Notes on AI for firms and teams
Practical guides on AI adoption, data security and training. No hype.
AI in agencies: produce faster without losing the voice
Generative AI speeds up the mechanical part of creative work — but speeding up without judgment is the fastest way to sound like everyone else or to publish a mistake with a client's logo on it. The difference isn't the tool. It's the flow.
Read →
AI in the architecture studio: where it accelerates and what not to delegate
AI in architecture is not one thing. It hugely accelerates conceptual iteration and writing, and it is genuinely dangerous in structural judgment, code compliance, and professional responsibility. The skill is knowing which is which before you deliver.
Read →
AI in Mexican accounting: where it helps and where it is risk
Not all accounting is the same problem. Classifying invoices is one thing, preparing a tax filing is another, and confusing the two is the most expensive mistake I see in firms. Here is where AI pays off and where it gets you into trouble.
Read →
AI and personal-data protection in Mexico: what your firm must watch
Your clients' data does not lose its protection because you run it through an AI model. I go through what the law requires, how public and private tools differ, and leave a checklist for your firm.
Read →
What an AI training syllabus that actually works looks like
Most syllabuses open with theory and leave nothing behind. Mine opens with the good part — the person's first real workflow — because that order is the difference between a course people forget and one that changes how your firm works.
Read →
What does running AI on your own servers really cost?
When a partner asks me the price of owning AI outright, I usually disappoint them, because the honest answer is not a number, it's a list of costs that never appear on the graphics-card quote. Here is that list, without the gloss.
Read →
AI for finance and accounting: where it helps and where it's a risk
In a finance or accounting practice, the line between help and harm is thin. Here is an honest map of the tasks AI can carry, the ones it must never own, and the operating rule that keeps a firm safe.
Read →
How to write an AI use policy your compliance team will approve
A usable AI policy is not a one-page disclaimer. It is a small set of decisions your people can follow on a busy day and your compliance team can defend to a client or a regulator. Here is what those decisions are.
Read →
Why most corporate AI rollouts fail (and what to do instead)
You bought the licenses, sent the memo, and three months later almost no one uses them. The problem is rarely the tool. It's how it entered the firm.
Read →
Whole-wafer chips: what a chip the size of a plate means for you
A partner sent me a headline about a chip the size of a dinner plate that supposedly runs the world's fastest AI, and asked whether it should matter to him. The short answer is no, not to buy. The longer one explains why that slab of silicon says something useful about where your data should run.
Read →
How to train your team to use AI without a 6-month rollout
You don't need a six-month transformation program or a big platform purchase. What changes behavior is hands-on practice on the work your team already repeats, plus enough reinforcement that the habit holds.
Read →
What an AI training engagement looks like (and what it costs)
A plain account of how we run a hands-on AI training engagement for a professional-services firm: the on-site months, the remote reinforcement that follows, what the firm walks away with, and why the price is quoted to fit rather than listed per seat.
Read →
Is it safe to put client documents into ChatGPT?
It depends on which tier you use and what the document is. The honest answer separates the consumer version, which is not safe by default, from the enterprise and API tiers, which can be, and gives you a rule for telling them apart.
Read →Bring this to your firm?
We train your team on their own real work, on-site, with three months of remote reinforcement.
Start a conversation