Reskilling Your Team in AI: A Starting Point for Small Businesses
Sign in to watch the recording
The write-up below stays open to everyone. The video itself needs a free account.
About
A free 45 minute session with Breakthrough Social Enterprise on where AI genuinely saves a small business time, and a practical pack to plan your next steps.
Recap
Jasmine, Head of Partnerships and Growth at Breakthrough Social Enterprise, ran a practical session on bringing AI into a small team. Breakthrough teaches AI and digital skills to people the digital economy often leaves out, and everything in the session comes from what they teach their own learners and how they changed their own team.
Her main point: treat AI adoption as a work design project, not a technology project. When people hear "technology project" they assume they need skills they don't have, and they worry about being replaced. When you start from the work itself, people can see how their role gets easier, and they are far more willing to try.
She set out four levels of use. Ask (one off questions, like summarising a document), assist (reusing saved context for the same task again and again), automate (a set workflow, such as drafting a reply to every website enquiry) and act (an agent working towards a goal on its own). More autonomy is not automatically better. Many small teams get most of the value from the first two.
AI should not lead when the cost of being wrong is high, when the work depends on trust or empathy, when you can't say what good looks like, when the data is sensitive, or when a decision affects people's jobs. Her answer is the human sandwich: a person sets the goal, the standard and the limits, AI does the drafting or analysis, and a person checks the result and stays accountable for it.
On tools, she advised starting with the simplest stack possible, using what you already pay for (Copilot if you're on Microsoft, Gemini if you're on Google), checking where your data goes, and counting the real cost, which is usually the time it takes people to learn a tool rather than the subscription. Breakthrough has no training budget, so the team runs a weekly hour where colleagues show each other what they have built.
Key takeaways
- Start with your work, not the tools. Automate tasks, not jobs.
- More autonomy is not automatically better.
- Context makes AI useful. Give it the background it needs, not just a clever prompt.
- Automate processes, never judgement. The responsibility always stays with the person.
- Write a short AI policy: which tools are approved, what data can go in, what needs checking, and who owns the outcome. Review it every quarter.
- Pick first projects that are frequent, time consuming, low risk and easy to check.