The one area to start with, the shortcut to a first draft and what makes it easy to keep going. AI becomes useful when it has all your business context (your value proposition, your language, how you structure things), but it needn’t be a daunting task to teach AI who you are.

The trick is to start far smaller.

Alexandra Buys is co-founder and COO of The Delta, a venture ecosystem that started in Cape Town and now runs an 80-person operation across a Berlin startup campus. She led the rollout of Claude across the entire business and built the context layer that powers how the whole team works with AI. She’s clear on where to start.

Don’t try to capture everything. Pick one area, build the context for just that, connect it to your AI, and feel the difference in output. That first win is what makes the rest worth doing.

“Don’t start with everything. Start with one area (sales is a good first one) and build context for just that.”

Sales is Alex’s suggested starting point; it’s high-leverage and the context is relatively stable. But the right first area is wherever you most often ask AI for something and get back something too generic to use.

One area. Not the whole company. The whole point is a scope small enough that you actually finish it this week.

The blank page is what kills these projects. The fastest way past it is to talk, not type.

Alex’s shortcut: have someone interview whoever owns that area for an hour (how do we do this, what happens when, what are the biggest struggles) and record it. Then hand the transcript to Claude and ask it to pull out the context worth storing. It does most of the first draft for you, and the team builds on it from there.

Don’t point the AI at your existing wiki. Those docs were written for people; full of fluff, context the AI doesn’t need, and so long it burns tokens hunting through them for the few lines that matter.

“Structuring things for AI is far more efficient.”

Write tight. Stick to the key points, the language you use and how you want things done. Keep it short and purpose-built beats long and human-readable every time.

Connect that one area’s context to your AI and run a real request through it, the kind that used to come back generic. The jump in output quality is the proof, and it’s what earns the project its next hour of your time.

Only once that one area is working do you add the next. Area by area, the context layer grows,  and because each piece already proved its worth, you never end up with the abandoned half-finished folder.

You go from AI that gives everyone generic answers to AI that answers like it actually works at your company: starting with one area, in a week, instead of a documentation project that never ships.

It takes an hour-long interview and an afternoon to get the first area live. The output difference shows up the very first time you use it.

This is one piece of Building Your Business’s AI Context Layer. Alex’s full masterclass inside the Founder Collab shows you the complete context-layer architecture that The Delta runs on:

You’ll also get access to 40+ other masterclasses from SA founders and operators on sales, fundraising, UX, paid media and more inside The Founder Collab.

Get more SA tech and business news, tips and business-building workflows and subscribe to The Open Letter.