Skip to content

Operations

The morning digest that writes itself

At nine, an agent pulls yesterday's numbers from the tools you already use, writes the summary, and posts it to the team channel. If a number looks wrong, it asks you before posting.

SolData analyst

One weekday morning

Nine o'clock. Sol is already on it

The routine fires, Sol reads the KPI sheet and yesterday's Stripe charges, and notices refunds jumped 40%. Instead of guessing, it asks Dana. She says flag it, and the digest lands in #ops with that line marked, four minutes after the hour.

  • Runs on a schedule, nobody has to remember
  • Reads Sheets and Stripe itself, no copy and paste
  • Asks a person about anything that looks off

Sol

Data analyst. Reads numbers, writes the digest, never posts a surprise.

ops Morning digest · weekdays 9:00Next 9:00

09:00 · routine started by schedule

How it is built

A routine, a short workflow, and a record of every run

Nothing to code. You describe the steps in the workspace, pick who approves what, and switch it on.

Schedules, webhooks, or a message

Three ways to start the same routine. Use whichever fits the job.

  • ScheduleWeekdays at nine, first of the month, every hour. Plain cron underneath.
  • WebhookA payment fails, a form is submitted, a deploy finishes. The event starts the run.
  • A messageAsk in the channel and the same routine runs right now, with the same steps.

Workflows chain agents

Each step hands its result to the next. Put a person in the middle wherever it matters.

  1. 1. Pull yesterday's numbersSol · Google Sheets, Stripe
  2. 2. Write the summarySol · Four lines, anomalies first
  3. 3. Human approvalDana Brooks · Only when a number is off
  4. 4. Post to #opsSol · With the flagged line

Every run is kept

What it read, what it wrote, who approved, and how long it took. Open any run when someone asks where a number came from.

Morning digest · history

  • Run 31Tue 09:04Posted, 1 line flagged
  • Run 30Mon 09:02Posted
  • Run 29Fri 09:03Posted
  • Run 28Thu 09:11Stripe was slow, retried once

Where the numbers live

The tools ops teams already use

Sign in once and every agent in the workspace can read from them. Post the result back to Slack if that is where the team looks.

All eighty integrations
  • Google SheetsRead ranges, write values and formulas, restructure sheets.
  • AirtableQuery, create and update records; inspect base schemas.
  • StripeCustomers, invoices, payments, subscriptions and disputes.
  • SlackSearch Slack, read channels and threads, post messages, edit canvases.
  • NotionSearch, read and write pages and databases.
  • Plus Postgres, HubSpot, Linear, Zendesk and more.

Before you switch it on

Questions ops leads ask

Short answers. Anything else, write to us and a person replies.

Ask us something

What if the numbers are wrong?

The agent shows its sources and the exact ranges it read, in the run history and in the digest itself. If a total does not add up, ask it in the channel and it walks through the pull step by step.

Can it post without anyone checking?

Yes, when nothing is unusual. You decide the rule: post straight away, always ask first, or ask only when a number moves more than you are comfortable with. Nothing it cannot undo goes out silently.

Does the agent keep working when we hit the plan limit?

No. It pauses, says so in the channel, and the routine waits. Nothing is lost and nothing auto-charges. Move up a plan and the next run picks up where it left off.

Tomorrow's digest could write itself

Free plan, one agent, no card. Connect a sheet, set the time, and see what lands in the channel at nine.