Small teams rarely lose on ideas. They lose on hours. I’ve seen it from both sides, working with startups and with well-known brands, first as a freelancer and then as a head of marketing. Along the way I kept meeting founders with sharp ideas and small teams, buried in work that should run itself. A big team isn’t smarter. It has more hands, more structure and someone who’s always on. AI is finally good enough to close most of that gap, and that’s why I build what I build. What follows is how I think about closing it, and which parts should never be handed over.
What big teams actually have
Take away the office and the budget, and a well-run big team has three things a small one usually doesn’t.
Specialists. Someone whose whole job is SEO. Someone who only thinks about paid ads, or email, or the sales pipeline. In a small team, one person covers all of it, so each job gets whatever attention is left over.
Process. Briefs, calendars, review steps, checklists and handovers. Big teams write down how the work gets done, so it happens the same way every week, even when people are away. Small teams keep the process in someone’s head, usually the founder’s.
Coverage. Someone is always on. Enquiries get answered in the evening and over the holidays. Follow-ups don’t slip because one person is off sick. Nothing waits for the only person who knows how.
None of that is talent. Small teams often have more of it per person. What they lack is depth.
Why hiring never quite closed the gap
The usual answers were to hire, outsource or buy more software. Each helps, and each has a catch. You can’t hire a specialist for every job on a small budget. Freelancers and agencies need briefing, managing and chasing, and I say that as someone who started out as a freelancer. Software gives you capability, but every new tool is one more thing somebody has to run. So the founder ends up as the glue between all of it, and glue doesn’t scale.
What changes when AI can do real work
For the first time, software can do the work, not just store it. AI can draft the post, write the ad variations, research the keywords, reply to the enquiry and update the CRM. That starts to close all three gaps at once.
- Specialists. Each job can have its own AI employee with its own brief, instead of one person spread across everything.
- Process. An AI employee needs its goal, steps and standards written down. The process finally leaves the founder’s head, because it has to.
- Coverage. It doesn’t get tired, take holidays or get pulled into something else. Every enquiry gets a reply in minutes, in your voice.
That’s the idea behind Gridwork: AI employees for content, sales, ads and SEO that do real work inside the tools you already use. It reaches beyond startups, too. A family juggling school emails and forms is a small team. So is an estate agent working on their own. Leila and Onshow are the same bet, made for parents and for agents.
People still make the calls
Handing AI the work doesn’t mean handing it the business. The principle I build everything around is simple: AI should do the work, and people should make the calls.
A call is any decision where judgement, taste or accountability matters. What we say, and to whom. What we spend. What we promise. When to stop. In practice, that means an approval step. The AI prepares the work, and a person says yes before anything is published, sent, spent or promised. In Gridwork, nothing ships until you approve it.
In a normal week, that might look like this. On Monday you approve the plan. Through the week, drafts arrive for review and enquiries get answered. On Friday the numbers land with a recommendation, and you decide what changes. The AI did most of the work. You made every call that mattered.
So the job of a small team changes shape. Less doing, more deciding. You become the editor, the strategist and the person who knows the customer best, which is the work most founders wanted to be doing all along.
What should never be automated
Some things stay with people, however good the models get.
- What you stand for. Positioning, values, what you will and won’t do. AI can draft the options. Choosing is the job.
- Talking to customers. AI can summarise every call you record. It can’t sit in the conversation for you. Build with the people you’re building for, and that means hearing them yourself.
- Relationships that matter. The first conversation with a key customer. The apology when something goes wrong. Hiring, letting someone go, a hard conversation with a partner.
- Anything you can’t take back. Large spends, contracts, prices and commitments with your name on them. AI can prepare the recommendation. A person makes the decision.
- Taste. Knowing what good looks like for your brand, and noticing when something is technically fine but isn’t you.
- Accountability. When something goes wrong, a person answers for it.
“You can hand AI the work. You can’t hand it the responsibility.”
Start with one role
You can start this week.
- List the work that repeats every week, including the small jobs nobody mentions.
- Mark each item as a call or as execution. Calls stay with people.
- Pick the execution job that eats the most hours and is easiest to check.
- Write the brief as if for a new hire: the goal, the steps, what good looks like and the words to avoid.
- Hand it to an AI employee with an approval step, and review everything it does for the first couple of weeks.
- Add the next role only when the first one runs without you chasing it.
Don’t rush that last step. Big teams are built one reliable role at a time, and so is this one.
Come and figure out the rest with us
None of this is finished. The tools change fast, and the best playbooks come from people trying things and sharing what worked, and what didn’t. That’s what my free community on WhatsApp is for: founders, marketers and operators learning to grow with AI, with me in there every week. Come and join us.