AI for Small Business: What a Company OS Looks Like at 10 People, and at 200

At 10 people, the founder is usually the company brain, so AI setup is about getting what they know out of their head. By 50, several people hold the knowledge and their AI starts to disagree. With 200, the business needs owners, rules and shared workflows. The architecture is the same at every size; only the weight changes.

Does a 10-person business need a Company OS?

It needs a small one. A 10-person business rarely needs formal infrastructure, but it benefits from the same idea: a few shared files describing the business that everyone's AI reads, and one or two written workflows. Without them, the founder's knowledge stays in the founder's head and everyone else's AI starts from nothing.

Size changes the problem, not the principle. In a small team, inconsistency is easy to spot and fix over coffee. Even so, the moment a second or third person starts producing customer-facing work with AI, the business needs one version of the facts.

Adoption is still uneven by size. According to the Australian Bureau of Statistics, about 12% of Australian businesses used AI in 2024-25. That rose to about 22% of medium businesses and 35% of large ones, against about 11% of small and micro businesses combined. The National AI Centre's broader tracker, which counts any level of adoption, put SMEs at 43% across December 2025 to February 2026.

What does a Company OS look like at 10 people?

At 10 people it is light: three or four shared files, such as product facts, the ideal customer and the brand voice, plus one written workflow for a job done every week. The founder is usually the main source and the main approver. Everyone uses the same business AI account.

A sensible 10-person setup:

• Shared files: product facts with approved claims, the ideal customer, common objections and the voice.
• A single workflow: the weekly job that wastes most time, written down step by step.
• A business assistant: one plan for everyone, not personal accounts.
• One approver: usually the founder, for anything that goes out under the company's name.

The biggest gain at this size is getting the founder's knowledge written down. Founders answer the same questions repeatedly. Writing those answers once, where everyone's AI can read them, frees the founder and lifts everyone else's work. The guide to AI for founders covers this in detail.

What does a Company OS look like at 50 people?

At 50 people, knowledge is spread across several leaders and teams, and inconsistency becomes expensive. The setup needs owners for each set of facts, a handful of written workflows across sales, marketing and operations, regular updates from calls and tickets, and clear rules about data and approval.

This is often where the problem bites hardest. There are enough people for ten versions of the business to exist. Meanwhile, there is usually no AI lead, and each team has adopted its own tools. According to Microsoft and LinkedIn's 2024 Work Trend Index, 78% of people who use AI at work bring their own tools, so every new hire adds another private version.

At this size the main additions are ownership and sweeps. Each set of facts gets an owner who settles conflicts. New calls, tickets and emails are processed regularly, so the shared knowledge keeps up with a business that changes weekly.

What does a Company OS look like at 200 people?

At 200 people it becomes infrastructure. The company brain has a full set of canonical files with owners, many workflows run across teams, agents connect to core systems with carefully scoped access, and there are written rules, training and regular reviews. Governance matters as much as content at this size.

Three things change most. Access becomes a serious design question, because agents connected to the CRM or finance systems need permissions set one tool and one action at a time. Training becomes a programme, not a conversation. Finally, ownership of the whole system needs a named person, usually in operations.

Ownership of the system itself matters more too. A 200-person business cannot have its institutional knowledge sitting with a supplier or in a few employees' personal accounts. It needs the system in its own accounts.

What stays the same at every size?

Four things stay constant: one shared source of business knowledge, written methods for frequent jobs, a named person approving anything published, sent or spent, and company work in company accounts. Size changes how much of each you need, never whether you need it.

The comparison of Company OS vs no OS shows what this changes in a working week.

Skein builds a Company OS sized to the business, typically for teams of 10 to 200 people. A team of two or three, with one well-organised person, may need nothing more than a shared folder and a good instructions file.

Frequently asked questions

Is AI worth it for a small business?

Usually, if it is pointed at the right work. Start with one weekly job that wastes time, give the AI your business facts, and measure the result. ABS data shows about 11% of Australian small and micro businesses used AI in 2024-25, so many are still at the start.

How should a small business set up AI?

Keep it light. Put everyone on one business AI account, write three or four shared files describing the business, document one frequent workflow and name an approver for anything that goes out. Add more files, workflows and agents only as the team and the work grow.

When does a business need formal AI governance?

Usually somewhere between 50 and 200 people, when several teams use AI, agents connect to core systems and knowledge is spread across many leaders. Before that, a one-page policy, an approver and shared files are enough. After it, owners, access rules and regular reviews become necessary.

Can a Company OS grow with a business?

Yes. The architecture is the same at every size: shared knowledge, written workflows and agents with approval points. A small business starts with a few files and one workflow, then adds owners, sweeps, agents and access rules as it grows, without rebuilding.

What is the biggest AI mistake small businesses make?

Letting everyone use personal AI accounts with no shared knowledge, so each person's AI holds a different version of the business. The fix is cheap at small scale: one business account, a few shared files and one written workflow, before the inconsistency spreads.

Related reading