Should you have an AI strategy? Here's ours

Ok, AI is a thing. It might even eliminate capitalism. Should you therefore have a strategy? Burned by leaning into the future already by your late 2021 crypto investments? I have been...

We do not like the idea of a solution looking for problems, so have been hesitant to leap in fast.

However, AI is meaningfully solving problems today – Copilot, for example, is super useful. This is different to what we saw with many blockchain companies previously.

How did we create an AI strategy?

We started by considering how AI would impact us.

For a little context, our mission is to help engineers build better products. We do that in three ways:

1. Provide all the tools they need in one to evaluate feature success
2. Get in first
3. Be the source of truth for customer and product data

Tim and I first asked ourselves what our biggest threat was, regardless of AI.

A while ago, we considered it to be someone doing a cooler more developer focused job of our product and smashing a launch on Hacker News – i.e. something with dark mode, keyboard shortcuts, SQL, warehouse-y features, and better memes.

But now we've shipped those features, we now think we are that company. We have a Godzilla hedgehog on our homepage after all, and I've gotten into memes.

Today, we think our biggest threat is an AI-powered suite of tools for each of the products we provide. That'd take a long time to build and would be hard without any user data.

The good news? We're in a better spot than anyone else to do this.

So, what difference will AI make?

The above was the next question we asked ourselves.

We can see three thoughts to incorporate into how we work and what we build.

1. We should consider AI a valid product enhancement. If we want to provide all the tools in one, they need to match up individually to the major competitor in each category. We think there are some useful improvements it can make to our users, such as natural language to SQL is getting a lot better. We'll build AI features into our products that already have product-market fit.

2. AI changes how software is built and thus we need to change how product analytics, for example, works to accommodate this. However, this feels more like it's an accelerant than a complete 180, so we think no big change is needed here. AI native products do have some additional data to monitor (i.e model performance), but this isn't a huge deal as a feature to ship – our warehouse, BI and SQL combo of newer products could easily handle this when we see enough demand.

3. AI enables us to build much faster. This will happen organically, we'll adopt as necessary. It will improve our ability to execute our "all in one" strategy – we can build 300 products instead of 30. I think we'll manage to sustain our moat of being the widest as we already have the traction, architecture and existing audience to pull this off.