Most of the AI pitches landing in Kenyan businesses right now are pitches for something else with the letters A-I added. That's not snark — it's a genuine problem for the buyer, because the wrong partner can burn six months and seven figures before anyone notices the thing doesn't work.
Here's the short version of how to tell them apart.
1. Ask to see a production system, not a demo.
Demos are easy. Production systems are hard. If an agency can't send you a staging URL or a short video of the tool being used by a real human in a real job, assume it doesn't exist yet. A production system answers these questions:
- Who uses it every day?
- What was the metric before, and what is it now?
- What happens when the model is wrong?
2. Ask what they don't use AI for.
An AI-native team has a long list of problems they wouldn't solve with an LLM. Reconciliation. Simple lookup. Deterministic pricing. If the agency reaches for AI for every problem you describe, they're not experts — they're enthusiasts. Those are different things.
3. Check the handover plan on day one.
If your team can't run the system without them, you don't own the system.
The question to ask on the first call: "What's in the handover package?" An AI-native team has an answer — docs, SOPs, training videos, a shutdown plan if the vendor disappears. A hopeful team will tell you not to worry about it.
4. Follow the money.
Ask how they charge. Time and materials on an AI project with no fixed scope is a known pattern — it ends with a bill four times the estimate and a system that works "mostly." Fixed price against named milestones forces the agency to actually scope the work before starting. You want that friction.
5. Get a reference who's one year in.
Anyone can ship something that works the week after launch. The test is: does it still work a year later when the edge cases have stacked up and the data has drifted? Ask for a reference who started with them 12+ months ago.
If you'd like a second opinion on an AI proposal you've received — paid or not, we're happy to read it — send it to us at info@alitdigital.com. No pitch back, just honest feedback.