214 messages sorted, 61 flagged
Lilach Bullock runs a five-person business. An agent reads every new message from LinkedIn, website forms, and email, and checks each one against written rules: does it name a service, name a budget, and come from a company domain? It drafts a reply for messages worth answering, tags the rest, and logs each one in Airtable with a summary line. It sends nothing. A person clicks approve first.
Start with one job only. Fill in the three lines. Keep the bot on that job.
The one job this bot does: ______________________________
What it may do on its own: ______________________________
What it must ask me first: ______________________________
Two lines from this case, side by side. No new figures.
Lilach Bullock runs a five-person business. An agent reads every new message from LinkedIn, website forms, and email, and checks each one against written rules: does it name a service, name a budget, and come from a company domain? It drafts a reply for messages worth answering, tags the rest, and logs each one in Airtable with a summary line. It sends nothing. A person clicks approve first.
Over four weeks the agent did the first pass on 214 messages and flagged 61 for a human reply. Assistant time fell from about 45 minutes a day sorting to about 8 minutes a day checking, about 14 hours saved in a month. Catch, from the owner: it once drafted a reply with an old price from a page pulled in weeks earlier, and only the approval step caught it.
The after line is the source's claim, not our measurement.
smalloffice.space · Get a case like this in your inbox each morning. Subscribe free
Read this case on your phone now, and keep it for later. Type your email. We send this case to that address, and you get the next cases each morning too.