Case study · 11 July 2026

How I run a business on AI agents

By Heinoux Roux, founder of NexBDM™, a South African AI automation partner.

NexBDM runs on AI agents every day: the inbox, the marketing engine, the client documents, and a 2,879-node knowledge graph of the whole business that updates itself. One operator, one agent system, no theory. This is the exact setup, and what I would tell you to copy first.

The whole stack in one paragraph

The agent is Claude Code, but the system is plain files. One root folder holds everything the agent needs: reusable methods (skills), persistent memories, and one intent-keyed index that maps all of it. Scheduled agents run the recurring work. Nothing lives in a chat window, because chats evaporate and files compound.

The four layers

1. 1. The operating system

   A single organised home. Methods in one folder, memories in another, projects in a third, one index at the top. The agent starts every session from the map, not from zero.

2. 2. The memory system

   One fact, one file: a name, a one-line description, a type, and the fact itself with its why. The agent writes memories when it learns something and recalls them when relevant. Decisions, standing rules, client preferences, and project state all survive across sessions. This layer is what the knowledge graph grows from: memories become nodes, their links become edges, and a nightly job keeps the graph current. It sits at 2,879 nodes today.

3. 3. Skills

   Any process I have explained twice gets written down once: when to use it, the steps, the rules, what done looks like. Proposal formatting, brand application, publishing pipelines, follow-up drafting. The method that rebranded 200 live documents in one sweep is a skill file.

4. 4. The toolkit index

   The single most useful file in the system: everything above, keyed by intent. You search by what you are trying to do, so the index is organised that way too. Hard rules live at the top where they cannot be missed.

A normal day

• ·Every two hours, an inbox agent triages email: it drafts replies in my voice, files what needs filing, and archives what is done. It never sends. I review drafts and hit send myself.
• ·Daily, a marketing engine publishes to an approved content calendar and prepares social posts. The calendar is approved weekly by me; the engine executes it with scripted quality checks that block anything off-brand.
• ·Nightly, the knowledge graph rebuilds from the memory store, so the map of the business is never stale.
• ·On demand, project work: proposals on the letterhead template, client platforms, brand systems. The agent applies the skills; I make the calls.

The guardrails that make it safe

Three rules carry all the safety. First: draft, never send. Anything a human will see gets a human yes before it moves. Second: hard rules live in one file the agent loads every session, brand rules, legal rules, lines never to cross, and scripted checks verify them before anything ships. Third: never let the agent invent a number. Every figure must trace to a real source, and arithmetic is verified by script, not by trust.

What to copy first

1. The folder structure and index. Half a day. This alone changes how every session starts.
2. The memory format. Start writing one-fact files the first time the agent learns a preference or a decision. Within a month it feels indispensable.
3. Your first two skills. Pick the two processes you explain most often. Write them down once, in the skill format, and stop re-explaining.

Frequently asked questions