Boost Your Web3 SEO with AI-Driven Strategies [guest on Hot Off The Press with Randy The News Guy]

Published: December 4, 2025
Victoria Olsina
Updated: April 22, 2026

Ask questions about this post:

Most people talk about AI for SEO like it is magic.

It is not.

I sat down with Randy The News Guy on his podcast: Hot Off The Press, to talk about it.

It is systems, prompts and a ridiculous amount of testing until the robots finally start doing the work for you. If you are in SEO, content or Web3 and you are tired of staring at NeuronWriter, Surfer or Ahrefs for hours, this is for you. You will see how I stack custom GPTs, APIs and automation to go from idea to SEO content that actually ranks, without sacrificing quality or brand voice.

Table of Contents

Watch the video: Hot Off The Press with Victoria Olsina

Who am I and why do I care about AI, SEO and Web3?

Quick version:
I am Victoria, Argentinian born, based mostly in the UK, and I have been in crypto and Web3 since 2018. I started in advertising and e‑commerce, worked at agencies like Saatchi & Saatchi, then moved into SEO for big brands and financial services, and later joined ConsenSys, the marketing arm around Ethereum and the company behind MetaMask.

Couple of relevant bits so you know where I am coming from:

I grew up in Argentina, where the banks literally shut the doors in 2001 and people lost access to their savings overnight. Trust in fiat money and institutions is low. That is why crypto makes sense to me.
I have done the “immigrant with zero contacts” thing twice, in New Zealand and the UK, so scrappy problem solving is baked into how I work.
I am lazy in a very specific way: I hate repetitive tasks, so I obsess over making robots do them for me.

Three years ago I went all in on AI and automation for SEO and Web3 content. Not as a shiny toy but as “how do I get rid of 70 percent of the boring work and keep the thinking part”.

How did I break into global marketing with no local network?

This matters because it is the same mindset you need for AI: show up, test, adapt, repeat.

When I landed in New Zealand with a working holiday visa, nobody knew my universities, my previous agencies or my clients. I spammed every job board until there were literally no more roles to apply for. Then I went manual:

Printed my CV
Pulled a list of 70+ agencies
Walked into reception at each one

Problem: receptionists block you and tell you to “send your CV by email”. Solution came when I saw someone walk in with flowers and get waved through. So I started turning up with flowers, asking for decision makers by name:

“Hi, can I speak to [Name], please?”

Nobody wants to interrupt a potential love story, so they either called the person down or let me in. Once I met them, I did a 30‑second pitch and left one flower plus my CV.

That is how I got hired by Saatchi & Saatchi: last agency on the list, last flower. The digital director literally said, “You arrived at the right time, we just had a budget increase.”

Same pattern later with ConsenSys in London:

I was already trading Ethereum out of boredom during an SEO role at Barclays
I saw a ConsenSys booth at a job fair
Then I followed up with targeted LinkedIn DMs to their marketing leadership explaining exactly how their SEO sucked and what I would fix

He said, “No budget yet, but I will call you.” Two months later, he did.

The point: persistence plus a specific offer beats generic “I want a job” every time. Exact same rule for AI: vague prompts and vague use cases get you garbage. Specific roles, specific workflows gets results.

What does AI‑powered SEO look like for Web3 and crypto?

When people hear “AI SEO for Web3” they usually think “spit out 50 generic blog posts about Bitcoin”. That is not what I do and it is not what works in 2024.

For my Web3 and crypto clients, typical activities look like this:

Keyword discovery and clustering
I use a custom GPT wired into the Keywords Everywhere API. Writers or SEOs type something like “Bitcoin loans” and specify:
- Search engine: google.com
- Country: US
- Currency: USD
- Page type: commercial landing page
The GPT calls the API, pulls real data, and returns:
- Top 10 competitors
- Related keywords
- Suggested keyword map
- Draft title and meta description
  No hallucinated volumes, because it is hitting an external API.
On‑page SEO briefs with NeuronWriter
For content scoring and semantic coverage I plug NeuronWriter’s API into another custom GPT:
- Create a query in NeuronWriter for a term like “Bitcoin loans”
- GPT pulls in: target terms, competitor content, headings, questions, recommended length
- All that comes back into ChatGPT as a structured brief rather than you clicking around the NeuronWriter UI
Brand‑correct Web3 copy
Web3 audiences spot AI fluff in two seconds. So I maintain a “marketing brain” GPT per client that:
- Has detailed knowledge base on the product, tone of voice, disclaimers, internal links, APRs etc
- Knows if the brand is “crypto‑native degen” or “regulated fintech for institutions”
- Can coordinate with the SEO GPT, the content GPT and writer personas
Result: when it writes a Bitcoin loans page, it mentions actual product details and flows, not generic “crypto can be volatile” college essay filler.

How do I make different custom GPTs work together?

Most people build one custom GPT and then copy paste between tools. That is fine for testing, but it gets messy quickly.

I prefer orchestration. Think of it as a Slack workspace where each colleague has a specific job:

Marketing Brain GPT
Knows everything about the client: products, tone, internal link rules, compliance notes. This is the “strategist”.
Keyword Research GPT
Connects to Keywords Everywhere. Jobs:
- Suggest topics
- Pull SERP competitors
- Build keyword maps with volume and intent
SEO Content Assistant GPT
Connects to NeuronWriter. Jobs:
- Create or pull queries
- Import term lists, SERP analysis and scores
- Turn that into a clear content brief
Writer persona GPTs
For one client we had:
- “Joey” who writes punchy, crypto‑native posts
- An academic writer for deep, educational content
I trained them by feeding real samples and breaking down:
- Typical openings
- Sentence length and patterns
- Favourite phrases and mental models

Inside ChatGPT, I can “@mention” another GPT in the same thread, so I chain them:

Call Keyword GPT to build a map
Call SEO Content Assistant to pull NeuronWriter data
Call Marketing Brain with all that context to draft a landing page
Call Writer Persona to rewrite as a blog post or Twitter thread

Zero copy‑pasting. Just like passing a task around a small team.

How do I build prompts that do not suck?

If your prompt is “write a blog post about crypto”, of course you get rubbish.

Good prompts include:

Role: “You are a content writer for a Web3 company that offers Bitcoin loans”
Audience: “Write for Bitcoin natives, already understand self‑custody and volatility”
Purpose: “Landing page for a new Bitcoin‑backed loan product”
Constraints: “1,000 words, include real use cases, no investment advice, mention LTV and APR ranges”
Tone: “Crypto‑native, confident, no hype, no emojis”

I also like using a Prompt Maker GPT that I built. You feed it your vague idea and it spits out a structured prompt with:

Context
Length
Audience
Style
Output format

So even if you are bad at prompting, the robot fixes your brief before sending it to another robot.

How do I connect AI, NeuronWriter and Google Sheets into real workflows?

Custom GPTs are nice, automations are better. For recurring content, I plug everything into Make.

Example stack for an SEO content pipeline:

Input: Google Sheet with:
- Target keyword
- Target country
- Page type (blog, product, category, brand page)
- Notes on angle or use cases
Make scenario:
- Sends keyword to NeuronWriter API, creates or pulls query
- Sends NeuronWriter data into an OpenAI Assistant that mimics my “marketing brain”
- That assistant:
  - Does SEO analysis
  - Calls research via Perplexity or similar for supporting sources
  - Drafts a strategist brief: outline, internal links, external links, CTAs
- A writer assistant then turns this into a full draft
- A compliance / reviewer assistant checks:
  - Claims, especially for finance or supplements
  - That we are not saying “best”, “guaranteed”, or making medical promises
  - Link rules and disclaimers
Output:
- Draft content back into the Sheet
- Score from NeuronWriter (for one client, target 52, draft hit 68)
- Ready for a human editor or straight to WordPress via another module

One client in the supplement space had over 500 categories and 15,000 products. Doing that level of on‑page work manually would be insane. The automation lets us treat categories, brand categories, products and blogs differently but off the same spreadsheet.

Does this AI SEO actually work in search?

Short answer, yes.

Example: a small Spanish site in crypto / finance content.

Before: ~8 clicks a day from organic search
After implementing the AI + NeuronWriter workflow at the start of the year: ~100 clicks a day
Data: straight from Google Search Console, not a tool screenshot

Volume is not huge because this is a niche editorial project, not a programmatic local lead gen monster. But the growth curve is clear and the pages are being indexed and ranking.

For my Latin American crypto podcast client, we publish around 10 high‑quality pieces per month using this system, not hundreds of spam pages. Editorial integrity stays, production time shrinks.

If you fire out thousands of Koala‑style articles without any brand training or review, you might get a quick win then watch a future Google update tear it down. Ask me how I know.

Quick wins

Use Keywords Everywhere + a simple custom GPT to get real keyword volumes directly inside ChatGPT when brainstorming content.
Plug NeuronWriter’s API into a GPT so writers can work from conversational briefs instead of logging into another SEO tool.
Build one “marketing brain” GPT per brand with product docs, tone of voice and link rules so every AI draft sounds like you.
Start a Google Sheet + Make scenario to auto‑generate briefs and drafts for one content type, for example blog posts only, before scaling to products or categories.
Train a reviewer / compliance assistant with your do‑not‑say rules so nothing embarrassing or illegal hits “Publish”.

Want to see how this would look for your brand?

If you are running SEO or content for a SaaS, Web3, fintech or e‑commerce brand and you want robots doing the grunt work while humans do the thinking, let us talk. Book a free 45‑minute strategy call here and we can map an AI + SEO workflow around your actual stack and constraints:
https://calendly.com/victoria_olsina/45min

Frequently Asked Questions

Do I need to be technical to build custom GPTs for SEO?

No. You need to be specific. Good custom GPTs come from clear roles, examples and boundaries, not from coding skills. If you can describe how you brief a junior SEO or writer, you can turn that into a solid GPT instruction set and start using AI in your keyword research and content workflows.

Which tools do you actually use daily for AI SEO?

For Web3 and SEO work my daily stack is: ChatGPT Plus for custom GPTs, Keywords Everywhere for quick keyword data, NeuronWriter for content scoring and semantic coverage, Make for automation, and sometimes Perplexity or Claude for deeper research and more natural writing. All of these plug together so you are not copy‑pasting between 10 tabs.

How do you stop AI content sounding like AI content?

You train it like a writer. I feed AI long samples of real articles, landing pages and tweets from the brand, then annotate openings, sentence patterns, favourite phrases and angles. That becomes a style guide inside the GPT plus a human editor at the end. Combined with NeuronWriter for structure and search intent, you get SEO content that fits the brand and does not scream “ChatGPT”.

Can this approach work outside crypto and Web3?

Yes. I already run similar stacks for e‑commerce supplements, finance and SaaS. The specifics change, for example compliance rules in health or loan products are harsher, but the pattern holds: one marketing brain, SEO data via API, research, writer personas and a reviewer. It is ideal for any niche where you need both accuracy and speed.

Is AI SEO safe from Google penalties?

The problem is not “AI content”, the problem is low quality, unreviewed content. Google has said repeatedly they care about usefulness and experience, not the tool you typed into. When you combine AI with real expertise, NeuronWriter‑style semantic coverage and a proper review layer, you create pages that answer queries properly and match search intent, which is what Google rewards over time.

How many articles do I need to publish to see results?

You do not need hundreds a month unless you are doing programmatic SEO. For editorial brands, I see strong results at 8–20 high‑intent pieces per month, provided the keyword research is solid and the content is tied to real products or offers. Start with your most commercial terms, build one workflow around them, then scale once you see traction in Search Console.