Build vs Buy

Build vs Buy

One of the biggest questions for enterprises right now is to build or to buy. Sadly, there is not one best way to do it. This short article is based on what I've seen work and not work; it favours the option to buy - albeit, I think the term buy is misleading about the real value of working this way.

Big enterprises, north of $1bn turnover, have the resources to build a new AI tool from scratch. A team of engineers, coding agents, API keys and a database - bosh. You have a new tool. Maybe, you even spin up a custom ChatGPT or Claude project with your own skills and that will suffice.

This can be quicker, cheaper and more customised on the surface. But, my opinion here is while this can work for simple and low priority areas - internally developed tools fundamentally lack the design, unwavering commitment and eclectic stakeholders that externally built specialised tools have. Although for more simple systems, this may not actually matter.

The real advantage of buying in AI is not the product itself, or whether it is built internally or externally. The real advantage is the collaboration and partnership. AI and frontier technology is not just moving incredibly fast but it's also evolving in new forms every quarter. The firms that want to do the best by embedding technology and innovation into their business practices won't fly ahead because of a new magic tool; it's because of who they're working with. When buying from an external specialist firm you're working with a team on the ground capturing the evolutions of technology, running at its development speed and working with your specialist teams to ensure they get the results they need.

The first version will not be great

It is more than likely the first version of your internally built tool will not be great.

Personally I've spoken with:

• $16 billion infrastructure consulting firm - said their internal AI tool is not great
• $4 billion consulting firm - said their internal AI tool is not great
• $30 billion soft service firm - said their internal AI tool is not great
• $80 million public transport firm - said their internal AI projects are not working
• $14 billion infrastructure consulting firm - said their internal AI tool is terrible

Legal firms are the one sector that gets it

• 80% of America's 100 largest law firms bought a specialist AI tool.
• 40/40 of US firms with 500+ attorneys have bought a specialist AI tool.
• Ropes & Gray, one of America's leading law firms, said:

Complex problems are difficult to solve

Complex problems are difficult to solve, which is the crux of why I side on the buying route. If you can find the right specialist team to invest in, they become your internal development team specialised in the vertical application of technology to the way you work. This team isn't just there to build but they're there to build, research, innovate, analyse your data and deliver the transformation - which is a lot more than an internal development project can offer.

The strongest distinction is between owning a tool and having a partner who continuously evolves how you use technology. That's a much more compelling commercial argument.

This doesn't mean to say all teams are great. That's a business decision to understand who you believe in and back to get you to where you want to be. It also doesn't mean built AI and bought AI can't co-exist; they really should. General AI and Vertical AI working together is the optimal position in today's climate. My final take is that the most dangerous position is one that imposes a singular hard rule and way of working. To thrive you need to be open, consider your options and work with people you can trust to deliver for you.