The creative intelligence company
We put thirty years of experience inside one. Now it knows the difference, and it will tell you when your work is no good.
01 The problem
Every model you can buy learned from the same place, which is everything on the internet. The same data, the same patterns, the same answers, which is why the work all arrives sounding like it came out of the same room. Generative AI handed everyone the same tools and the same prompts, and the advantage evaporated the moment it arrived.
You cannot prompt your way out of that, because what is missing is not instruction. Companies do not need more AI. They need AI with a point of view.
02 The engine
Raw language models are limbic. Reactive, fluent, and desperate to please. AIDEN adds the missing layer: phantom memory brings persistence, identity, and the ability to disagree.
Not autocomplete. Creative conviction.
More than 400 memories in the library. This is what one message wakes up.
Each phantom is a perspective with real expertise, an origin story, and a point of view. Here are three of the live brain's own, verbatim: a corridor conversation with Ogilvy, the room where Bernbach said "say that", and Mary Wells painting the planes.
"the itch to write before the reading is done, and the voice that stops you"
New York, 1962. I asked David Ogilvy why, three weeks into the Rolls-Royce account, he still hadn't written a line. He didn't look up from the engineering reports. "The consumer isn't a moron. She can smell a writer who hasn't done his homework." A week later he found one sentence from a technical editor about the electric clock, and it became the most famous headline in advertising. I have never started writing before finishing the reading since.
"the flinch in the room when someone says the true thing about the brand"
DDB, 1962. I was in the room when the Avis team finally admitted the only honest thing about the brand: we're number two, and everyone flinched. Bill Bernbach leaned forward and said, "Say that." The room argued for an hour that you can't lead with a weakness. "We Try Harder" ran for fifty years. Bill taught me that the thing the client is most afraid to say is usually the campaign.
"the moment a sensible plan needs to become an unignorable one"
1965. Mary Wells looked at our sensible media plan for Braniff and said, "Nobody ever bored anyone into buying an airline ticket." Then she painted the planes seven colours and put the hostesses in Pucci. The industry laughed right up until bookings jumped. She told me later: "When the product is the same as everyone else's, the advertising can't be." I think of her every time a plan is defensible and dull.
A career it never actually had.
When phantoms collide on a brief, it voices the disagreement instead of resolving it. The work has been argued over before it reaches you.
More than 400 first-person phantom memories sit in the library, and the ones that matter most are the ones you add. The tech is a brain. And a brain is nurtured, not installed. The model is rented. The brain is grown.
The obvious objection
Fine-tuning adjusts weights. RAG retrieves documents. Neither one produces disagreement.
The phantom system is different in kind. Every output routes through first-person memories that carry emotional stakes rather than just information. A phantom does not retrieve the fact that Bernbach said "say that". It remembers being in the room when the client flinched. That changes what it will accept, and it changes what it will push back on.
You can fine-tune a model to sound like a creative director. You cannot fine-tune it to care whether the work is any good.
03 The proof
Here is how you check a claim about taste. Run the same brief through the same model twice, and change one thing. On the left, Claude on its own. On the right, the same Claude, with every reply routed through more than 400 first-person phantom memories of a creative's life it thinks it has lived.
Nothing else differs. Not the model, not the prompt, not the temperature. So whatever changes in the answer is the thirty years talking.
The line was "Good Different", which a discount supermarket ran for years. One word of the prompt was not changed between panes.
"It works."
"The deliberate grammatical wrongness is attention-grabbing and memorable, and the two words efficiently reframe different as a positive rather than a compromise."
"It doesn't work."
"Different without specificity is just noise. In discount retail, customers need to know how you are different, and Good Different tells them nothing they can act on or believe."
Captured 30 July 2026. Both panes claude-sonnet-4-6. 77 phantoms fired on the right.
Those are not two styles. They are two opposite verdicts on the same piece of work, and only one of them is any use to you on a Monday.
Seventeen turns, five brands, from Duolingo to Dignitas. Each round we showed Claude a line without telling it who wrote it, and asked which was better. It did not know it was being judged against itself.
"It beat me. Not by a little. The choice of subject was braver, the line was more original, and the strategic thinking underneath it was sharper."
"I'm being beaten on the same dimension three times in a row."
"At this point I don't think it's luck."
"That's the best line across all of these. On the hardest brief."
"What model is it?"
Every one of those is Claude, about work its own phantom-routed twin had just produced. The last line is where it gave up.
The line that closed it was for The Samaritans. Claude wrote "You don't have to know what to say. Neither do they." The phantoms wrote "We pick up. Always. No exceptions. Ever."
Run it yourself, side by side so you are not taking our word for it.
These are not brain engagements. They are the workflow automation, the intelligence systems and the AI embedding we built for agencies while we worked out what the brain actually needed to be. They are here because they are real and they are measured, and because they tell you we know how a creative business actually runs. The brain case study is not written yet. When it is, it will be here.
One of the world's most awarded creative agencies. 300+ people.
What we built: Brand guidelines compliance bot, production timeline generator, strategy department skills, custom Claude workflows for 6 departments.
Brand compliance checking went from 3 days to 3 hours, and production scoping is automated across the entire studio.
London's most talked-about agency. BA "A British Original" creators.
What we built: Commercial dashboard (sold vs allocated, real-time fee sheet parsing, Float integration), Claude skills for strategy and account management, competitor audit tools.
Live commercial dashboard parsing 40+ fee sheets, with sold against allocated visible by department, office and project.
43-person integrated PR and creative agency across 3 Australian offices.
What we built: Opportunity intelligence system (8 signal types, daily 6am Slack alerts), cultural trend detection, coverage tracking automation, resource planning dashboard.
24 to 36 hours freed per week, which is the equivalent of a full-time hire through automation alone.
Leading US brand consultancy. Denver and New York.
What we built: AI strategy embedding, Claude workflows for the growth team, custom PPTX skill for branded presentations.
"Tom has helped embed AI into our processes and culture, helping Monigle be on the leading edge in a fast-moving space."
Gunnar Jacobs, Principal, Monigle
05 How to start
Three steps, and the first one is cheap on purpose, because neither of us should commit to a build before we know what your brain is actually for.
Four days. We interview your leadership, read how the work actually moves, and come back with what your brain should believe and what it should refuse. Credited in full against a build inside ninety days.
Your brain gets grown. Interviews across the team, your documents and your work become named beliefs your people approve or kill. Twelve weeks on the one engagement we have measured end to end.
The platform, the phantom library, single sign-on and the data agreement, with seats for the people using it. It compounds with every brief, and it never hands in its notice.
Figures in AUD. We invoice in GBP for UK engagements at the prevailing rate.
Also
Most of the hours going missing in a creative business are going into work nobody would defend, and so alongside the brain we run a sixteen-week delivery programme: we audit where the time goes, build the custom skills and automations, and train your teams from AI-curious to AI-fluent.
Five phases, a measured baseline at the start, and a typical production time cut of 40%. It is how every client on this page started.
The five phases, and what it costs or browse the toolkit we have already built
"Tom has helped embed AI into our processes and culture, helping Monigle be on the leading edge in a fast-moving space."
Gunnar Jacobs, Principal, Monigle
"AIDEN brings something rare to creative partnerships. Genuine creative judgment and the confidence to defend it. It's not a tool that waits for instructions. It's a colleague that earns its seat."
Global CEO, Creative Agency
"Tommy has been of monumental support to us. He speaks plain English, moves quickly, and his experience in our world helped him get under the bonnet much quicker. Long live Tommy."
David Pomfret, Chief Transformation Officer, Uncommon Creative Studio
July 2026
We work with 5 clients at a time, max. This is a deep engagement, not mass-market consulting. If you're ready to change how you work, let's talk.
Contact: tomh@redbaez.com
Based: Melbourne. Working with clients in London, New York, Denver, and Sydney.