AI and Entrepreneurship: The Tool Does Not Build the Company

A field experiment involving 758 consultants at Boston Consulting Group, carried out by researchers from Harvard and elsewhere, found that those who used artificial intelligence completed 12.2% more tasks on average, and that the quality of their work was more than 40% higher than that of the control group. They were then given a task deliberately designed to fall outside the model's capability, and those using the tool were about 19 percentage points less likely to reach the correct answer (Dell'Acqua et al., 2023). The same tool helped in one place and misled in another, with the same confidence.

What has become cheap, and what has grown dear?

What artificial intelligence gives a young company is cheap trial: drafting a text, building a prototype, analysing hundreds of replies in hours. Two studies suggest the gain leans towards the less experienced: in the experiment above, lower performers improved by about 43% and higher performers by about 17%; and in a study of 5,179 customer-support agents, productivity rose by about 14% on average and by about 34% among the least experienced, with little change for the experts (Brynjolfsson et al., 2023).

Two things follow. The first is that the tool brings the beginner closer to the seasoned practitioner. The second, which is our own inference, is that what is available to everyone cannot be a basis for distinction: if your competitor can use the same tool tomorrow, the difference lies elsewhere, in how well you know your customer, in the soundness of your judgement, and in the strength of the relationship you have built.

Start from the need, not the tool

In July 2025 the MIT NANDA project published a report concluding that about 95% of organisations had achieved no measurable return from their investments in generative artificial intelligence, and that only about 5% of integrated initiatives had produced a tangible effect on profits. The report attributes this to a "learning gap": most systems do not retain feedback, adapt to context or improve over time. The figure should be read with caution; its authors describe it as directional, based on self-reported interviews (52 organisations), a survey of 153 senior leaders and a review of more than 300 publicly announced initiatives, and it measures organisational initiatives rather than start-ups specifically.

Yet the lesson agrees with old advice from Paul Graham in his essay on starting companies: that the founder should make what customers actually want. Artificial intelligence does not cancel this advice; it raises the price of neglecting it, because it lets you build the wrong product quickly and cheaply, so you imagine you are advancing while you drift away from the market. The better question is not "How do I use artificial intelligence?" but "Which real need can I honestly meet?".

Practical habits for a founder on a limited budget

• Treat the tool as an assistant, not a manager: review what it produces, know where it errs, and do not delegate anything that touches your reputation or your customers' trust.
• Make learning your advantage: while the cost of an experiment is low, test more hypotheses and measure the market's response before you commit.
• Guard what cannot be copied: customer trust, data generated by use of your product with their permission, and an understanding of a local context that a general model may not handle well. We discussed the language and context of models in the article "Arabic Artificial Intelligence: Do We Make or Consume? Sovereignty and Arabic Language Models".
• Be honest with data: what you are entrusted with from people's data must not be treated lightly (see "Principles of Responsible AI").

In Saudi Arabia

The Monsha'at authority reports that small and medium enterprises numbered about 1.3 million at the end of 2023, and it aims to raise their contribution to GDP to 35% by the end of the decade (Arab News, 2024). That expansion leaves room for anyone who understands the local market and serves it in its own language and context, and the availability of a tool is not enough for that. The opportunity belongs to those who start from a real problem in their community and serve it with excellence.

This is not only a race for resources, but a matter of sincere intention and sound management. Whoever takes the necessary precautions, relies on God and listens well to people will find their tools a help to their wisdom, not a substitute for it.