Buy, build, augment or keep: deciding what AI does
Four choices for each step of your work. We look at when each one fits, using examples from our own projects to show what works in practice.
Clinton Onyekwere5 min read
Buying AI tools does not mean your team will use them. The BCG AI at Work 2025 survey found that more than three-quarters of leaders and managers use generative AI several times a week. Meanwhile, regular use among frontline employees has stalled at 51 per cent.
Look at what to do with each step of the work. Then you can decide whether to buy, augment, build, or keep that task entirely with your people.
Buy a tool
From what we've seen, buying a ready-made tool is usually the quickest and cheapest route. It makes sense when a task is common and good tools already do it well. The main thing is safety. Always check how a tool handles your data before anyone uses it.
Augment your tools
We put AI inside the tools you already use, so your team doesn't have to learn a new platform. The AI writes the first draft, and a person approves it.
Take a start-up preparing for launch that needed to handle more leads without hiring. We built a system that captures every lead and routes it down one of five paths. High-fit leads get a draft email ready for review. Inbound and outbound share one pipeline, keeping a person in control.
Build something new
It makes sense to build something new when your work is specific to your business and the volume is high. We usually suggest this when ready-made tools just don't fit what you need.
Take a maritime business we helped. Writing ship part requests took up to 30 minutes by hand. Our system formats five enquiries in under 90 seconds, passing an AI check and 18 code checks before human review. We also built a pipeline producing over 2,400 illustrations for a language platform. The client's notes always override the system, keeping people in charge.
Keep it with people
Some steps need human judgement, relationships, or accountability. In our experience, it's best to keep work with your team when a mistake costs more than the time saved, or when clients pay for your expert view. Choosing to keep a step with people is a deliberate decision.
Why decide step by step
McKinsey's 2025 survey found redesigning workflows has the biggest effect on whether you see an earnings impact from generative AI. And Microsoft's 2026 index showed organisational factors have more than twice the reported impact of individual ones, 67% against 32%. From what we've seen, they both point the same way. The work and the organisation matter more than the tool.
Our AI Readiness Audit takes one to two weeks. We rate every step of your work as one of four decisions. You get three ranked recommendations and a costed first move. After that, you can build with us, your team, or someone else.
Sources
Get our newest articles
We only email you when a new piece goes live.