CallableAI

Documentation

Tasks

Background agents that run your operation — analysing, allocating, auditing and curating, with every run visible.

A Task is a managed background agent doing an ongoing job for your business. Not the wrap-up of a single call — Specialists handle their own calls end to end — but the jobs above the calls: the analysis, allocation, auditing and outreach that a great operations or marketing manager would do if they had unlimited hours.

You describe the job once, and from then on it's simply CallableAI doing that task. No scripts, no flowcharts.

What Tasks actually do

Real examples of Tasks businesses run:

  • Reactivation sweeps — find the contacts in your CRM that have slipped through the cracks and add them to a reactivation Specialist's calling queue.
  • Lead source intelligence — track which sources convert highest and deliver a monthly report.
  • Lead allocation — route every new lead to the right person or Specialist, by rules or by judgment.
  • Compliance auditing — review your team's calls for compliance issues and flag anything that needs attention.
  • Warm-lead analysis — analyse conversations, find the warmest buyers, and update the CRM to reflect it.
  • Inbox curation — monitor your email inbox and draft personalised emails that help prospects become customers, customers become reviews, and reviews become referrals.

How you configure a Task

Every Task is set up the same way, and none of it is technical:

  1. The goal. A plain-English description of the job: "Every month, find clients we haven't spoken to in 6+ months and queue them for reactivation."
  2. The output. What a successful run should hand back, and where: a queue built for a Specialist, a report in your inbox, updated CRM records, drafted emails for approval.
  3. Connectors & permissions. Which systems the Task may touch — CRM, calendar, email, call records — and what it's allowed to do in each. Everything starts read-only; you grant write access explicitly, per connector.
  4. Triggers. When it runs: on a schedule ("first Monday of the month"), on an event ("a call ends", "a lead arrives"), or on demand.
  5. Knowledge. What it should know to do the job well — your definitions of a warm lead, your compliance rules, your tone for emails. The same briefing you'd give a capable new hire.

Runs — see everything it does

Every time a Task fires, that's a run — and every run is visible, like a routine you can audit. The Task's page shows its run history: when it ran, what triggered it, what it did, and what it produced. Open any run for the step-by-step detail — which records it touched, what it found, what it drafted, what it queued.

This is the deal Tasks make with you: full autonomy on the job, full visibility on the work. "What did it do while I slept?" is never a mystery — it's a list.

Tasks and Specialists, together

Tasks think; Specialists talk. When a Task's job produces calls to make — a reactivation list, a batch of warm leads worth a same-day call — it queues them into a Specialist's Calling Queue with full context attached. The Specialist takes it from there, handles the conversation and its own follow-through, and the Task's next run picks up the results in the data.

Guardrails

Tasks act only inside the permissions you granted, and anything customer-facing (like curated emails) can be set to draft-for-approval until you trust it to send. Every action in every run is logged — trust is verified, not assumed.