- Studio Vektor
- Interactive eBooks
- Stellar: UX of AI
Case study · Interactive playbook
UX of AI, from agency playbook to six working plays
Stellar’s playbook sets out six plays for bringing AI into a brand’s digital ecosystem. The interactive edition lets readers pressure-test a brand voice, simulate how AI assistants cite a page and plan how to measure an AI investment, then send their readout to book an audit.
- Publisher
- Stellar (Stellar Agency)
- Original format
- 29-page PDF playbook
- Edition
- 6 plays, 12 interactives, a self-check
- Status
- Independent portfolio demonstration
An interactive edition of a playbook Stellar published. The playbook’s text, stances and page numbering are as published; the summaries, simulations, fictional brands and quiz were written for this edition by Studio Vektor. Not commissioned by, affiliated with or endorsed by Stellar. Assistants, pages and curves in the simulations are illustrative.
The starting point: an agency playbook
The playbook covers six plays: where AI belongs, what a brand sounds like, what it does, what it knows, how it is found and how to know it’s working, each with a stance and watch-outs, and closes with a six-point AI audit. In print it is 29 pages, and its most practical ideas, from tuning a voice to generative engine optimization, are described rather than tried.
The brief
Brand leaders need to feel the trade-offs before they commit. The edition had to make each play a decision the reader can try.
Simulate the trade-offs
Let readers tune a brand voice and test it under pressure, or choose the data they’d use and see value against trust.
Show discovery changing
Simulate how assistants answer from a page, and what each optimization tactic does.
Close with an audit
Turn the six-point audit into a self-check and a readout to send.
Who it is for
CMOs, brand and digital leaders, and the teams who run websites and customer experience.
Brand leaders
Pressure-test a voice and sort guardrails.
Digital leaders
Run the GEO page lab and the visibility timeline.
Executives
Read the stances on the Quick track and take the self-check.
From PDF to edition, chapter by chapter
Each play shows the page range it covers in the original.
| Chapter | The play | In the interactive edition |
|---|---|---|
| Opening (p. 3) | You don’t need to be convinced that AI matters | A read-aloud introduction and a choice of track |
| Play 1 · Where AI belongs (pp. 4–7) | Start with invisible operations; the chatbot comes last | A ring-by-ring ecosystem map and three 18-month roadmaps |
| Play 2 · What your brand sounds like (pp. 8–11) | Voice is words, topics and behavior | A voice pressure test and a guardrail sorter |
| Play 3 · What your brand does (pp. 12–15) | Move from catalog to concierge | A fictional paint brand that switches from catalog to concierge |
| Play 4 · What your brand knows (pp. 16–19) | Personalization is a journey up a five-rung trust ladder | A trust ladder in a fictional store and a data diet |
| Play 5 · How your brand is found (pp. 20–23) | Content made for discovery | A GEO page lab and a twelve-month visibility timeline |
| Play 6 · How you know it’s working (pp. 24–27) | Decide what success looks like before launch | A measurement plan with a holdout |
| The six-point AI audit (p. 28) | Where to go from here | A six-point self-check, a five-question quiz and a readout to send, copy, share or print |
What Studio Vektor did
Turned each play into a decision the reader can test.
Built a voice sandbox
Readers tune warmth, length and behavior under pressure, then run five pressure prompts against a simulated assistant for a fictional retailer.
Simulated generative search
The GEO page lab applies three tactics from the research the playbook cites, citing sources, adding statistics and quoting an expert, to a fictional page and shows the simulated answer change.
Showed value against trust
The data diet finds the smallest set of customer data that delivers 80% of personalization value.
Ended with an audit
The six-point self-check and the reader’s answers make a readout to send to Stellar, copy, share or print.
The 12 interactives
Each play has at least one interactive; all twelve stay in both tracks.
| Interactive | Type | What the reader does |
|---|---|---|
| 01 · Ecosystem map | Self-assessment | Marks where each of six rings stands today, from invisible operations to visible brand voice |
| 02 · Three roadmaps | 12-second loop | Plays chatbot-first, inside-out and wait-and-see over 18 months |
| 03 · Voice pressure test | Sandbox | Tunes a voice and runs pressure prompts |
| 04 · Guardrail sorter | Sorter | Sorts what the assistant should and shouldn’t do |
| 05 · Catalog → Concierge | Before and after | Switches a fictional paint brand’s site from catalog to concierge |
| 06 · Trust ladder | Fictional store | Turns on rungs of personalization and watches relevance rise against customer comfort |
| 07 · Data diet | Optimizer | Picks the data to use and balances value against trust |
| 08 · GEO page lab | Simulation | Applies optimization tactics to a page |
| 09 · Visibility timeline | Illustrative curves | Sets refresh cadence, a model update and monitoring over twelve months |
| 10 · Measurement plan | Planner | Picks an initiative and its metrics, measured against a holdout |
| 11 · Six-point self-check | Assessment | Rates the brand on the six-point AI audit |
| 12 · Checkpoint quiz | Quiz | Answers five questions in two minutes |
Reading experience and accessibility
Twenty-nine print pages become one responsive playbook with the print numbering kept.
Two tracks
Quick adds a 30-second summary to each play and trims supporting paragraphs; Deep keeps every word, with every stance and interactive in both.
Listen along
Read-aloud of the introduction with word or sentence tracking at 0.8× to 1.5×.
Reading settings
Text size, Auto, Light, Dark and Sepia themes, dyslexia-friendly type (Atkinson Hyperlegible) and bionic emphasis.
Print numbering kept
An editor’s note explains how the edition keeps the print contents numbering.
Announced results
Twenty-six live regions announce simulation and self-check results.
Readout
The reader’s answers become a readout to send, copy, share or print.
Design system
Stellar’s navy and lime, with concentric rings from the cover as a motif.
- Navy
#0C2340Cover, headings and dark panels - Lime
#D4F562Actions and highlights - Ring scale
#DCE5F2 → #1E3A6FOrdered data and the cover rings - OK
#137A4CPositive results - Risk
#BA1B3AWarnings and watch-outs
Type
Newsreader for display, Inter Tight for reading and interface, and Atkinson Hyperlegible for the dyslexia-friendly mode.
Stances up front
Each play opens with Stellar’s stance and its watch-outs, as in the print.
Fictional by design
Brands, pages and assistants in the simulations are fictional and labelled.
How it’s built
Plays are self-contained, so a new play or an updated simulation can be added without touching the rest.
| Aspect | Approach |
|---|---|
| Delivery | One self-contained HTML file (about 295 KB) with its images embedded. No platform, plugin or reader app. |
| External requests | The web fonts, with system-font fallbacks, and a page-view analytics tag |
| Narration | The reader’s device voice through the Web Speech API, with word or sentence tracking |
| State | Answers, progress and settings saved in the reader’s browser |
PDF vs interactive edition
The PDF sets out the plays. The edition lets readers try each one.
| Question | Original PDF | Interactive edition |
|---|---|---|
| How should our AI sound? | Guidance on voice | A sandbox with pressure prompts |
| How do we get cited by AI? | A description of GEO | A page lab with three tactics |
| How much data do we need? | A stance | A data diet with value and trust |
| Where do we start? | A six-point audit | A self-check and a readout to send |
What we would measure
This edition is an independent demonstration with no audience data, and we don’t report results for it. For an agency, the self-check and readout are a qualified brief.
Audit scores
How readers rate themselves on the six points.
Priorities
Which plays and simulations readers spend time in.
Intent
Readouts sent to book the audit.
Where this approach fits
Agency playbooks gain the most when readers can try the method.
Agency thought leadership
Plays that become working simulations.
Brand guidelines for AI
Voice and guardrail sandboxes.
Audit-led lead generation
Self-checks that end in a readout to send.
FAQ
Common questions about this edition and projects like it.
Was this edition commissioned by Stellar?
No. It is an independent portfolio demonstration built from a playbook Stellar published. The playbook is Stellar’s; the editorial layer, interactions, design and build are Studio Vektor’s.
Do the simulations use a real AI model?
No. They are designed simulations with fictional brands and pages that show the trade-offs each play describes.
What would an edition like this cost?
Interactive eBooks start from $635, and the Advanced package with one defined quiz or scoring tool and interactive charts starts from $1,170. An edition at this depth, with twelve interactives, a self-check and a readout, goes beyond those packages and is quoted on scope after we see your document.
Can the readout go straight to our team?
Yes. In this edition readers can send it to book an audit; a commissioned edition can route it to your form or CRM with consent.
Publishing an agency playbook?
Let prospects try your method before they hire you.
Agency playbooks and audit-led guides work well. Send the PDF and we’ll show you how.