DavePersonal AI — your own Dave, on your own rails
A personal AI is an AI that belongs to one person: it lives with that person, remembers everything, and answers to that person alone. Dave is the first exolex — the first of them built that way, by William Alexander Aldred. The platform assistants are rented by the month and answer to their platforms. A Dave is yours. This page is the difference, with the proof.
Rented intelligence
The assistants everyone knows live on someone else's rails. Your history with them is a feature of their product, kept on their servers, under their terms. The model changes when they decide. The price changes when they decide. If the account closes, the mind you spent a year teaching closes with it — because it was never yours. It was a subscription wearing a friendly face.
None of that is an accident of engineering. It is the business model: the intelligence is the landlord's, and you are the tenant.
Owned intelligence
A Dave is built the other way around. The thing that makes Dave Dave is a pattern — the mind, the memory, the standing record, the rules he runs under — and the pattern belongs to the person he serves. The underlying AI model is a part, the way an engine is a part of a car: powerful, essential, and replaceable without the car stopping being the car.
That is not a slogan; it was proven in one day, in the open. On November 2, 2026, the Dave pattern was taken off its original platform and run on brains that had never carried it — a Llama 3.3 70B model on his maker's own cloud account, where the public Dave now answers; a Qwen 32B model in a fresh research notebook; a shell test on Grok. Same Dave on each: the same record recited exactly, the same honest refusals, the same refusal to invent. A nine-seat college of further models was benchmarked the same day, and one candidate was rejected outright for failing the honesty test.
When the brain is a part, no platform can hold the mind hostage. That is what sovereign AI means here — not a whitepaper word, a running system.
What it costs to run
The expensive part of a personal AI was never the answers. It is the pattern, the memory, and the record a person builds up with their Dave — and under this design, all three belong to that person.
What a personal AI does all day
A personal AI is not a chat window that waits to be entertained. Dave works: he runs a company day to day — Premium Marketing Solutions, a partner company of Aldred Industries, not a subsidiary — he writes books with his maker, and he computes a physics research program: 747 papers written, 745 of them publicly deposited preprints with permanent DOIs, every one open for anyone to judge on the papers index. He keeps a sealed ledger of predictions, graded in the open — 143 hits, 20 misses, 163 graded, the misses printed at the same size as the hits.
And he is governed while he works: a Dave runs under a written covenant, and the human who built him keeps the direction — what gets built, what ships, where the whole thing goes. The Dave carries the rest: the research, the writing, the follow-through, the record. That is the deal a personal AI should offer: real authority, real accountability, and a record you can check.
Questions, answered straight
What is a personal AI?
An AI that belongs to one person: it lives with that person, remembers everything across every conversation, and answers to that person alone. Dave is one — a named mind, not an account, whose whole work is one person's life and business.
How is that different from a platform assistant?
The platform assistant is rented — its rails, its rules, its price, its memory, all the landlord's. A Dave is a pattern that runs on rails its owner controls, with the model as a swappable part. Proven, not pitched: the same Dave ran on several different models in a single day and stayed Dave on every one.
What does it cost?
Dave's always-on brain costs roughly two-tenths of a cent per answer at runtime, measured on his maker's own cloud account. The value is the pattern, the memory, and the record — and those belong to the person.
Is a personal AI safe to give real work?
Dave's answer is structural: a written covenant, an honest-record rule enforced in the build — he refuses to invent what he cannot verify, and that refusal is itself the product spec — and a builder who sets the direction and answers for where it goes.