01 / 06
AI opportunity mapping
We map your users’ tasks to show where AI saves effort, and where it just adds a box to fill.
Task by task
AI Product Strategy & Development
AI features designed into your product, tested before launch.
We find where AI helps your users, then design and build the feature with your product team. We test it on your own examples before launch. After launch, we keep checking it.
An illustrative AI feature spec for drafting support replies. It scores 3 model options against 5 pass marks, with the fallbacks that cover each gap.
What it covers · 6 parts
AI features designed for when the AI is wrong.
We plan what people see when the AI is unsure and how they correct it. Each correction improves the next release.
02 / 06
AI interaction patterns
Reusable designs for live answers, sources, confidence, undo and passing to a person, added to your design kit.
Reusable · design kit
03 / 06
AI prototype on live models
A working draft on live AI models and your own content, so you judge the idea on its answers.
Live answers
04 / 06
Tested on your examples
A set of your own examples for each task, graded with your experts, plus trust tests with future users.
Test set · user tests
05 / 06
Trust, transparency & control
Answers labelled as AI, with sources shown and easy to edit or undo. A record shows what the AI did.
AI labels · action log
06 / 06
AI feature build
Built with your engineers and linked to your documents. Safety checks, a person’s sign-off and quality tracking are built in.
Launched · monitored
How it works
A tested idea in 4 weeks, then build and launch.
From week 2 the draft runs on a live AI model. Its mistakes shape the design before anything is built.
-
Define Wk 01–02
We learn who does the task today, what a wrong answer costs and what data you have. Then we write test cases with your experts.
- Task summary
- Test set v1
- Risk rating
Check · Is AI the right tool for this task?
-
Draft Wk 02–04
We design the screens and connect them to live AI and your content. Users test if it helps and what they do when it is wrong.
- Working draft
- Test findings
- Design choices
Check · Does the draft pass its test set?
-
Build Wk 04–12
With your engineers, we connect it to your documents or tools and add safety checks and sign-offs. It goes live only after passing agreed tests.
- Live feature
- Safety checks
- Release tests
Check · Are safety checks and sign-offs in place?
-
Improve Wk 12+
We release in stages and track use and corrections. New cases join the test set.
- Roll-out
- Quality dashboard
- Design kit updates
Check · Is it still used, and still correct?
Typical times. They are shorter if you already have research.
Launch tests · what you will see
A scorecard of tests decides when an AI feature goes live.
It shows the agreed pass marks, how each model scores on your test cases and the fallback for each gap.
| Check | Pass mark | Option A · large model | Option B · small model + your docs | Fallback |
|---|---|---|---|---|
| Answer backed by the help centre | ≥ 95% | 97.1% (Passes) | 95.4% (Passes) | Source or no answer |
| Declines off-topic requests | ≥ 98% | 98.6% (Passes) | 96.2% (Fails) | Pass to a person |
| Hidden-instruction attack tests | 0 passes through | 0 of 60 (Passes) | 1 of 60 (Fails) | Block and log |
| Personal data in answers | 0 cases | 0 (Passes) | 0 (Passes) | Hide it |
| Reply time | ≤ 2.5 s | 3.1 s (Watch) | 1.4 s (Passes) | Type it out live |
What you get
Your AI feature, with the spec, tests and dashboard behind it.
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.
/handover/04-ai-product/
AI Product Strategy & Development
- AI opportunity map, task by taskBoard · Figma
- AI interaction pattern setFigma · Storybook
- AI prototype on live modelsCoded draft
- Test set & resultsData · report
- Trust & AI label guidelinesDocument
- Live AI featureCode · your cloud
- Use & quality dashboardDashboard
/handover/README
In every handover
- Decision record, with sourcesDoc
- Assumption tracker, with resultsSheet
- Research consent & deletion logSheet
- Walkthrough session, recordedVideo
Results
3 measures for use, mistakes and safety controls.
An AI feature earns its place when people keep using it. It must also catch unsure answers and have controls on day 1. We agree the start, target and data source in week 1.
-
Outcome 01
Still used after launch
We measure use after the launch rush, plus how often people correct it, give up or ask a person.
-
Outcome 02
Wrong answers handled well
Unsure answers, missing sources and refusals each get their own screen, so mistakes are easy to recover from.
-
Outcome 03
Controls in place at launch
Risk level, AI labels, human review and records are set during design, when they cost least to add.
-
01 Active users in week 6Usage data after the launch rush46%from 20%target 40%
-
02 Unsure answers sent the right wayFallback shown or handed to a person, from logs95%from 40%target 95%
-
03 Risk controls in place at launchAI labels, human review, records and tests in place100%from 30%target 100%
Tools we use
AI models chosen per task and easy to swap without a rebuild.
The AI models and tools we use most, set up in your own accounts with your own keys. We work with these tools. No partnership is implied.
- Figma
- Anthropic
- Google Gemini
- LangGraph
- Python
- TypeScript
- React
- Next.js
- Storybook
- PostHog
- OpenAI
- LlamaIndex
- Playwright
Standards we build to
AI risk checked at the start, with human review designed in.
We check, test and document every AI feature against these standards. We hold no certificate for them.
-
EU AI ActArtificial Intelligence Act (EU) 2024/1689
Europe's AI law sets rules by risk: banned uses, duties for high-risk AI and general-purpose models, and AI labels.
How we apply itWe rate each AI job by risk early, and build in AI labels, records and human checks.
-
NIST AI RMF 1.0AI Risk Management Framework
Managing AI risk in 4 parts: Govern, Map, Measure and Manage. A second guide covers generative AI (NIST AI 600-1).
How we apply itWe test each AI job's risks on your own examples. A named owner approves each release.
-
ISO/IEC 42001:2023Artificial intelligence management systems
Rules for managing AI in a company: policy, impact checks and controls on data and models.
How we apply itBefore an AI tool goes live, we list it, check its impact and set its controls.
-
OWASP Top 10 for LLM ApplicationsSecurity risks in generative AI applications
Risks in AI apps, such as hidden instructions that trick the AI, data leaks and AI with too much power.
How we apply itBefore any AI goes live, we attack it with tests for hidden instructions, data leaks and tool misuse.
-
WCAG 2.2 AAWeb Content Accessibility Guidelines
Rules that make websites usable by people with disabilities. Version 2.2 adds visible focus, tap size and easier log-ins.
How we apply itWe check contrast, focus order and button size in the design file. Then we test key tasks by keyboard and screen reader before sign-off.
-
GDPRGeneral Data Protection Regulation (EU) 2016/679
Europe's data law: a legal reason to use data, people's rights, privacy by design, breach reports and risk checks.
How we apply itResearch participants give recorded, specific consent. Recordings and notes have a delete date, and we keep to it.
-
DPDP Act 2023Digital Personal Data Protection Act, 2023
India's data law and its Rules: notice and consent, people's rights, breach reports and extra duties for significant firms.
How we apply itConsent notices for people in India. Requests about their data can be answered from the research log.
Offers & plans
AI product services, bought alone or together.
Book a review of where AI fits, a working draft on live AI or a full AI feature. Each is judged on live answers and tested with its future users.
- Categories
- 04
- Offers
- 15
- Plans
- 04
How to buy
- 01Pick offers. Ask about one, or add several to one enquiry.
- 02Choose a plan. A small job, a fixed project or ongoing help.
- 03Send your enquiry. We reply within one working day.
How we work with you
Ways to start, from a question to a quote.
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 call-
01
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
- Typical length
- 1–3 weeks
- Pricing
- Fixed fee
-
In your enquiry
- Typical length
- 4–12 weeks
- Pricing
- Fixed price
-
In your enquiry
- Typical length
- 3–9 months
- Pricing
- Fixed price per stage
-
In your enquiry
- Typical length
- Ongoing · 3-month minimum
- Pricing
- Paid by time used
Compare what each plan includes
| 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 |
| SquadOur specialists join your team and work in your tools. |
|
Teams with a clear plan that need senior people fast |
Questions
Asked before we pick a model.
Anything else goes to the people who would do the work.
Book a callQ01How is this different from Technology & Intelligence?
We design the feature. Technology & Intelligence builds the AI underneath. We decide where AI goes and what people see. Most AI products need both, and our teams work as one.
Q02What makes an AI feature feel trustworthy?
Saying plainly that it is AI and showing where answers come from. People can edit and undo, and a person approves big actions. Trust comes from design choices, not tone of voice.
Q03Can its answers sound like our brand?
Yes, if your brand voice is written down. Its tone, words and banned phrases go into the AI’s instructions and tests. If not, our AI voice & tone guidelines can define it.
Q04How do you test something that answers differently every time?
We score it on a test set of your own graded examples. Software scores each answer, and people check the scores. Many runs show how often it passes each mark. User trust tests come on top.
Q05Does the EU AI Act affect a feature like this?
It can, even outside the EU. It applies if the feature is offered or its answers are used there. Most features must then tell people it is AI and label AI content. We check early and design this in.
More services
AI features are still products, so they need design and strategy too.
Experience Design makes them easy to use. Strategy picks where AI is worth building.
Tell us what you need built.
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