Why AI Council is the most interesting free AI tool for UK SMEs right now
Most AI tools give you one voice that agrees with you. Our new free tool puts six advisers in a room to argue about your decision before you spend a penny on it. Here is why that approach is genuinely cutting edge, and how it helps a small business in practice.
Try it first: the tool is live and free at expedientai.co.uk/tools/ai-council. Three councils per hour, no login, no email. This article explains why we built it the way we did.
The problem with a single chatbot
Ask one AI model whether you should hire your first employee, and it will usually say yes, then write you a supportive essay. Large language models are agreeable by training. They mirror your framing, adopt your optimism, and rarely tell you the one thing you needed to hear: the objection you had not considered.
That is not a flaw you can prompt away. It is a property of asking a single voice. Real decisions in real businesses survive because they get challenged, by a sceptical accountant, an optimistic salesperson, a cautious operations manager, a customer who does not care about your plan. A small business rarely has that room full of people available on demand. AI Council builds the room.
What AI Council actually does
You describe a decision: hire the first employee, raise prices, open a second location, drop a product line. The tool convenes six advisers, each a distinct archetype with a different priority: growth, money, operations, customers, risk, and the long view. They deliberate in sequence, react to each other, and disagree where they genuinely disagree.
Then it hands you a structured verdict, not an essay:
- Ranked objections. The strongest arguments against your plan, ordered by how much they matter.
- Risk flags. The specific things that could go wrong, named plainly.
- A kill line. A concrete threshold that tells you when the decision stops making sense, for example how many new bookings must land to cover the new wage.
- Cheap tests. Small, low-cost experiments you can run this week to learn before you commit, like a paid trial month with a subcontractor instead of a permanent hire.
That structure is the point. An essay is easy to nod along to. A ranked list of objections with a kill line is something you can act on.
Why this is cutting edge, honestly
Multi-agent deliberation is one of the most active areas of AI research right now, because a single model's blind spots are exactly what a panel of differently-prompted agents can surface. The cutting edge part is not that we use many models. It is that we force divergence, then compress it into a decision-ready output.
Three things make it work in practice:
- Divergent archetypes, not clones. Each adviser is prompted to hold a different objective function. The money adviser cares about cash, the customer adviser cares about churn, the risk adviser cares about downside. They are designed to collide.
- Adversarial sequence, not parallel summaries. Advisers see each other's points and push back, so the output is a deliberation, not six independent opinions stapled together.
- A decision-shaped verdict. The deliberation is distilled into ranked objections, risk flags, a kill line and cheap tests. The raw argument is visible in the chamber; the verdict is what you take into Monday's meeting.
And because the free tier runs on free open-weight models, each council costs us a fraction of a penny in compute. That is what makes a genuinely free tier possible without a login wall or an email capture.
How it helps a UK small business in practice
The tool is built around the decisions SMEs actually face. A few examples of what a council will do with them:
- Hiring your first employee. The council will produce a kill line around how many new bookings must land to cover the wage, and a cheap test such as running a paid trial month with a subcontractor before committing to a salary.
- Raising your prices. The customer adviser will name the segments most likely to churn, the money adviser will model the margin gain, and the verdict will flag which customers to grandfather and which to move first.
- Opening a second location or adding a service line. The risk adviser will surface the fixed-cost exposure, the operations adviser will flag the staffing and process load, and the kill line will tell you the utilisation you need to break even.
None of this replaces judgement. It does something more useful for a busy owner: it makes the invisible objections visible before the money is spent, in about a minute, for free.
The honest bit
The free council argues from generic UK SME archetypes, not from your books. It is a pressure test and a devil's advocate, not a forecast. If you want the advisers to argue from your actual numbers, customers and market, that is the paid AI Council, trained on your data. The free tool exists so you can feel the difference a deliberation makes before you ever pay for one.
We would also rather you knew the limits up front: a council can miss domain specifics, and its confidence is not a probability. Treat the kill line as a tripwire to monitor, not a prophecy.
Try it
Put a real decision in front of the council at expedientai.co.uk/tools/ai-council. Three free councils per hour, no login. If it surfaces an objection you had not considered, it has already paid for itself.
