BYOAI

Since the dawn of this current Age of AI, there has been an assumption that at the highest level, there are two markets – the markets as old as time, or at least as old as tech: consumer and enterprise. Startups, when they're born, tend to pick one lane. And if they grow into large companies, they tend to stick in that lane.¹

AI, to date has been playing out similarly. While OpenAI may not have set out to be a consumer company, ChatGPT shoved them into that bucket and market. Anthropic, perhaps in part because ChatGPT became the "Kleenex" of consumer AI, went mostly down the enterprise path.

Now, with Anthropic seeing massive success on their path thanks to the rise of "vibe coding" branching into the first truly agentic workflows, OpenAI is scrambling to pivot-to-enterprise. They undoubtedly wouldn't frame it that way, and it is slightly unfair, but it's also not entirely untrue. It's why they keep touting how the enterprise business should match the consumer business this year. In this way, coding may be to 'Big AI' what the cloud was to the last generation of Big Tech. That is, their inroads to enterprise.²

Still, I'm somewhat skeptical of the strategy because OpenAI is aiming to shove Codex inside of ChatGPT itself. Yes, this has worked for Anthropic with Claude and Claude Code (and now Cowork) residing in the same desktop app, but that's also because Claude doesn't have nearly the consumer business that ChatGPT has. And I'm just worried ChatGPT, after spending the past couple of years trying to simplify their product, is about to complicate things considerably.

That said, they sort of have to try? Anthropic seemingly has such momentum that they only obvious lever OpenAI has to pull in order to jump into the race is to leverage that ChatGPT installed base. To leverage their Kleenex position, as it were. The hope would be that we're early enough in the agentic revolution that ChatGPT and not Claude – and certainly not OpenClaw – can be the one to introduce the masses to it. You can see the logic, but there's a ton of execution risk.

Thinking through this has led me down another path that's tangential, but related: what if AI plays out similarly to the smartphone? That is, whereas everyone used to have their work computers and home computers, the iPhone changed this dynamic. Because the smartphone took over for many people as their main computer, and most people didn't want to carry around two smartphones, companies had to start adopting 'BYOD' – bring your own devices – strategies. There have obviously been trade-offs – namely in the form of security and compliance – but there was no fighting the convenience tide here. Even if they had a work phone, everyone was doing everything on their main device – see: any number of headlines about any number of politicians over the years.

Undoubtedly thanks to the inherent cost savings as well, this movement has since spread to computers/tablets and to schools and other walks of life.

Anyway, what I'm wondering now is if this dynamic plays out in AI too. As we all increasingly have the one AI we use the most, and that builds up a moat in the form of memory, might we start insisting on using that AI in the workplace? Yes, 'BYOAI'.

Once again, security will be the main, obvious issue here. And a subset of that is privacy – which has also been the case with BYOD, of course. But might convenience win the day again? If, say, you have the ways you like to work with your AI and the workflows established in that memory, wouldn't you want to bring that to work as well?

Some people undoubtedly would say "no", that they want to separate work from home in that regard. Maybe it's more similar to email in that way. But even there, the lines have clearly blurred over time. (Again, see: politicians.) So it probably just depends on if workplaces end up implementing more rigid harnesses on top of the AI models they choose. Which is to say, they don't just choose the chatbots out-of-the-box – and soon the "superapps" from OpenAI and the like.

And sure, for large enough companies, they'll obviously have tailored AI solutions. Certainly in certain more highly regulated industries. But if you really believe AI is going to permeate everything – much like the smartphone has – doesn't it stand to reason that a small mom-and-pop store in Ohio is going to simply go with the AI brand they use at home? Especially if they're already paying for it there.

Obviously, I don't know how this will play out, but my instinct is that for many businesses, and certainly smaller ones, there will be this 'BYOAI' policy. Perhaps the AI tools themselves even implement a "work" mode to complement the "incognito" mode that they all borrowed from web browsers. Many web browsers now, of course, also offer multiple profiles to split up work and home.

With all that in mind, of course OpenAI needs to ensure they snuff out the Anthropic threat in enterprise. Because it also stands to reason that a lot of people will get their start using AI at work, especially if there's any level of coding and eventually agentic needs there. And if that's the case, it could almost be the opposite of the BYOD movement – it could flow the other way, from work to home. And that could seriously imperil ChatGPT's position...

¹ Yes, much of Big Tech blurs such lines, but that's out of necessity: these companies are so large that in order to grow, they need to go after any and all customers. Still, they're usually bucketed into one of those markets: Apple = consumer, Microsoft = enterprise, Meta = consumer, etc. ↩

² Google and Amazon are perhaps the most diversified Big Tech companies thanks to the rise of AWS and GCP after they built massive consumer businesses. ↩