Users try your AI once.
They don't come back.
104 cards for designing AI tools users trust, understand, and come back to — tuned to the moments where adoption actually breaks.
Featuring AI product tactics used by:
Microsoft
Canva
OpenAIMicrosoftCanvaMost AI adoption advice is hype.This deck is the opposite.
Generic AI playbooks tell you to “add chat” and “train on user data”. They don't help when your activation collapses on day two, your power users stop opening the product, or your enterprise pilot stalls in week three.
Adoption breaks at specific moments. The empty prompt. The plausible-but-wrong answer. The correction that never sticks. The connector that nobody grants permission to. Each has a known shape — and a known action card to apply.
Diagnose. Apply. Workshop.
The deck is organized around the workflow that actually changes outcomes: identify the adoption problem before choosing action cards, apply the cards that target the symptom, then run a workshop to turn the action card into product decisions.
Diagnose
Start with a Diagnostic that matches your symptom. The deck includes 12, each pointing to the action cards most likely to help.
Apply
80 action cards across 10 stacks — from product fit to retention. Each card carries the same structure: what it is, why it works, when, how, what to avoid, example, metrics, pair-with.
Workshop
12 workshops turn the action card into product decisions, copy, experiments, or specs. Each ships with a fillable deliverable template.
Every card, same shape.
Pickable in 30 seconds, applicable in an afternoon. Each action card carries enough structure to be argued about in a product review — and enough specificity to ship.
Evidence Trail
Show the user the path the AI took from input to output — sources, steps, and the claims each one supports.
Trust is built on traceability. When users can see which source backed which sentence, they stop having to verify the whole output and start verifying the ones that matter.
The AI synthesizes across sources, and users need to defend the output to someone else — a client, a regulator, a manager, themselves.
- Attach citations at the sentence or claim level, not the document level.
- Make every source clickable into a focused, highlighted view.
- Show retrieval steps as a collapsible trace.
- Flag claims with no supporting source as "unsupported."
Don't bury citations in a footnote panel. Don't show every retrieved chunk — show the ones the answer actually used.
The complete table of contents.
Stacks map to where AI adoption breaks — from the first prompt to team rollout.
Product Fit & Value
e.g. AI-Native Job to Be Done
Mental Models & Onboarding
e.g. Model of the Model
Prompting & Input Design
e.g. Clarifying Question Gate
Explainability & Confidence
e.g. Evidence Trail
Control, Correction & Feedback
e.g. Editable Memory
Failure Recovery & Safety
e.g. Prompt Injection Tripwire
Agent Permissions & Oversight
e.g. Agent Scope Card
Evaluation & Quality Loops
e.g. Golden Task Set
Privacy, Data & Governance
e.g. Delete and Forget
Retention & Workflow Habit
e.g. Return Hook
+ 12 Diagnostics · + 12 Workshops · 124 pages
Recognize the signal. Find the diagnostic.
The deck includes a full triage table — start from a user quote, a behaviour pattern, or a metric signal. Each maps to the diagnostic most likely to apply. Below: the verbatim patterns.
A 124-page working playbook, not a slide deck.
Designed to be printed, annotated, and put on the wall next to the roadmap. Hyperlinked between cards, indexed by symptom, stack, and metric.
Built as a playbook. Works as a skill file.
Same PDF, second life — paired with a SKILL.md companion file that turns it into a drop-in skill for any LLM. Attach both to Claude, ChatGPT, or any agent loop, and the deck becomes a reasoning surface your model can search by symptom, retrieve by card, and apply to your transcripts, your roadmap, your last user interview.
Drop it in a chat
Attach the deck to a Claude or ChatGPT thread. Paste a user quote, a support transcript, or a funnel drop — ask which diagnostic applies. The model answers in cards, not platitudes.
Mount it as a skill
Ships with a SKILL.md file you can drop into Claude Skills, a Custom GPT, or your agent's retrieval stack. The Diagnostic → Action → Workshop structure is addressed by name — agents reason in the same shapes the deck does.
Indexed for retrieval
Every card carries the same fields — symptom, stack, metric, pair-with. Models locate the right card without rereading the deck, and surface the action that matches the moment your team is in.
Works with Claude · ChatGPT · Gemini · Cursor · any tool that reads PDFs
Three ways to bring it in.
One-time purchase. Lifetime updates to v1.x. Org-wide sharing rights at every tier.
- · Full 104-card deck + 12 workshop templates
- · SKILL.md companion file for LLM & agent workflows
- · Triage tables, stack index, glossary
- · Org-wide sharing rights
- · Lifetime v1.x updates
- · Everything in Indie
- · Invoice & PO workflow
- · Priority email support
- · Everything in Startup
- · Custom license & legal terms
- · Team onboarding call
- · Volume & multi-team pricing
30-day money-back guarantee, no questions asked.
If the deck doesn't help you move the needle on AI adoption, email us within 30 days and we'll refund you in full.
Paulius Uza
CTO · AI & Fintech
“Most teams shipping AI today are not blocked on the model. They're blocked on the hundred small product decisions around it.”
This deck is the compressed version of those decisions: the failure modes I've watched teams hit again and again, paired with the specific action cards that move the metric. It exists because there is no shortage of AI hype — and a real shortage of concrete, applicable product work.
— Compiled from product reviews, postmortems, and adoption interviews across teams shipping AI in 2024–2026.
Things people ask before buying.
Q.Who is this for?+
Q.Is this a framework or a checklist?+
Q.Do I need to read it cover to cover?+
Q.What format is it?+
Q.Will it work for my AI product?+
Q.Can my team use it?+
Stop guessing.
12 diagnostics. 80 action cards. 12 workshops. One workflow that gets you from symptom to ship.
Get the deck — $249 →