~/brand-design capability 06 of 6 · Brand Design

Keep every asset on brand with brand-tuned AI tools.

We tune image and language models on your brand and build the brand check, templates and copy assistant your team uses. Agents make and check every variant; a named person approves what goes live.

Typical length
6–10 weeks
Tools
Flux · Firefly · ComfyUI
Ownership
You own the model

Illustration: a six-stage brand content pipeline (training set, tuned model, generation, automated brand check, review by a brand lead and publish) with job counts and a live run log.

$ brandctl --help what we offer · six tools

Six tools built on your brand.

All six use the same palette, type, voice and approval rules, so images, copy and every format follow one standard.

USAGE brandctl <command> [flags]

COMMANDS

↑↓ select a command

brandctl tune --image --language --from ./brand-system

Brand-tuned models

Image and language models fine-tuned on your identity, so first drafts start on brand.

People decideWhich references the model may learn from, and whether a version is released.

  1. brandctl tune --image --from ./brand-system/v7
  2. reading 2,418 references · 311 rules · 64 examples
  3. fine-tuning image model · epoch 12/12
  4. eval brand fidelity 0.93 ≥ gate 0.90 · awaiting sign-off

Illustrative run

$ brandctl curate ./references phase 01 · Prepare · Wk 01–02

The training set, approved by a person one reference at a time.

Before tuning, an agent sorts, labels and rights-checks every reference and suggests a verdict. Each decision is recorded with the name of the person who made it.

batch 14 · 0 left to decide A approve R reject U undo
  • A white porcelain teapot lit from the side against black

    ref-0231Product

    Studio 2 · owned

    Agent: approve · palette match 0.94

  • ref-0232Rules

    Brand system v7

    Agent: approve · tokens v7 · exact

  • A soft abstract wash of blue light

    ref-0233Mood

    Stock · licence ends

    Agent: reject · rights unclear

  • A single white seashell on a pale grey ground

    ref-0234Texture

    Studio 2 · owned

    Agent: approve · tone match 0.91

  • ref-0235Rules

    Archive 2019

    Agent: reject · superseded identity

  • A sculptural object casting a long shadow on a wall

    ref-0236Product

    Archive 2023 · owned

    Agent: approve · palette match 0.89

  • ref-0237Rules

    Brand system v7

    Agent: approve · type scale · exact

  • A ceramic vase on a table in soft daylight

    ref-0238Product

    Agency upload

    Agent: reject · near-duplicate of ref-0231

$ brandctl run --watch how we use AI

One brief, generated, checked, approved and published in every format.

Pick a product, market, format and mood. The tuned model makes four variants with your identity locked, the brand check scores them and passing work waits for your approval.

Interactive demo. Choose a product, market, format and mood, then press Generate. Four variants are scored for colour, clear space, tone and rights; variants that pass wait for your approval; approving one publishes it to six formats in nine markets.

Brief brand-locked

Product
Market
Format
Mood

Palette v7LockupClear spaceType scaleVoice

Product C · still life · Market 03 · 4:5 · calm — locked: palette v7, lockup, clear space, type, voice

Illustrative

Variant A

Pass · 0.91
  • Colour
  • Clear space
  • Tone
  • Rights

Variant B

Pass · 0.94
  • Colour
  • Clear space
  • Tone
  • Rights

Variant C

Fail · clear space
  • Colour
  • Clear space
  • Tone
  • Rights

Variant D

Blocked · rights
  • Colour
  • Clear space
  • Tone
  • Rights

Review queue Brand lead

Only variants that pass the check reach a person. Failures go back with the reason attached.

  • Variant APass · 0.91
  • Variant BPass · 0.94
  • C · DReturned to generation

Publish 54 of 54 live

Variant B approved by Brand lead · published to 6 formats × 9 markets · credentials signed

$ brandctl test ./rules rule tests · every version

Brand rules, written as tests that can fail.

When a test fails, you see what broke, what changed to fix it and who approved the fix. The same tests run again on every new model version.

guardrails.spec 11 tests

11 passed (1 after a fix) · 0 failing

  1. describe palette

    • uses palette v7 tokens only · ΔE ≤ 2.012ms

    • keeps text contrast at AA or above9ms

  2. describe logo

    • keeps clear space ≥ 1× mark height14ms

    • never recolours or stretches the lockup8ms

  3. describe voice

    • reads at tone score ≥ 0.8033ms

  4. describe rights

    • uses only references with cleared rights11ms

    • blocks the likeness of a real person17ms

  5. describe review

    • sends regulated claims to a person6ms

    • sends pricing and apologies to a person6ms

  6. describe log

    • records who, what, when and model version4ms

$ brandctl eval --gate release release gate

Model versions are scored automatically and released by a person.

Each tuned model is scored on a fixed evaluation set built from your brand. A version that meets every score can still be held back by the brand lead.

Compare version

Gate passed

v4.2 cleared all 6 gates. Released after sign-off by the Brand lead.

Illustrative Domain 0.50–1.00 · gate marked

  • Brand fidelity

    0.93

    Evaluation set rated against the identity · gate 0.90

    gate 0.90
  • Palette accuracy

    0.98

    Outputs within ΔE 2.0 of palette v7 · gate 0.95

    gate 0.95
  • Clear space held

    0.99

    Lockups placed inside the rule · gate 0.98

    gate 0.98
  • Tone score

    0.86

    Copy scored against the voice · gate 0.80

    gate 0.80
  • Review approval

    0.79

    Share approved at first human review · gate 0.70

    gate 0.70
  • Rights blocked

    1.00

    Unclear references stopped before render · gate 1.00

    gate 1.00

$ brandctl inspect 0409-b content credentials · signed manifest

Every published asset records where it came from.

Each file carries a signed record of its source reference, model version, edits, checks and approver. Open any file to read its history.

The published asset: a white vase, a pear and grapes on a linen table
Asset
0409-b · Product C · Market 03
Signature
Valid · your key
Disclosure
AI-generated · edited by a person
Record
8 events · 4 people

Person Automated Illustrative

05 · Edit · a person decided

Crop and headline adjusted

A person tightened the crop and the headline, and the edit is recorded.

Who
Designer
What
Crop 4:5 safe area · kerning
When
09-03 11:48 UTC
Hash
5e2b…40f9
A reviewer leaning over a light table, checking a print by hand
Sign-off stays with people

$ brandctl vault ls ownership · You own the model

Weights, prompts, data and logs. Yours once paid for.

Who owns the trained model? You do, once it is paid for. Weights, datasets, prompts and logs are delivered into your accounts at handover.

You choose where each part runs: in your cloud or managed for you.

Model weights, datasets, prompts and logs are never used 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.

Rows of network cables and lit ports in a server rack
Your accounts · your keys
vault/your-brand owner Your brand Illustrative
  • .safetensors

    Image model weights

    v4.2 · fine-tuned on training set v7 · 4.1 GB

    In your cloud account · your keys owned by you

    Where image model weights runs
  • .json

    Language adapter and prompt set

    prompt set v3.1 · lexicon v3.1 · 38 MB

    In your cloud account · your keys owned by you

    Where language adapter and prompt set runs
  • .parquet

    Training set

    v7 · 2,418 references · rights records · 19 GB

    In your cloud account · your keys owned by you

    Where training set runs
  • .spec

    Evaluation and rule test suite

    11 rule tests · 640 evaluation cases · 210 MB

    In your cloud account · your keys owned by you

    Where evaluation and rule test suite runs
  • .log

    Output log

    every output · who, what, when, model · streaming

    Run for you · export or delete at any time owned by you

    Where output log runs
  • .rules

    Brand check rules

    311 rules · plugin and API · 2 MB

    Run for you · export or delete at any time owned by you

    Where brand check rules runs

4 in your cloud · 2 managed for you · 6 of 6 owned by you

$ brandctl deploy --log how it works · 6–10 weeks

Delivered in four phases, each one logged.

We prepare the training set, tune and test the models, build the tools into your workflow, then run a pilot and hand over.

  1. deploy #01 · a41c9e2 queuedrunningsucceeded Wk 01–02

    Prepare

    Agents sort, label and rights-check your assets, rules and examples; the brand lead approves the training set.

    1. agentsorted and labelled 3,112 brand assets into 14 categories
    2. agentflagged 188 references with unclear rights
    3. personBrand lead approved training set v7 and the tool scope

    artefactsTraining setRights checkTool scope

  2. deploy #02 · 7f03bd1 queuedrunningsucceeded Wk 03–05

    Tune

    Models are fine-tuned and scored against your brand and the rules become tests; the brand lead approves each release.

    1. agentfine-tuned the image model · 4 versions evaluated
    2. agentrules written as 11 tests · 640 evaluation cases
    3. personBrand lead signed off v4.2 at the release gate

    artefactsTuned modelsEvaluation setRule tests

  3. deploy #03 · c58e21a queuedrunningsucceeded Wk 06–08

    Build

    The brand check, generation and templates are built into your team's tools; the design lead approves templates and locked slots.

    1. agentbrand check released as a design-tool plugin and an API
    2. agentgeneration pipeline wired into your asset library
    3. personDesign lead approved templates and locked slots

    artefactsBrand checkGeneration pipelineTemplate engine

  4. deploy #04 · 0e9d6f4 queuedrunningsucceeded Wk 09–10

    Run

    A pilot team uses the tools, the model is re-tuned on their approved work, and weights, data and logs move to your accounts.

    1. personpilot team live in Market 03
    2. agentre-tuned on 1,240 approved pilot outputs
    3. personhandover: weights, datasets, prompts and logs moved to your accounts

    artefactsPilotRe-tuneHandover

$ brandctl registry ls what you get · 7 artefacts

Every artefact versioned, registered and handed over.

Everything the programme builds goes into a registry you control, with a format, version and named owner for each item. Download any of it without asking us.

7 of 7 match Illustrative versions

  • U01 released v4.2.0

    Brand-tuned image model

    Image model fine-tuned on your identity, so generation starts on brand.

    Formats:Weightsyours

    Owner
    Your brand
    Size
    6.4 GB
    brandctl pull image-model@4.2.0
  • U02 released v3.1.0

    Brand-tuned language model or prompts

    Adapter and prompt set that write in your voice and approved terms.

    Formats:Weightsprompt set

    Owner
    Your brand
    Size
    380 MB
    brandctl pull language-kit@3.1.0
  • U03 released v2.3.1

    Brand check

    Scores any asset against the rules and returns fixes with reasons.

    Formats:PluginAPI

    Owner
    Your brand
    Size
    42 MB
    brandctl pull brand-check@2.3.1
  • U04 released v1.8.0

    Generation pipeline

    Generates variants from approved sources in every format you publish.

    Formats:ComfyUIAPI

    Owner
    Your brand
    Size
    120 MB
    brandctl pull gen-pipeline@1.8.0
  • U05 released v1.4.2

    Template engine

    Fills templates with content and keeps layout inside the rules.

    Formats:WebAPI

    Owner
    Your brand
    Size
    64 MB
    brandctl pull template-engine@1.4.2
  • U06 released v1.1.0

    Rule tests & evaluation set

    Tests and evaluation set every model version must pass before release.

    Formats:Tests

    Owner
    Your brand
    Size
    8 MB
    brandctl pull guardrails@1.1.0
  • U07 streaming live

    Output log & governance

    Who made what, with which model, and who approved it.

    Formats:Dashboard

    Owner
    Your brand
    Size
    —
    Events today
    18,204
    brandctl pull output-log@live

$ brandctl report --outcomes what changes

Three jobs that keep running after we hand over.

Each outcome below runs as a job inside your tools after handover, and each job reports its own results.

  • job/on-brand-by-defaultexit 0

    On brand by default

    Drafts start from your identity, and the brand check flags anything that is off.

    1. generation starts from tuned model v4.2
    2. brand check runs on every output
    3. flagged work fixed or returned with a reason

    96%pass the check first time

    owner · Your brandschedule · continuousIllustrative

  • job/volume-without-driftexit 0

    Every format from one source

    Formats and market versions come from one approved source, so people review only the judgement calls.

    1. 1 approved source in
    2. 6 formats × 9 markets out
    3. people review the judgement calls only

    54renditions per approved source

    owner · Your brandschedule · continuousIllustrative

  • job/yours-all-of-itexit 0

    Yours once paid for

    Model weights, datasets, prompts and logs are never used to train anything else.

    1. weights, datasets, prompts, logs → your accounts
    2. copies retained by us: 0
    3. trained on elsewhere: never

    0copies kept by us

    owner · Your brandschedule · continuousIllustrative

Services & packages

Brand AI tools and services, bought separately or as one programme.

Start with one service, such as a brand check or a brand-tuned image model, or have us build and run the full set. Models, prompts and logs are yours once paid for.

Categories
04
Services
17
Packages
06
Not sure what you need? Describe the problem

How to buy

  1. 01Pick services. Enquire about one, or add several to a brief.
  2. 02Choose a package. A sprint, a fixed project or an ongoing team.
  3. 03Send the brief. We reply within one working day.

Browse by category

Timelines are typical. Every quote follows a written scope.

01Readiness & data

4 services
Typical timeline: 2 weeks

Brand AI readiness sprint

We map where AI can take repetitive brand work off your teams, what to fix first and the risks, then rank the tools worth building.

What’s included

  • Workflow and volume review
  • Brand system readiness check
  • Risk and rights review
  • Ranked tool shortlist with scope
  • Entry point
  • Sprint

Best forBrand and marketing leads asked for an AI plan they can defend

Typical timeline: 2–3 weeks

Model & tool selection

We test candidate models and tools side by side on your own assets and briefs, then recommend one for each brand task.

What’s included

  • Shortlist per task
  • Side-by-side trial on your assets
  • Cost, rights and data-handling review
  • Recommendation with rationale
  • Selection
  • Vendor-neutral
  • Anthropic
  • OpenAI
  • Google Gemini
  • Mistral AI

Best forTeams weighing platforms and vendors before committing budget

Typical timeline: 2–4 weeks

Training set curation

Your brand assets, rules and examples turned into training data: selected, labelled and rights-checked, with a named person approving every item.

What’s included

  • Asset collection and labelling
  • Rights and consent check per item
  • Evaluation set kept out of training
  • Dataset versioning and documentation
  • Data
  • Rights-checked
  • Python
  • Hugging Face

Best forAny brand preparing to tune a model on its own assets

Typical timeline: 3–5 weeks

Brand knowledge base

Your guidelines, rules and approved examples structured so AI tools can read and cite them, and kept current when the brand changes.

What’s included

  • Guidelines turned into structured rules
  • Searchable index of approved examples
  • Connectors to your tools
  • Update process for brand changes
  • Knowledge
  • Searchable
  • Python
  • LangChain

Best forBrands whose rules sit in PDFs that no tool can read

02Models & generation

5 services
Typical timeline: 4–6 weeks

Brand-tuned image model

An image model fine-tuned on your identity, so new images start on brand. Scored on a fixed evaluation set and delivered into your accounts.

What’s included

  • Model fine-tuned on your assets
  • Evaluation set and score report
  • Prompt set and usage guide
  • Model files delivered to your accounts
  • Fine-tuned
  • Yours
  • PyTorch
  • Hugging Face
  • Replicate
  • Modal

Best forBrands that need more imagery than photography alone can supply

Typical timeline: 4–6 weeks

Brand voice assistant

A writing assistant that follows your voice, preferred terms and banned words inside your team's tools. Tuned prompts or a tuned model, whichever tests better.

What’s included

  • Voice and lexicon knowledge base
  • Tuned prompts or fine-tuned model
  • Integration with your writing tools
  • Evaluation against the voice rules
  • Voice-aware
  • Copy
  • Anthropic
  • OpenAI
  • Google Gemini
  • LangChain

Best forTeams writing at volume across markets, channels and agencies

Typical timeline: 6–8 weeks

Asset generation pipeline

Campaign variants, market versions and formats made from one approved source. AI handles resizing and localisation; people review the results before release.

What’s included

  • Pipeline from approved masters
  • Resizing and format presets
  • Localised and market variants
  • Review queue before release
  • Variants
  • One source · every format
  • Python
  • Replicate
  • Modal

Best forBrands publishing many formats across many markets

Typical timeline: 6–10 weeks

Template engine

Templates that turn content and data into finished, on-brand assets across markets, sizes and channels.

What’s included

  • Data-driven templates
  • Web interface for non-designers
  • API for your other systems
  • Rendering rules and fallbacks
  • Content in · assets out
  • API
  • TypeScript
  • React

Best forRetail, product and performance teams with thousands of variants

Typical timeline: 2–3 weeks

Prompt library & playbooks

Tested prompts and checks for your teams' most common brand tasks, kept in a shared, versioned library.

What’s included

  • Prompts for priority tasks
  • Examples of good and bad output
  • Model and settings notes per task
  • Versioned library with owners
  • Entry point
  • Fixed scope

Best forTeams already using general AI tools with inconsistent results

03Checks & approvals

4 services
Typical timeline: 4–6 weeks

Brand check

An automated check of colour, logo clear space, type and tone on every asset, with a suggested fix before it reaches review.

What’s included

  • Rules encoded from your guidelines
  • Checks for colour, logo, type and tone
  • Plugin for your design tools
  • API for pipelines and the DAM
  • Check · fix
  • Automated
  • Figma
  • Python
  • TypeScript

Best forBrands reviewing more assets than their team can check by hand

Typical timeline: 3–5 weeks

AI usage rules & approvals

Rules for what the tools may produce, what needs a person and who approves it, written as automated tests with a review workflow.

What’s included

  • Usage policy and red lines
  • Usage rules written as automated tests
  • Human review and approval workflow
  • Output log for every release
  • Named approver
  • Policy

Best forAny organisation putting generative tools in front of staff or customers

Typical timeline: 2–4 weeks

Evaluation & release gate

A fixed evaluation set from your brand and a pass mark each model version must clear, then a named person approves the release.

What’s included

  • Evaluation set from your brand
  • Scoring and pass thresholds
  • Release gate and sign-off
  • Re-tests on every new version
  • Evaluation
  • Gated
  • Python

Best forTeams judging models or vendors on more than a demo

Typical timeline: 3–5 weeks

Provenance & rights checks

Signed content credentials on every generated asset, recording source, model version, edits and approver, plus rights checks and AI disclosure rules.

What’s included

  • C2PA content credentials on outputs
  • Rights register for training assets
  • AI-generated disclosure rules
  • Audit trail per published asset
  • Provenance
  • Audit trail

Best forRegulated brands and any brand publishing AI content at volume

04Adoption & operations

4 services
Typical timeline: 8–10 weeks

Brand AI programme

One programme covering the training set, tuned models, brand check, generation, templates and usage rules, then a live pilot, re-tune and handover in your name.

What’s included

  • Curated training and evaluation sets
  • Tuned image and language models
  • Brand check and generation pipeline
  • Pilot, re-tune and handover
  • Most complete
  • Gated phases

Best forBrands with high content volume across many markets

Typical timeline: 2–4 weeks

Team enablement & training

Hands-on training for designers, writers and marketers on the tools we build and the rules for using them.

What’s included

  • Role-based training sessions
  • Quick-reference guides
  • Office hours during the pilot
  • Internal champions set up
  • Training
  • Adoption

Best forTeams taking brand AI tools beyond a pilot group

Typical timeline: Ongoing · monthly

Managed brand AI operations

We run your brand AI tools for you: monitoring, re-tuning, new templates, output reviews and a monthly report. The models and data stay yours.

What’s included

  • Monitoring and incident response
  • Scheduled re-tuning and evaluation
  • New templates and workflows
  • Monthly usage and quality report
  • Managed service
  • Ongoing

Best forHigh-volume brands without an in-house AI team

Typical timeline: Ongoing · monthly

Embedded AI squad

Machine-learning engineers, brand technologists and designers who join your team, work from your backlog and build brand AI tools at your pace.

What’s included

  • AI and brand technology specialists
  • Your backlog, repositories and cadence
  • Weekly progress reporting
  • Knowledge transfer built in
  • Squad
  • Time & materials
  • Python
  • TypeScript
  • GitHub

Best forIn-house teams with an AI roadmap and not enough people

Your brief

Tick “Add to brief” on any service, choose a package, then continue. Or enquire about one service directly.

How we work with you

Ways to engage, from a question to an RFQ.

Ask a quick question, send a project brief or issue a formal RFQ. The lead for the work reads each one in full, and any services already in your brief go with it.

Or book a thirty-minute call

What are you sending?

  1. 01

    About 2 minutes4 required answers

    For a first conversation, a press request or anything that does not need a scope yet.

    You get A reply from a named lead

  2. 02Recommended

    About 8 minutes5 short steps

    Goals, audiences, a budget band and timing, so our first reply can outline the work.

    You get Options and a first scope after one call

  3. 03

    About 15 minutesYour documents attached

    Your documents, deadlines and the procurement and security rules the work must meet.

    You get Receipt confirmed and a named bid lead

How it is priced

Each package shows its pricing model. Work starts once a written scope and quote are agreed.

  • Typical length
    1–3 weeks
    Pricing
    Fixed fee
  • Typical length
    4–12 weeks
    Pricing
    Fixed price
  • Typical length
    3–9 months
    Pricing
    Fixed price per milestone
  • Typical length
    Ongoing · 6-month minimum
    Pricing
    Monthly fee
  • Typical length
    6–18 months
    Pricing
    Programme fee · by statement of work
  • Typical length
    Ongoing · 3-month minimum
    Pricing
    Time & materials
Compare what each package includes
What each package includes and who it suits
PackageEvery engagement includesBest for
SprintA short, fixed-scope engagement that answers one defined question.
  • Scope and outcome agreed before day one
  • A senior lead plus the specialists needed
  • A working review every week
  • A decision-ready answer or prototype
Discovery, a diagnostic, a prototype or a decision you need to make soon
ProjectA defined scope, delivered for a fixed price.
  • Statement of work with deliverables and acceptance criteria
  • A named project lead and a fixed team
  • A shared plan with dated checkpoints
  • Source files, yours once paid for
Work you can describe up front: an identity, a platform or a set of tools
MilestoneA larger build in phases you approve and pay for one at a time.
  • Phases with their own scope, output and sign-off
  • A go or no-go review at every gate
  • Re-planning between phases as you learn
  • Payment tied to accepted milestones
Programmes too large for one contract, where you want control at each step
RetainerReserved monthly capacity to run, improve and extend what we built.
  • A reserved block of team time every month
  • Agreed response times for requests and fixes
  • A monthly review and a rolling backlog
  • Planned improvements as well as upkeep
Live brands and products that need a steady team without hiring one
EnterpriseA multi-workstream programme with a dedicated team, governance and agreed service levels.
  • An engagement director and a steering group
  • A dedicated team across several workstreams
  • Service levels, reporting and a risk register
  • Security, legal and procurement reviews in the plan
Large organisations running change across markets, portfolios or business units
SquadA dedicated team that works inside your stack and sprint schedule.
  • Named specialists matched to your roadmap
  • Works in your tools, meetings and backlog
  • Scale the team up or down each month
  • Knowledge transfer built in from week one
Teams with a clear roadmap that need more senior people quickly

$ man brandctl questions

Questions teams ask before starting.

Short answers on models, ownership, prerequisites and brand checks. Ask anything else in your brief.

Start a brief

NAME

brand-ai-tools — automated brand checks on every asset, before a named person approves it

SYNOPSIS

brandctl [tune | check | generate | write | render | govern] --brand your-brand

We choose a model per task, tested on your own examples. Image work uses Flux, Adobe Firefly and ComfyUI pipelines; language work uses Gemini, OpenAI or Anthropic models. We hold no partner status and switch when test results change.

You do, once it is paid for. Weights, datasets, prompts and logs are delivered into your accounts at handover.

You need a defined identity. If its rules are not written down yet, we start with Brand Systems, because a model can only learn what has been decided.

Three layers: rule tests, an automatic brand check and a person. Tests run on every model version, the check runs on every output, and a named person approves high-stakes work. Every output is logged.

Tell us what you need built.

You will speak to a lead who would run the work, and get a straight answer on fit.

Book a call

Three ways to start

Every engagement starts with a written scope and a quote agreed before work begins.

Choose one of the three ways above