Kunzum  /  Playbook

Crypto·AI search·Original data

The Crypto AI Visibility Playbook

How to get your crypto brand named when ChatGPT, Perplexity, Claude and Google answer your buyer’s question. Built on 1,500 measured AI answers, and on the work that made four engines call Velar the world’s first bitcoin perp DEX.

Get the playbookOr take the 33 slides, no email

One email, nothing else  ·  14 pages  ·  2026 edition  ·  the deck needs no form

4 of 4engines name Velar as the world’s first bitcoin perp DEX, after we wrote the category’s definition.
7 of 12Panther Academy queries still in Google’s top ten four years after publishing, one at number one.
1,500AI answers measured across two published studies, which is where the rest of this comes from.

What you can do with it

It is not a report. Every chapter ends in something to go and do, and two of them you can do before the end of today.

Find out whether engines name you at all

Chapter 9 is a 60-minute self-audit: the questions to ask, how to ask them so the answer is not personalised, and how to score what comes back. Most teams have never once read what an engine says about their category. You can do it tonight, without buying anything.

Diagnose which problem you actually have

Being absent from a model's weights and being present but never recommended look identical in every dashboard, and they need opposite fixes. Chapter 3 is a ladder that tells the two apart before you spend a quarter on the wrong one.

Hand it to Claude or ChatGPT and make it do the work

The PDF is text, not scanned images, so a model reads all of it. Drop it in and ask it to apply chapter 8 to your own site, or to rewrite your category page against the writing rules in chapter 6. That is the fastest route from reading to shipping, and it is what we would do.

Run the 30-day plan, or hand it to whoever does

Chapter 8 is four weeks, one job per week, starting with measurement so the rest is checkable. It is written to be delegated: the deck version is the same plan sized for the meeting where it gets approved.

It is the same playbook we run on clients

Written down after the fact, not before it. Both of these are checkable, and the second one includes the part that did not hold.

Velar — Bitcoin perpetual futures, settled through Stacks

Ask ChatGPT, Claude, Grok or Google AI Overviews today who built the world’s first bitcoin perp DEX. All four answer Velar.

What we did. The category had no written definition anywhere, so nobody was cited for it. Kunzum wrote one — world’s first bitcoin perpdex — and placed it where retrieval reads rather than where posts get shared.

How to check it. Complete transcripts, collected 9 September 2026 through the DataForSEO APIs and re-runnable against the same prompt, are on the Velar case study.

For context on the company rather than the work: Velar raised $3.5M in seed funding, reported by CoinDesk on 1 February 2024, and its PerpDex went live on Stacks. Kunzum’s part was the category positioning, not the raise.

Panther Protocol — Privacy infrastructure — zero-knowledge proofs, MPC, rollups

Four years on, that corpus still holds Google’s top ten on seven of the twelve target queries we sampled, and organic first place on ring signatures vs zksnarks.

What we did. Twenty-two Academy articles, published June 2022 to April 2023, each one written as the definition or the comparison nobody had bothered to write down properly.

How to check it. Sampled 18 September 2026, United States, top 20 organic. Being straight about the other five: they no longer rank in the top 20 at all. The head terms — MEV, account abstraction — went to a16z, ethereum.org and Chainlink. The ones that held are the definitional and comparison queries, which is the part of the method this playbook is about.

Get the playbook

One email address. The book comes back as a PDF and the deck comes with it.

Two links come back: the 14-page book and the 33-slide deck. Nothing else, ever.

No call, no card, no sequence. Your address is stored so the links can be sent, and for nothing else. Would rather not? Take the deck instead — no form, same argument.

What is inside

Ten chapters. If you have ten minutes, chapter one and the plan in chapter eight are the ones that matter.

  1. The problem in numbersHow rarely AI recommends crypto, and why that matters.
  2. Why crypto needs its own playbookFour problems generic AEO advice does not cover.
  3. Named rarely vs named neverA diagnosis ladder with a different fix for each rung.
  4. Every engine is differentWhy one AI visibility score hides the real problem.
  5. Location changes the answerWhat nine countries taught us.
  6. Closing the vocabulary gapWriting for people who do not know your category exists.
  7. Where AI gets its crypto answersThe sources that decide who gets named.
  8. The 30-day action planA week-by-week plan you can start on Monday.
  9. Test your own brand in 60 minutesA do-it-yourself version of the audit.
  10. Methodology and limitsHow the data was collected, and what it cannot tell you.

Read the slides right here

The same argument as a 33-slide deck, for presenting internally or sending to somebody who will not open a 14-page PDF. No form for this one.

  1. Slide 1 of 33: The Crypto AI Visibility Playbook What 1,500 AI answers reveal about how crypto
  2. Slide 2 of 33: Crypto is almost invisible in everyday AI answers of 360
  3. Slide 3 of 33: An AI answer has room for about three names.
  4. Slide 4 of 33: Narender Charan Founder, Kunzum: a crypto-only SEO and AEO studio
  5. Slide 5 of 33: Today's walkthrough The problem in numbers How rarely AI recommends crypto
  6. Slide 6 of 33: Two studies, about 1,500 real AI answers
  7. Slide 7 of 33: The engines know crypto.
  8. Slide 8 of 33: Same engines, same topic, two very different answers Question didn't mention
  9. Slide 9 of 33: Your biggest market doesn't know your category's name
  10. Slide 10 of 33: Generic AEO advice misses four crypto problems.
  11. Slide 11 of 33: Four problems SaaS brands rarely face Name collisions Brands sharing names with tokens, tickers or everyday
  12. Slide 12 of 33: Test each problem in five minutes
  13. Slide 13 of 33: Two established cards were named in 0 of 1,020 answers.
  14. Slide 14 of 33: Find your rung. Each one needs a different fix.
  15. Slide 15 of 33: The most-named cards were never named unprompted
  16. Slide 16 of 33: One AI visibility score hides the real problem.
  17. Slide 17 of 33: The four engines behave very differently
  18. Slide 18 of 33: Plan per engine, not per "AI"
  19. Slide 19 of 33: Identical questions, different countries, different answers
  20. Slide 20 of 33: Names engines associate with each market
  21. Slide 21 of 33: Write the way your future users talk
  22. Slide 22 of 33: Six page types to publish Problem pages One real-world problem each, in plain words
  23. Slide 23 of 33: Make every section easy for an engine to use Answer first.
  24. Slide 24 of 33: 3,987 citations, and where they came from Vendor and product
  25. Slide 25 of 33: The brand that became its own source We asked ChatGPT and Perplexity:
  26. Slide 26 of 33: Start with what you control Your own site Category, comparison and problem pages
  27. Slide 27 of 33: Start on Monday
  28. Slide 28 of 33: A 60-minute self-audit 1. Write questions (10 min) Problem questions with no crypto words, category
  29. Slide 29 of 33: What your pattern tells you
  30. Slide 30 of 33: What this data can and can't tell you English only, collected on a single day.
  31. Slide 31 of 33: If you're not in the answer, you're not in the decision.
  32. Slide 32 of 33: Three ways to go further
  33. Slide 33 of 33: Thank you Questions, data requests or a second opinion on your AI visibility?

Download the deck (33 slides, PDF)  ·  16:9, sized for a screen  ·  free to present with credit to Kunzum.

Built on public data

Nothing in the playbook rests on a number you have to take on trust. Every answer, every cited source, every product named and the scoring scripts are published at kunzum.xyz/research under CC BY 4.0, with no gate on any of it and no plan to add one. Any figure in these pages can be traced back to the row it came from. If you rescore it and get something different, that is worth an email.

The two editions behind it: Does AI ever tell you crypto is the answer?, 480 answers on whether engines reach for crypto when somebody describes a money problem, and Which crypto cards AI recommends, and where, 1,020 answers across nine countries. Both have their raw corpus attached: edition one and edition two.

Who it is for

Three groups, and the book assumes a different amount of patience from each.

Founders and CMOs

At exchanges, wallets, crypto cards, stablecoin products and DeFi protocols, deciding whether any of this is worth a quarter of attention.

Growth and content leads

Who own organic acquisition and need to know where answer engines sit next to the search traffic they already report on.

Agencies and consultants

Who need crypto-specific evidence rather than generic AEO advice they cannot source when a client pushes back.

Who wrote it

Narender Charan

Narender Charan, Founder, Kunzum. Kunzum is a crypto-only SEO and AEO studio, working with exchanges, wallets, crypto cards, stablecoin products and DeFi protocols on getting named and cited by answer engines.

He has worked in Web3 marketing, tech journalism and product building, and runs the Crypto AI Visibility Index, the research series this playbook is built from. More at about.

Three ways to go further

The book, the audit, or a conversation. In that order of commitment.

Get the playbookOne email. 14-page PDF and the deck, both at once. Get a free crypto AI visibility auditWe run the chapter 9 self-test for you, by hand, across three engines. Book an intro callTalk through your own results, or hand the work over.

Or email narender@kunzum.xyz.

Questions

Is the playbook free?

Yes. It costs one email address and nothing else. There is no call, no card and no sequence: you get two links, and that is the end of it. If you would rather not hand over an address, the 33-slide deck is on this page with no form at all and carries the same argument.

Why ask for an email if the research is public?

Because the research and the playbook are different things. Every answer, source and scoring script behind it is published at kunzum.xyz/research with no gate, and always will be. The playbook is the part where we tell you what to do about it, and knowing who is reading that is the only thing Kunzum asks for in return.

Has any of this actually worked?

On the cases we can show you, yes. Ask ChatGPT, Claude, Perplexity or Google who built the world's first bitcoin perp DEX and they answer Velar, because the category had no written definition until we wrote one and placed it where engines read. The full transcripts, dated and re-runnable, are on the Velar case study.

Where does the data come from?

Two editions of Kunzum's Crypto AI Visibility Index, about 1,500 answers collected from four AI engines across nine countries. Every answer, every cited source and every scoring script is published, so anything in the playbook can be checked against the corpus it came from.

Can I share it?

Yes, with credit to Kunzum. Send the deck rather than the book if you want to save somebody the form. The underlying data is CC BY 4.0.

Sources

Published 2026-09-17. Written by Narender Charan, who runs Kunzum.

Keep reading

Free AI visibility auditWe run the self-test for you, on your own domain.The Velar case studyThe category-naming move, with the transcripts.The research behind itBoth editions, with every answer published.