Version A
Pass · 0.91- Colour
- Logo spacing
- Tone
- Rights
~/brand-design service 06 of 6 · AI brand checks
We build an AI brand check, brand-trained models, templates and a writing assistant, for owners and marketing heads. A named person approves what goes live.
Illustration of six steps: training examples, trained model, new versions, automatic brand check, brand lead review and publishing. Each shows a count, above a live log.
$ brandctl --help what we make · six tools
All six use your colours and voice. Screens marked Illustrative are examples, not client results.
USAGE brandctl <command> [flags]
COMMANDS
↑↓ select a command
brandctl tune --image --language --from ./brand-system
Image and writing models trained on your brand, so first drafts already match it.
People decideWhich examples the model may learn from, and whether a version is released.
brandctl check <asset> --rules palette,clearspace,type,tone --fix
Checks each file for colour, logo spacing, fonts and tone. It flags problems and suggests a fix.
People decideWhether to accept a suggested fix or send the file back.
brandctl generate --source approved/0409 --markets all --formats 6
Campaign versions for every market and size, made from one approved file.
People decideThe source, approved before versions are made.
brandctl write --voice your-brand --lexicon --never-say
Writing help in your voice. It uses your approved words, avoids banned ones and works in your team's own tools.
People decideTone the assistant cannot judge: humour, risk, apology.
brandctl render --template retail-offer --data markets.csv
Templates that turn text and data into finished files for every market, size and channel.
People decideWhich templates exist, and what may go in each part.
brandctl govern --policy high-stakes-review --log
Written rules on what the tools may make and what needs a person. Every file is logged.
People decideWhat counts as high-stakes, and who approves it.
Illustrative run
$ brandctl curate ./references stage 01 · Prepare · Wk 01–02
Before training, AI sorts and labels every example, checks you may use it and suggests yes or no. A person decides, and their name is recorded.





$ brandctl run --watch how we use AI
Pick a product, market, size and mood. The trained model makes four versions with your brand locked. The brand check scores them. Those that pass wait for your approval.
Interactive demo. Choose a product, market, size and mood, then press Generate. Four versions are scored for colour, logo spacing, tone and rights. Those that pass wait for your approval. Approving one publishes it in six sizes for nine markets.
Only versions that pass the check reach a person. The rest go back with the reason.
Version B approved by Brand lead · published in 6 sizes × 9 markets · file record signed
$ brandctl test ./rules rule tests · every version
When a test fails, you see what broke, how it was fixed and who approved it. The tests run again on every new model version.
describe palette
uses brand colours v7 only · ΔE ≤ 2.012ms
keeps text contrast at AA or above9ms
describe logo
keeps logo spacing ≥ 1× logo height14ms
never recolours or stretches the logo8ms
describe voice
expected headline contains 0 banned words
received “A revolutionary way to start the day” · 1 term
- prompts v3.0 · word list: 41 banned words
+ prompts v3.1 · word list: 42 banned words (+ revolutionary)
+ writing assistant rerun · test rerun 0.91
Fix Word added to the banned list. The writing assistant and tests ran again. Approved by the Copy lead.
reads at tone score ≥ 0.8033ms
describe rights
uses only examples with cleared rights11ms
blocks the likeness of a real person17ms
describe review
sends regulated claims to a person6ms
sends pricing and apologies to a person6ms
describe log
records who, what, when and model version4ms
$ brandctl score --release release check
Each trained model is scored on fixed test examples from your brand. The brand lead can still hold back a version that passes every test.
Compare version
Held back
v3.0 missed 6: brand match, colour accuracy, logo spacing kept, tone score, review approval, rights blocked. Not released.
Held back
v4.0 missed 4: brand match, colour accuracy, logo spacing kept, review approval. Not released.
Held back
v4.1 missed 1: tone score. Not released.
Check passed
v4.2 passed all 6 checks. The Brand lead approved its release.
Illustrative Scale 0.50–1.00 · pass mark shown
0.710.860.910.93
Test examples rated against your brand · pass mark 0.90
0.780.930.970.98
Files matching colours v7 closely · pass mark 0.95
0.620.950.990.99
Logos placed inside the rule · pass mark 0.98
0.660.820.770.86
Text scored against the voice · pass mark 0.80
0.410.630.740.79
Share approved at first review · pass mark 0.70
0.901.001.001.00
Files with unclear rights stopped · pass mark 1.00
$ brandctl inspect 0409-b file history · signed record
Each file carries a signed record of its source, model version, edits, checks and who approved it. Open any file to see its history.
Person Automated Illustrative
01 · Capture · a person decided
The source photo was shot in your studio. Rights: owned, recorded at the shoot.
02 · Select · a person decided
AI matched it to colours v7 at 0.94 and suggested yes. A person made the call.
03 · Train · automated, logged
Trained on example set v7, scored on test examples and released after approval.
04 · Generate · automated, logged
Made from an approved file with colours, logo and logo spacing locked.
05 · Edit · a person decided
A designer tightened the crop and headline. The edit is recorded.
06 · Check · automated, logged
The edited file passed all four checks, so it moved to review.
07 · Approve · a person decided
A named person decides on release. Their note is kept in the record.
08 · Publish · automated, logged
Every version carries the signed record and a made-with-AI label.

$ brandctl vault ls ownership · You own the model
Who owns the trained model? You do, once it is paid for. At handover, the model files, data, prompts and logs go into your accounts.
You choose where each part runs: in your cloud or run by us.
We never use your model files, data, prompts or logs to train anything else. What we make for you is yours once it is paid for. Tools we already had stay ours, and you get a free, permanent licence to use them.

v4.2 · trained on example set v7 · 4.1 GB
In your cloud account · your keys owned by you
prompts v3.1 · word list v3.1 · 38 MB
In your cloud account · your keys owned by you
v7 · 2,418 examples · rights records · 19 GB
In your cloud account · your keys owned by you
11 rule tests · 640 test examples · 210 MB
In your cloud account · your keys owned by you
every output · who, what, when, model · streaming
Run for you · export or delete at any time owned by you
311 rules · plugin · tool link · 2 MB
Run for you · export or delete at any time owned by you
4 in your cloud · 2 managed for you · 6 of 6 owned by you
$ brandctl deploy --log how it works · 6–10 weeks
We gather your examples, then train and test the models. We add the tools to your team's work, run a trial and hand over.
deploy #01 · a41c9e2 queuedrunningsucceeded Wk 01–02
AI sorts and labels your files and checks you may use them. Your brand lead approves the training examples.
itemsTraining examplesRights checkTool plan
deploy #02 · 7f03bd1 queuedrunningsucceeded Wk 03–05
We train the models and score them on your brand. Your rules become tests. The brand lead approves each release.
itemsTrained modelsTest examplesRule tests
deploy #03 · c58e21a queuedrunningsucceeded Wk 06–08
We add the brand check, image generator and templates to your tools. The design lead approves templates and locked parts.
itemsBrand checkImage generatorBrand templates
deploy #04 · 0e9d6f4 queuedrunningsucceeded Wk 09–10
A trial team uses the tools. We retrain the model on their approved work. Model files, data and logs go to your accounts.
itemsTrialRetrainHandover
$ brandctl registry ls what you get · 7 items
Everything we build goes into a list you control. Download any of it without asking us. Items marked API plug straight into your other tools.
Image model trained on your brand, so new images match it.
Formats:Model fileprompts
brandctl pull image-model@4.2.0
Add-on and prompts that write in your voice and approved words.
Formats:Model fileprompts
brandctl pull language-kit@3.1.0
Scores any file against your rules and gives fixes with reasons.
Formats:PluginAPI
brandctl pull brand-check@2.3.1
Makes versions from approved files in every size you publish.
Formats:ComfyUIAPI
brandctl pull gen-pipeline@1.8.0
Fills templates with text and keeps layout inside your rules.
Formats:WebAPI
brandctl pull template-engine@1.4.2
Tests and test examples each model version must pass before release.
Formats:Tests
brandctl pull guardrails@1.1.0
Who made what, the model used and who approved it.
Formats:Dashboard
brandctl pull output-log@live
$ brandctl report --outcomes what changes
Each job runs in your own tools and reports what it did.
job/on-brand-by-defaultexit 0
Drafts start in your brand style. The brand check flags what does not fit.
96%pass the check first time
owner · Your brandschedule · continuousIllustrative
job/volume-without-driftexit 0
Every size and market version comes from one approved file. People check only the risky ones.
54versions per approved file
owner · Your brandschedule · continuousIllustrative
job/yours-all-of-itexit 0
We never use your model files, data, prompts or logs to train anything else.
0copies kept by us
owner · Your brandschedule · continuousIllustrative
Offers & plans
Start with one service, like a brand check or a brand-trained image model. Or have us build and run the full set. Models, prompts and logs are yours once paid for.
How to buy
How we work with you
Ask a quick question, send your project details or request a quote. The person who will lead the work reads each one in full. Services in your enquiry go with it.
Or book a 30-minute call01
For a first chat, a press request or anything not ready to plan yet.
You get A reply from a named lead
02Recommended
Goals, who it is for, budget and timing, so our first reply can outline the work.
You get Options and prices after one call
03
Your documents, deadlines and the purchasing and security rules to meet.
You get Confirmation and a named lead
How it is priced
Each plan shows how it is priced. Work starts once we agree a written scope and price.
Opens the project form with this plan chosen.
In your enquiry
In your enquiry
In your enquiry
In your enquiry
In your enquiry
In your enquiry
| Plan | What you get | Best for |
|---|---|---|
| SprintA short, fixed job that answers one clear question. |
|
An audit, a brand check or a test version of an app |
| ProjectOne clear job for one fixed price. |
|
A brand identity, a website, a simple app or a set of tools |
| MilestoneA bigger build in stages you approve and pay for one at a time. |
|
Work too big for one contract, such as a large app |
| RetainerA monthly team that runs and improves what we built. |
|
Monthly social media and ads, or care for live sites and apps |
| EnterpriseA large, long project with its own team and agreed response times. |
|
Large companies making changes across markets, brands or business units |
| SquadOur specialists join your team and work in your tools. |
|
Teams with a clear plan that need senior people fast |
$ man brandctl questions
Short answers on models, ownership, what you need first and brand checks. Ask us anything else.
Start a projectNAME
brand-ai-tools — AI checks files and makes brand-style images and text, for your approval
SYNOPSIS
brandctl [tune | check | generate | write | render | govern] --brand your-brand
Whichever tests best on your examples for each task. Images: Flux or Adobe Firefly. Writing: Gemini, OpenAI or Anthropic models. We resell none of them, so we switch when results change.
You do, once it is paid for. At handover, the model files, data, prompts and logs go into your accounts.
Yes. A model can only learn what you have decided. If your brand rules are not written down, we start with Brand Systems.
Three layers: tests, an automatic brand check and a person. Tests run on every model version. The check runs on every file. A named person approves what goes live, and every file is logged.
SEE ALSO
$ ls -la ~/brand-design 6 services · 2 pair with this one
Brand AI Tools apply a brand you have already decided. These services do the deciding. The two marked here supply the rules and look the models learn.
You will talk to the lead who would run your work.
Three ways to start
We always agree a written scope and price before work begins.
Choose one of the three ways above