AI: Build vs. Buy
When you are building the AI ecosystem of your enterprise, the build vs. buy fork needs deep consideration.
Your AI strategy will be driven more by internal factors than by what’s in the market. This is more true for AI solutions than tried-and-tested solutions for standard enterprise use cases for two reasons (1) Irrespective of what it says on the brochure, out of the box AI solutions don’t yet have a foothold in large enterprises and they are still evolving. (2) In most enterprises, employees are still figuring out how to effectively use AI.
Instead of ‘Build vs. Buy’, I proposed the ‘Create vs. Co-create’ approach in another post here.
More thoughts on this..
Budget
Depending on licensing model, AI costs escalate faster with growing adoption. With fixed cost licenses, you know the bill. You can opt for gradual roll out, focusing on teams and people best equipped to make the most of what the AI features offer. Flexible pricing models are a major risk. AI cost can escalate rapidly at enterprise scale when the reference dataset compounds and the token costs look like a nuclear chain reaction.. Estimate the total cost of ownership rather than simply multiplying the cost derived from a PoC.
Data is king
Most large, tech driven enterprises run on massive amounts of data. Data pipelines, approach to data access, synchrony between work roles and data access affect the value AI solutions can bring in. The demo and reality of traditional systems look very similar. Results and user experience for AI tools can be vastly different from what is shown. The difference is data.
Soft corners
A good AI system allows for customization and training. A bad one tries to build a walled garden and force users to adapt to it and buy more to make things work. Any decent internal tech team can tune AI solutions to work according to the culture, workflows and habits of the internal users PROVIDED the system has an open interface to work with.
Rushing through this kind of evaluation can cause frustration and insane amounts of time and money.