A Small Firm Handed Its Back Office to AI. Here Is What Happened.

Most small firms have a back office that runs on human attention: someone chases the new enquiries, someone cross-checks the calendars to find a meeting slot, someone drafts the same proposal for the fiftieth time. None of it lands on an invoice, all of it eats the week, and for years the only way to do more of it was to hire another pair of hands. That is the assumption a growing number of firms are now quietly walking away from.

In late June 2026, Anthropic published a customer story on its Claude site (claude.com) about a lean professional services firm of around thirty people that has moved a large slice of its daily operations onto what it calls AI coworkers: assistants that do not just answer a question but carry out a whole multi-step job in the background. The details are worth a small business owner's attention, because this is not a giant enterprise with a research budget. It is a small team, and the tasks it handed over are the same ones eating your week.

The point is not the specific tools. It is that the work a small firm treats as unavoidable overhead is increasingly the work AI is best at, and the firms wiring it in are pulling ahead of the ones still doing all of it by hand.

The busywork it quietly took over

The firm pointed its AI coworkers at the least glamorous corners of the business, which turned out to be exactly where the hours were hiding. New enquiries that once meant days of manual research and data entry now get scored for fit, filed in the customer database, and summarised to the team automatically, with the daily volume handled in a couple of hours rather than swallowing whole afternoons. An hour a day spent cross-referencing two separate calendars to find a free slot collapsed into a single request. Proposals and scopes of work that took hours to draft now take minutes.

The most telling part is who built it. The lead who set up the enquiry-handling system was not an engineer, she runs revenue. She had an idea, described what she wanted, and had a working assistant a short time later, without waiting in a queue for technical help. That is the shift underneath the headline: the people who feel the busywork most acutely can now do something about it directly, instead of adding it to a wish list nobody gets to.

From answering questions to doing the job

This is the change that snuck up on a lot of business owners in 2026. For a couple of years AI was a clever thing you asked for a draft or a summary, one question at a time. It is now something you hand a whole task to, and it works through the steps on its own: read the enquiry, check the details, update the system, tell the team. It is a shift the developer and independent AI writer Simon Willison has tracked closely through the year, AI crossing the line from answering questions to carrying out multi-step work with far less hand-holding. For a small firm, that is the difference between a tool you have to sit and operate and a coworker that just gets on with it.

That is also why the upside is so lopsided in favour of small teams. A big company has whole departments to absorb the admin. A small firm feels every hour of it, so every hour handed to an AI coworker flows straight back into the work that actually earns money. The leverage is largest exactly where the team is smallest.

What good looks like for a firm like yours

You do not need to run an engineering firm for this to apply. Any business with repetitive intake, scheduling, quoting or follow-up has the same hidden hours to reclaim. Here is what good looks like once AI coworkers are genuinely doing the back-office grind, rather than sitting unused in a browser tab:

• Every new enquiry is handled the moment it lands: sorted, logged, and summarised to the team, so nothing slips through on a busy day.
• The scheduling, quoting and follow-up that used to need a person now run quietly in the background, and free that person for the work only they can do.
• The volume a small team can handle goes up without another hire, because the admin no longer scales with the workload.
• The know-how lives in the system, not in one overloaded person's head, so it keeps working when someone is on leave or moves on.
• The owner gets their attention back for customers and craft, the parts of the business that do not run themselves.

Work that used to take days now gets done in minutes.Elli Rader, Chief Revenue Officer, Blank Metal

Easy in a case study, fiddly in real life

Here is the honest part a polished customer story tends to skip. An assistant that handles your enquiries is genuinely powerful, and genuinely easy to get wrong: pointed at the wrong task, it wastes effort; left unchecked, it makes confident mistakes on the things that matter most, a risk we wrote about in why an AI agent will do what it is told, even by a stranger. The firms getting real value are not the ones who bought the most tools. They are the ones who chose the right jobs to hand over, wired them in properly, and keep a careful eye on the parts that matter. That judgement is the whole game, and it is the piece a case study never sells you.

This is exactly what we do at NextAura. We work out which corners of your week are safe to hand to an AI coworker and which are not, build the agents and automations that quietly carry the load, and keep them pointed at the right work so the time you save is real and the mistakes stay rare. If reading this made you think of the three jobs that eat your week, get in touch and let us take them off your plate while you get back to the business itself.