Ever so often, first time founders surprise us with their ambition. It’s only fair, artificial intelligence works well in pitches, so you can use it to get more attention from VCs. And with the advent of AI frameworks such as TensorFlow, it’s easier to use AI than it’s ever been. Here’s how to think about getting the execution right.
There are two important differences between artificial intelligence projects and “regular” tech ones:
- Most artificial intelligence projects you’d want to embark on are heavily front-loaded. A lot of thought and effort has to be put into the first iterations, and you’ll only get to see the result much later.
- Artificial intelligence projects mostly fall in the research & development category. You never really know how long the research part will take before you come up with anything to sell.
The difficulty of R&D projects is that it’s hard to estimate the return on investment. You might have an idea about the return side of the equation, but can’t guess how long it will take to get there. Or even if it’s possible at all.
Small differences in input data or ambition can make a huge difference too. Even the simplest and most generic recommendation engine can be derailed by a perfectly normal looking site’s perfectly clean data — if that particular niche or traffic is somehow less predictable than others.
To sum up, with AI:
- You have to spend a lot of cash upfront, and
- There’s a chance that all you’ll learn is: it can’t be done
Agile and AI
If you do embark no AI, my preferred approach is based around the agile methodology. Here you have an opportunity to change direction after every sprint. This acts as a safety check where the potential downside is capped, but creativity isn’t.
The point with using the agile approach is to limit the scope of work, and have a clear goal for each 1 or 2 week sprint. Each of these chunks is to be completed with a shipped product. Think of things like “generate similar tweets from this dataset” or “categorize users based on their characteristics”.
After every sprint, the know-how transferred from the freelancer to the company. And after every sprint, you will see whether you’re getting closer to something valuable. Designing the scope right will go a long way in meeting commercial targets. Or, in a worst case scenario, capping the investment to a minimum.
This project is likely not your last one. All upcoming artificial intelligence experiments should use the knowledge gained in previous ones.
With future proof the R&D efforts, the goal is to make it easy for any one developer to pick up the job:
- It’s a lot easier to write code than it is to read it, and developers don’t enjoy writing documentation. To help the next developer, the absolute minimum is a short and easy to read README. It has to explain how the system works, and how to use it.
- Open standards and popular frameworks help. These days most AI developers will have similar tools in their toolbox. Most likely, Python with a framework like TensorFlow or Scikit-Learn. Always make sure that devs use a known framework. Home grown, proprietary code is better avoided.
- You can use automated or manual tests, but you can’t outsource all testing to the same people who build the system. Make sure you test every project in-house. This takes you halfway in knowledge transfer, and you’ll be well positioned to hire new devs.
Keeping the creative momentum
Once a project finishes and results are in, you’ll will be eager to start a new one. This is a good time to encourage innovation and idea generation: what could be the next opportunity to explore?
Use this time for a discussion to draw conclusions and gather a large number of ideas to continue. Form those ideas into more concrete potential projects, even if you don’t start them all at once.
There’s a temptation to continue as long as it’s fun, but make sure it makes sense for the business to continue. Keep the control on your side: only approve proposals when you’re ready to translate them into projects and actions. Figure out whether you’ve passed over into a phase where the added value is not worth the investment anymore.