Key takeaways
- Most of AI transformation in a smaller business is internal work that depends on knowing the business.
- Outside help pays for itself in three places: initial setup and governance, regulated-sector policy, and custom builds.
- Buy defined pieces at fixed prices, not open-ended programmes.
- The single biggest success factor is an internal owner with protected time.
A lot of businesses assume AI transformation has to be bought, because the phrase sounds like something a large consultancy does. In a business of 20 to 200 people, most of the work is better done by your own people, because it depends on knowing how the business actually runs. The trick is to buy help only for the parts where outside experience saves real time or risk.
What you can do yourselves
Choosing workflows to change
Nobody knows better than your staff which tasks are repetitive, which documents get rewritten every week, and where the bottlenecks are. Ask each team lead for the three tasks they would most like to stop doing by hand. That list beats any external assessment.
Redesigning those workflows
Once people have a good tool and basic training, redesigning a simple workflow is a few hours with the person who does it: write down the steps, try the AI-assisted version, refine the instructions, measure. See the AI transformation roadmap.
Training on real work
Generic AI training is widely available and mostly forgettable. The training that works is a colleague showing three tasks from their own week. You can run this internally once a few people are confident. See AI training for employees.
Usage policy
A one-page policy is something you can write using a good template. See AI usage policy for teams.
Where outside help pays for itself
The initial setup and governance
Choosing the right tier, configuring the admin console, retention and identity, setting spend caps at the right levels, and limiting connectors is work you do once. Someone who has done it many times will do it in a fraction of the time and avoid the mistakes that come from learning on your own tenant. This is a defined, fixed-price piece of work.
Regulated-sector policy
If you are in law, health, financial advice or another regulated field, the policy needs to reflect your professional obligations. Getting this right once is worth paying for.
Custom builds
If you need software built on a model API, such as an agent that reads your practice management system and drafts documents, that is engineering. Hire people who do it.
How to keep any engagement small
- Buy outcomes, not time. "Configured and live in 30 days" rather than "40 hours of consulting".
- Skip the assessment phase if you know what you want. Many businesses already know their tool and their first workflows.
- Insist on handover. Admin access, configuration notes, policies and training materials should all be yours.
- Do stage 2 yourselves. Once the foundation is in, workflow redesign is mostly internal.
The internal owner
The single biggest predictor of success is not the tool or the consultant. It is whether one respected, organised person has protected time to own the work. In a business of 30, that might be three hours a week. In a business of 150, it might be a day a week. Without that person, no amount of outside help will stick.
A realistic split: buy the foundation as a fixed-price package, then run the workflow programme in-house with your owner. That is the model our Implementation Package is designed for.
For a comparison of the kinds of firms you might buy that foundation from, see who should implement AI in your business.
Frequently asked questions
Can a small business do AI transformation without consultants?
Yes, much of it. Choosing a tool, writing a usage policy, training staff on their own work and redesigning simple workflows can all be led internally. Outside help is most useful for the initial configuration and governance setup, regulated-sector policy, and custom builds.
What parts of AI implementation should we outsource?
Outsource work that is done once and is easy to get wrong: provider configuration, admin and security settings, spend caps, and identity integration. Keep in-house work that depends on knowing your business: choosing workflows and redesigning them.
How do we keep AI consulting costs down?
Buy a defined, fixed-price piece of work rather than open-ended time, insist on handover of everything built, and do the workflow redesign yourselves once the foundation is in place.
What skills does an internal AI owner need?
Organisation, credibility with staff, and curiosity. Technical depth helps but is not essential. They need to be able to run a pilot, measure a task, write a clear page of instructions, and say no to scope creep.
Get a straight answer for your business
The Clarity Package gives you a written business case and a detailed implementation plan for 500 USD. The Implementation Package does the whole build, live in 30 days, for a fixed 5,000 USD.