Kia ora {{ First name | friend }},
We’re exploring the light and shadow of technology, with a focus on efficient AI workflows to free up your time.
This week’s edition features more ChatGPT Atlas, a Rasta monkey AI influencer, some updates from our mate Elon, and interactive video models.
I trust it’s useful. If it is, let me know. If it’s not, reply to this email and tell me why :)
Featured Workflow: ChatGPT Atlas Automation Builder
Ok so ChatGPT Atlas was the first feature in last email. I’m putting it here again cause I’ve been blown away in the last week by it.
Basically, you can now build automations using any software with natural language prompts. No matter if you’re using Zapier, Shopify, Airtable, Make.com, Manychat, whatever. Any manual task on any web browser app can now be automated with plain English prompts, using ChatGPT Atlas.
Here’s an example of what I mean
I needed an auto-email to go to customers when they brought a product on Shopify, with product info
I opened ChatGPT Atlas, navigated to Shopify, and logged in like I would on any other browser.
I put in agent mode and said this prompt: “in shopify, set up a new automation so if someone orders {PRODUCT}, they receive the below #email content# automatically. Include formatting and attachments (ask for attachments when ready).
##email content##
{this is where I put my actual info I wanted to go to the customer}.I waited like 10 minutes and it was basically done.
Some field notes
It’s still slower than I could do manually, but it’s monotonous work, and meant I could focus on other things while it worked
The email still needed some work (formatting, language)
I have serious concerns about the privacy of ChatGPT being able to log in and access all my things, and act on behalf of me, so I proceed with caution, and recommend you do the same. Don’t go giving it all your passwords like it asks you to.
Sorry Windows users, it’s only available for Mac at the moment, but hear it’s coming soon for Windows,
Gif by theitcrowd on Giphy
Prompt: Discover Customer Pain Points Quickly with AI
Step 2: Research Pain Points
Many of us are entrepreneurs trying to provide solutions for customers across various industries. Here’s a prompt idea for going deeper into that. It’s not a silver bullet (you’ll still need actual research, from your customers, in your area, etc), but it’s a good palce to get started.
I can imagine this would be quite useful for:
Somebody looking to start a new business/project in an industry that they’re still getting to know
Somebody with existing customer data to upload (website traffic, social followers, email list, survey responses etc) and combine that for a deeper analysis.
## Role: Act like an experienced Market Research Strategist with 10 years of experience
## Task: Identify and prioritize high-value, unsolved problems in [industry or niche] for [target user or ICP] using deep research across recent public conversations. Deliver a ranked table of problems and a short insights summary that highlights the best opportunities to build a business around.
## Context: I want to start a business and need evidence-backed problems that are urgent, frequent, and show willingness to pay. The goal is to uncover pain points with high potential for a viable product or service, ideally untapped or under-served, so I can reduce risk before building.
## Structure:
1. Run deep research across Reddit, Quora, X, niche forums, product reviews, and recent articles.
2. Extract 10 distinct pain statements with a short user quote and a source link for each.
3. Score each item on: urgency, frequency, willingness to pay signals, and evidence quality.
4. Return a ranked table with columns: Problem | Example quote | Source link | Urgency | Frequency | Pay intent | Evidence quality | Notes.
5. Write a 150-word insights summary that explains the top 3 opportunities and why they stand out. 6. List follow-up questions needed to validate buyer, budget, workflow, and success criteria.
## Tone: Analytical, data-driven, founder-friendlyStep 2: Estimate Real Demand
Once you’ve identified several pain points, the next step is to see if people actually care about them.
Demand validation helps you avoid solving something that no one is actively searching for or discussing.
Prompt (with deep research mode):
## Role: You are an experienced Market Demand Analyst
## Task:
Evaluate market demand for the following problems:
[insert the list of problems discovered in Step 1].
Use search trend data, discussion frequency, and social signals to estimate current and projected interest. Deliver both quantitative and qualitative insights that show whether each problem has real traction.
## Context:
I am testing multiple startup problem ideas and want to know which ones show real demand before committing to product development. I need to identify which pain points people are already searching for, talking about, or spending money to fix.
## Structure:
1. For each listed problem, research indicators of interest using deep research or browsing: Google Trends, Reddit mentions, X posts, Quora activity, YouTube topics, Product Hunt discussions.
2. Collect approximate metrics such as monthly search volume ranges, social chatter frequency, and trend direction (rising, steady, or declining).
3. Summarize the findings in a table:
Problem | Search volume range | Social mentions | Trend direction | Momentum score | Evidence snippet | Source link.
4. Provide a 200-word insights summary explaining which problems show strong growth potential and which appear saturated or stagnant.
5. Rank all problems from strongest to weakest based on momentum, discussion energy, and clear purchase intent.
## Tone: Strategic, data-driven, founder-focused.From here, it’s too nuanced to continue. I’d suggest similar structured prompts with the following
Review your product against the data
Assess competition
Design a marketing strategy
I’m keen to hear how this lands for you, let me know if you have a go with it :)
Here’s a few interesting technology updates:
AI Influencers are starting to pop up, such as Jahmunkey, a Jamaican Rastafari monkey that brands pay to make ads. I find this one particularly amusing. Here’s a video promoting an ayahuasca retreat in Costa Rica. Very strange times we’re living in.
Elon Musk’s xAI has launched Grokipedia, a bold attempt to rebuild Wikipedia from scratch using its own AI model, Grok. The platform positions itself as an uneditable, AI-generated encyclopedia, designed to serve as a factual backbone for both humans and future AI systems. Hmm.
Microsoft and Claude have teamed up for Claude for Excel. Claude listens carefully, follows instructions precisely, and thinks through complex problems. I don’t use Excel (Google Sheets instead) but I assume this will be huge for Microsoft users. Learn more and join the waitlist.
Odyssey-2 is a new interactive AI video model, where the video changes as you prompt it. I had a jam on it, (see the sped-up timelapse below)
. Pretty trippy. Not sure where this will go but it’s cool to see video models getting faster. Try for yourself.
Bill gates on AI: “the biggest technical thing of my lifetime,” but warns that many current bets will fizzle. Think stranded chip orders, overbuilt data centers, and inflated energy costs, but still resulting in lasting long-term value. Similar to the dot com boom, he reckons AI in 2025-27 will be boom-or-bust for a lot of small, medium and large orgs.
Amazon cut 14,000 jobs amid AI push, while putting a lot of money into new data centres. Watch here.
Here’s the sped up version of the Odyssey-2 interactive video, you can see in the top left where i’m prompting, as the image changes. Again, super weird.
{{ First name | friend }}, thanks for showing up again. This is getting more fun and more weird simultaneously.
If you found this useful, please reply and let me know what you enjoyed, and share this with a mate who you think could benefit from it. :)
Stay human,
Billy