The promise behind the book
Trusted AI starts with a business that knows itself.
The book's method, the Trusted Living Agent Framework, sets out how to build trust around your goals from the inside out, and that trust is what keeps a business durable as AI changes.
The method works from the inside out because a perimeter alone cannot keep up once the digital workforce is already inside the building, reading the data, taking actions and using the tools. The perimeter still says what must be true. Something inside the arrangement has to make it true at each decision and be able to show that it did.
If your main AI vendor disappeared tomorrow, how many days until you were running again, and who would decide?
A business that trusts its AI can answer ten questions about it:
- What business outcome does it serve, and what result would be unacceptable?
- What has it changed, at what full cost, and what can honestly be credited to it?
- Which AI is in scope, and what could we have missed?
- Is the information it uses current, correctly understood and permitted for this use?
- Who owns it, and what is it allowed to cause to happen?
- Can the responsible person see trouble coming and stop it in practice?
- Which limits are enforced by controls, rather than by the model's promise to behave?
- Can it recover, and does it stay inside its access and dependencies?
- What supports each claim it makes, and what can we not observe?
- If a provider, a model or a person changed tomorrow, what would the business keep?
These ten are the test, and the book is the field guide to answering them.
The argument starts with the problem, in the founding essay, and it rests on one map, which is drawn at Trust League. The essays that carry the rest of it are here on the book's site.
Everyone knows about the bears, the big, visible AI disasters that make the news. At work, somebody's whole job is watching for them. At home, you assume that whoever built the tool has it covered.
While everybody watched for bears, the mosquitoes moved in. The agent that drafts your contracts, reconciles the ledger or runs the family calendar acts on its own, under your name, with nobody checking its work. The agents are small, quiet and everywhere, and most people do not feel the bites yet.
Meanwhile the work itself changed. The day you asked an AI to draft, reconcile or build for you, you became a builder, and so did everyone you work with.
Bears & Mosquitoes is the field guide for everyone who just became a builder and wants to build AI they can trust.
Building Trusted AI from the Inside Out
It is a field guide in the old sense, the kind you keep within reach because it names the things around you. It is written in plain language for the person who just became a builder, not for engineers.
A circle of honest readers reviewed it before it went to press. It is available now on Amazon in Kindle, paperback and hardcover.
It is written for all of us, and if you know a college graduate stepping into this economy, it is for them too.
Read the back cover and two excerpts from the Introduction →
Get it on Amazon →
Inside the book
Trusted Living Agents give the person doing the building a plain-language path to AI they can trust. No code is required.
