AI experimentation everywhere
Teams try tools on their own with no shared goal, so effort is spread thin and lessons stay with individuals.
Problem 3: AI strategy
AI Strategy for Founder-Led Companies
An AI strategy for a founder-led company decides where AI is used, who is accountable and how results are judged. Without one, a company can end up where everyone is talking about AI, but nobody knows where to use it.
On this page
- What an AI strategy for a founder-led company contains
- Six symptoms of AI without a strategy
- Why AI efforts stall without a strategy
- What a workable AI strategy looks like
- Choosing where to use AI first
- A first-pass audit of AI use
- Who should own AI in a founder-led company
- How I approach AI strategy for founder-led companies
- Frequently asked questions
What an AI strategy for a founder-led company contains
An AI strategy can start from four questions. Where will AI be used first? Who owns each use? Which rules apply? How will results be measured?
I call the problem “Everyone is talking about AI, but nobody knows where to use it.” Of the three problems I work on, it is the third. It shows up in companies that are interested in AI but haven't operationalized it.
Six symptoms of AI without a strategy
Read these six symptoms with your own company in mind. Under each one, the text describes what it can look like.
No AI strategy
Nobody has decided which business problems AI should address first or how success will be judged.
Employees using random tools
People pick their own tools, which produces uneven results and open questions about company data.
No governance
There are no rules on which tools are approved, what data goes in or who checks the output.
Lots of talk, little ROI
Activity is high, but nobody has measured whether it changes cost, speed or revenue.
Opportunities for automation haven't been identified
Repetitive work that could be automated has not been found, listed or ranked.
Why AI efforts stall without a strategy
AI efforts can stall when nobody owns the outcome. A pilot with no owner and no measure is easy to start and easy to abandon.
Pressure can add to the problem. Vendors, competitors and staff may all be talking about AI. A company can then feel it must act before it has decided what to act on. Activity can stand in for direction.
Scattered experiments can also teach the company little. Each team may try a different tool on a different task, and the results may not be comparable.
Weak foundations can make it worse. If the underlying process is unclear or the data is scattered, AI has little to work with. A tool may not fix a process that nobody understands.
Missing rules add risk. Without a view on which tools are approved and what data can go in, employees may decide for themselves.
What a workable AI strategy looks like
A workable answer moves from opportunities to pilots to rules and oversight. No strategy can promise a return in advance, so each step should produce something the company can measure.
-
Assess the opportunities
Review processes for repetitive, high-volume work where AI could help and results can be measured. The article on the AI readiness assessment explains how to do a first pass.
-
Map and rank them
A map of AI opportunities lists where AI could help and ranks the candidates. The company can then pick a few starting points.
-
Run pilots with an owner and a measure
Start small. Each AI pilot gets a named owner, a defined task and a measure of success. Set a review point at the end of each pilot. Decide then whether to stop, adjust or extend it.
-
Set governance
Governance sets which tools are approved, what data can go into them, how outputs are checked and who is accountable. The NIST AI Risk Management Framework, built around Govern, Map, Measure and Manage, is one public reference. The article on AI governance for growing companies covers a starter structure.
-
Oversee implementation
Someone with authority oversees implementation, builds the results into AI-enabled processes and reviews them in the regular operating rhythm. Results that work move into the standard process, so the gain doesn't depend on one enthusiast.
Choosing where to use AI first
These criteria are general. They help rank candidates, and each company's answers will differ.
| Question | A good first candidate | A poor first candidate |
|---|---|---|
| Task | Repetitive and high in volume | Rare, one-off or highly varied |
| Result | Easy to measure | Hard to define or judge |
| Data | Data the company already holds and may share with a tool | Sensitive data with no rules on its use |
| Owner | A named person is accountable | Nobody owns the outcome |
| Failure | A mistake is caught early and is cheap to fix | A mistake reaches a customer unseen |
A first-pass audit of AI use
A first pass can be done inside the company with a short set of steps. It shows what is already happening and where to start.
- List every AI tool people already use, including personal accounts used for work.
- List the repetitive tasks in each department.
- Note what data each task involves and how sensitive it is.
- Rank the tasks by volume, measurability and risk.
- Choose one candidate and name an owner.
Who should own AI in a founder-led company
One accountable leader should own AI results across the company. In a founder-led company, that role can fit whoever runs operations.
The owner sets priorities, approves tools and reviews results in the regular leadership meeting. Without an owner, AI tends to fall to whoever is most enthusiastic, and the result can be scattered experiments.
A company missing senior operational leadership has a gap here. That gap can be one reason the AI question and the operating question arrive together.
How I approach AI strategy for founder-led companies
I treat AI as one part of how the company operates, alongside people, process and technology. The 90-Day Scale-Up Operating System assesses AI opportunities in Phase 1. It creates an AI opportunity map in Phase 2, then helps execute AI pilots and automation in Phase 3.
The Fractional COO + AI Transformation engagement includes everything in the Fractional COO engagement. It runs at approximately 2 days per week and costs $10,000–$12,500 per month. It adds the following.
- AI opportunity assessment
- AI implementation
- Workflow automation
- AI-enabled processes
- Technology architecture
- Vendor selection
- AI governance
- Implementation oversight
Frequently asked questions
Where should a small company start with AI?
A small company should start with a business problem, not a tool. Look for repetitive, high-volume work where results can be measured, and run a small pilot with a named owner. Agree basic rules on tools and data before more people start using AI.
What is an AI opportunity assessment?
An AI opportunity assessment is a structured review of where AI could help a company's work. It can end with a ranked list of candidate uses and a view on what to try first. AI opportunity assessment is one of the areas in the Fractional COO + AI Transformation engagement.
Why do AI pilots stall?
AI pilots can stall when nobody owns the result and nothing is measured. They can also stall when the pilot is not connected to how the company runs. Scattered tools, unclear processes and missing rules add friction. A named owner, a measure and a place in the regular operating rhythm can keep a pilot moving.
What is AI governance for a small company?
AI governance for a small company is the set of rules and responsibilities that decide how the company uses AI. It can be light: approved tools, limits on data, a check on outputs and a named owner.
Do we need clean data before starting with AI?
A company can start with AI before its data is fully clean. A pilot needs data that is adequate for the task. Choose a first use where the data is already available and understood. Fix data problems as the pilot shows which ones matter.
How should a company handle employees who already use AI tools?
A company should handle employees who already use AI tools by finding out what is in use. It should then set clear rules. Ask what people use, what data goes in and what value they get. Approve the tools that meet the rules and retire the rest.
Keep reading
Start with where AI is used today
Describe how AI is being used in your company and where it is stuck, using the contact form.
Book a conversation