Skip to main content

The rise of artificial intelligence has had a profound impact on product development and the general zeitgeist of the internet economy lately. The hype train is in full effect, but its a bit different than the crypto or NFT hype cycles we recently experienced.

The big difference is we can now see relevant use cases that aren’t some abstract polymath function none of us really understands. The principal argument of blockchain was that it enabled trust, but most of us already trusted the tools we use every day. The barrier to trust we place on our tools isn’t that high that we need to due diligence everything we use. If we did, no startup would ever pass the sniff test and get any amount of traction.

What changes now though with these new AI tools that have recently hit the market is that we can now ingest sentiment from the user in so many non-traditional formats. We’re no longer restricted to purely keyboard or mouse or tap interfaces. We can have someone share an image of what they are thinking about or what they want, we can have them talk to the software to describe what they want. These multi-modal commands make it all the more dynamic and flexible and change the game in how we think about software development. Gone are the days when you’re explicitly limited to whatever data a human or a robot can feed you.

The emerging era of the AI copilot.

In what may be the most exciting and rapid paced arms race in the tech world, everyone and their mother seems to be racing to add a copilot function to their apps. Whether its Github, or Microsoft (who owns github), or Google, or Adobe, or any number of other well known software companies adding a copilot function, the trend is clear: people are going to expect in-app help.

So how does this affect how you design and develop your products?

First let’s categorize how we can expect to see this generation of AI tools working with your product. There are going to be the helpers that give people better tutorials, answer questions, and show people how to use the tools better. (These will end up being trained off of extensive and more readable documentation and knowledge bases). This can end up being a cost savings from a tech support side over the long run. Watch this space for an emerging crop of human and hybrid powered solutions to this in the form of next generation zen desk type tools or intercom.

AI generated robot image

Robot assistants made by AI

The other big bucket we’re going to see is generative tools that take varying inputs from a user or different input types (asking the user questions vs expecting them to learn your UX) and then helping them accomplish the end goal of whatever your product does. Whether its filling in a spreadsheet, drawing a poster, or explaining what the contents of a document are. None of the off the shelf generative libraries really solve for this entirely, and you’re seeing a rush of people adding image generators and text expanders into their tools, but that’s not really going to make a huge leap like the solutions that do things like “take my meeting notes and generate a project template and notify the key people” type of prompts. These prompts most likely will be both typed and spoken.

So how do you as a product manager/founder adapt?

Lean into these tools, they already exist, ignoring them would be like doubling down on horse and buggy accessories when the automobile is already here. The next few years will give you enough cover to come up with novel interfaces as moats, or force you to get creative on defensibility.

Exciting times.

Note the featured image was generated with AI.

Leave a Reply