Saw an interesting job post from Wise yesterday that was hiring for an SEO specialist for a project called "owned sites".
This got me digging, and I think I've uncovered something big about their AEO/GEO strategy:
Wise seems to be building a network of sites beyond their main domain. I did some digging and found that out of 10 results for a transfer query on ChatGPT, 4 were owned by wise.com (including exaip.com), and 1 was from their partner remitfinder.
This strategy allows them to dominate more of the conversation in their industry, appearing in multiple contexts when users ask AI systems about international money transfers.
It's a smart move for AEO (AI Engine Optimization) and GEO (Generative Engine Optimization). By controlling more sites, they're increasing their chances of being cited by AI systems.
This approach goes beyond traditional SEO. It's about creating a wider content footprint to influence AI perceptions and responses.
The job posting specifically mentions an "Owned Sites" team, suggesting this is a deliberate, long-term strategy for Wise.
Also did a few more interesting experiment to uncover how ChatGPT forms an opinion from search & how to find the queries used by ChatGPT - happy to share more.
Anyone else here performing your own experiments to understand how AEO/GEO is gonna be different from SEO?
I’ve been digging into how ChatGPT pick their sources when answering questions.
What I’ve learned: traditional SEO isn’t enough anymore. It’s not just about ranking #1. It’s about getting used by AI when it pulls together answers.
A few notes:
- These AIs are firing off super long, weirdly specific queries. You can see them in your GSC if you filter right
- Content that ranks today might end up embedded in the next LLM update
- Sentiment matters. Reddit threads and off-site mentions shape how AI talks about you
- No ads (yet), which means organic visibility is still wide open... for now. Sam actually release a video today about his stance on ads.
I was learning to write Vue and use Nuxt for the first time yesterday.
Coming from the Next.js + React ecosystem it took a while to get use to the new syntax and also an entire ecosystem of plugins. Was trying out Nuxt to see how the developer experience is like and I can't figure what to make of it yet - can't say I like it for now.
Some interesting things I've noted:
- There's a little bit more of "magic" involved, ie components are automatically made available globally
Question: how do you navigate to the component code from where it's referenced? When writing in react components need to be explicitly imported and I can Ctrl+click to go to the component code easily.
- I still have trouble grasping environment variable. I've loaded some values in the .env file but then on the page it will resolve twice, once to undefined and once to the actual value.
Question: I must be doing something wrong here, how do you go around passing an env variable into a util class (I'm building a reusable component there).
Anyway, I took about 8 hours to get this blog template up and really hope to solve the issue with the env and share the template with the rest.
I found solutions out there were either full fledged cms which are cumbersome to setup and honestly distracting.
I was looking for something that was easy to work with like medium or notion, supports markdown syntax, and is fully headless because I like tinkering with the other frontend stuffs. If you have similar issues, check out wisp: https://wisp.blog/.
I recently conducted an experiment using AI agents to send personalized emails to nearly 1,000 developers with public blogs on GitHub. The goal was to introduce them to my CMS product, wisp, and the results were surprisingly.
The tech stack included LangChain's LangGraph for building AI agents and LangSmith for monitoring. I used Node.js and TypeScript to create a workflow with three AI agents:
1. WebInfoAgent: Gathered contextual information about recipients
2. EmailAgent: Crafted personalized emails based on the gathered data
3. JobReportAgent: Documented the process and tracked results
Key results:
- 20% click-through rate (180+ clicks in a single day)
- 10+ signups attributed to the campaign
- $0.20 cost per personalized email
- Median time of 26 seconds per email
- Multilingual capability (emails sent in Chinese, Korean, Spanish, and French)
The experiment highlighted both the potential and challenges of using AI for outreach. While the results were impressive, it made me think about ethical considerations down the line when this becomes more prevalent. I've since stopped the experiment to figure out what's next.
I'm eager to hear your thoughts, experiences, or concerns about using AI for business processes. How do you see this technology shaping the future of work and communication?
