching-focus guidance for managers.

Playbook score movement alert - Tracks your team's playbook criteria scores week over week (or month over month) and surfaces what moved, which reps drove the change, the specific call behaviors behind it, and a recommended coaching action for each movement. Runs weekly or monthly.

Lead burning alert - Flags reps who cycle through live leads without meaningful engagement. It looks for a high rate of ultra-short live calls versus peers, clusters of consecutively burned leads, and early playbook sections marked Not Applicable. Reports include short-call rate vs. team average and transcript quotes drawn only from calls where a live customer actually picked up. It notifies only when lead burning is detected. Runs hourly or daily.

A Custom Agent watches calls on chosen playbooks, on a schedule you set, for any pattern, risk, opportunity, or theme you describe - an objection, a missed compliance step, a competitor mention, an exceptional call, anything.

Writing the prompt. The prompt field asks "What should this agent look for?" Describe both the pattern to watch for and what you want the summary to return. A useful structure is two parts: What to look for (the behavior or theme, in plain language) and What I want to receive (e.g., teammate name, the reason it qualified, and a call quote). Your description is the source of truth for what qualifies, so be specific - vague prompts produce vague reports.

What it has access to. The Agent runs across all calls for the selected playbook within the window since its last run. It can count and research customer interactions, pull playbook criterion scores, rank teammates by overall score, and ground its findings in transcript quotes and your knowledge base. It can filter the calls it examines by date range, call duration, outcome, playbook criteria, signals, teammates, contact name, and your custom fields. Reports come back in a clean summary with quotes that link directly back to the source calls.

Cadence. Custom Agents run hourly, daily, weekly, or monthly. Hourly runs fire only when there's new call activity and, like daily runs, notify only when something actually qualifies (alert mode). Weekly and monthly runs always deliver a report on schedule (digest mode). Weekly lets you pick one or more weekdays; all timed runs use US Eastern time.

Preview before you commit. Use the Run preview button on the Agent's detail page to test it. A preview exercises the Agent for real and saves its results, but sends no notifications - so you can refine your prompt until the output looks right before anyone receives it.

Past runs. Every run is recorded on the Runs tab of the Agent's detail page, with the run time and a status (Queued, Running, Completed, Failed, or Canceled); preview runs are tagged with a Preview pill. Open any run to see the full detail, including the Agent's reasoning, and to cancel a run that's still in progress. The Results tab shows the saved reports themselves, including who they went to and whether they were delivered by email, Slack, or both.

You control who gets each Agent's output and how it reaches them.

Who. Recipients depend on the Agent. Custom Agents let you notify individual teammates and groups (groups expand to their members at send time). The score movement and lead burning templates notify managers - all managers or specific ones - and let you CC additional Editors or Reviewers. The coaching template sends to each team member, each manager, or both.

How. There are two delivery channels:

• Email - always private, one report per recipient.

• Slack - either a direct message to each recipient (matched to their Slack account by email) or a post to a Slack channel you choose. Note that anyone in a chosen channel can see that Agent's output, so channels are best for team-wide summaries and DMs for anything sensitive. Personal reports like coaching goals are kept private and can't be posted to a channel.

Results are always saved in Alpharun regardless of the delivery channel.

Learn how to connect with Slack