Welcome back, Creators!
After a brief holiday break, we're diving straight into the deep end. Today we have loads of examples of non-coders building amazing stuff using AI.
We're also introducing a new section designed to cut through the hype called AI at work with real case studies of how businesses are using AI in reality.
Please take 2 minutes to tell us what you want to learn in 2026. We're giving a free 60-min consulting call to one respondent.
Don't forget we have upcoming meetups next week in Norwich and Manchester! Join us for great IRL conversations about vibe coding and AI agents:
Let's jump in...
On Your Radar
⭐ OpenAI's GPT-5.2 Pro Solves Three Erdős Problems
👉 Why it matters: This isn't just another AI milestone—it's a fundamental shift in how mathematical research happens. When AI can crack problems that stumped mathematicians for decades, we're looking at accelerated scientific discovery across every field. For builders, this signals that AI is moving beyond pattern matching into genuine reasoning and problem-solving at the highest levels. Read more
⭐ Ben Tossell wrote an incredible post about how he leverages AI agents
👉 Why it matters: This is proof that the "learn to code" barrier is crumbling. When someone can build personal sites, CLIs, and crypto trackers through AI collaboration rather than traditional programming, we're witnessing a new category of technical competency emerge. It's not about replacing developers—it's about expanding who gets to build and ship real projects. Read the full story
⭐ Claude Cowork Research Preview Launches
👉 Why it matters: Anthropic's new Cowork feature allows Claude to access and work with files in a chosen folder—it can read, edit, create files, and draft reports from notes. Built for non-coding tasks, Cowork represents a shift toward AI with genuine agency, making plans and executing them while keeping you informed. Currently available to Claude Max subscribers on macOS as a research preview. Learn more
AI At Work - How to transform an agency with AI
Andrew Bryce runs a 10-year-old company called 60 Seconds. What started as him running around with a camera asking businesses if they wanted videos has transformed into a leads agency that helps clients generate leads by creating personalised landing pages.
His small team now operates with the output of a large organisation. They send hundreds of thousands of personalised emails. They create custom landing pages for each prospect. They generate videos where an AI voice says the recipient's name correctly. And they track everything in a CRM that was entirely built by AI.
Andrew isn't a developer. But in the last nine months, he's shipped more software than most dev teams.
The video that knows your name
When a potential customer lands on one of Andrew's personalised landing pages, the system scrapes their details, renders a completely new video with an AI voiceover that pronounces their name, and has it ready by the time they hit play.
Not a video where someone holds up a whiteboard with "KIERAN" scrawled on it. An actual personalised video, generated on the fly.
Xnapper-2026-01-13-11.05.47
The system then tracks whether they watched 25%, 50%, or 95%. If someone only watches half, the next ad picks up where they left off.
The industry standard click-through rate for cold email is less than 1%. Andrew's approach crushes that.
The CRM that builds itself
Andrew built a full customer relationship management system from scratch. No developers. Just him, talking to AI.
After every sales call, AI automatically:
- Transcribes the conversation
- Extracts tasks and next steps
- Analyses sentiment (was the call positive or tense)
- Creates contact and company records
- Updates the pipeline
- Generates blog posts from interesting topics discussed
Sales reps don't log anything manually. The AI handles it all.
Xnapper-2026-01-13-11.10.29
How he actually builds this stuff
Andrew uses Claude Code (the command-line AI coding tool we covered recently) but with a system that makes it work at scale.
Working alongside actual developers, he created reusable templates and shortcuts. When he starts a new project, he types `/get-started` and the AI pulls a pre-built template, reads his project requirements, breaks it into phases, and starts building.
The phases act as "infinite memory". Even if AI loses context, it checks the phases file to see what's done and what's next.
Here's where it gets wild: Andrew runs a command that bypasses all the confirmation prompts, then goes to bed. He wakes up to find AI has completed hundreds of hours of development work overnight.
The team of AI agents
Andrew has built a collection of AI "agents", each with a specific job: backend architect, frontend expert, QA tester, copywriter, SEO specialist, and more.
One workflow generates entire blog posts automatically. It checks for queued topics, hands off to research, passes to the copywriter, finds images, optimises for SEO, and publishes. A 3,500-word blog post from a single command.
During our interview, Andrew showed a community platform he'd started building the Monday before. Slack-style chat, course creation, event management, monetisation tools. One week. No developers.
Xnapper-2026-01-13-11.08.55
How to start
Andrew's advice:
Start small. Pick something that takes you two hours a day. Spend 20-40 hours automating it, and you win back that time forever.
Break it into micro components. Don't build a CRM. First build a way to log contacts. Then add companies. Then deals. Layer by layer.
Keep your ideas somewhere. Note what could be automated. Tackle them in order of complexity, levelling up as you go.
The real insight
The most important thing Andrew said wasn't about AI or code.
Running his business for 10 years meant he'd worn every hat. He'd done sales, marketing, operations, finance. He'd trained humans to do all of it.
When AI arrived, he already knew exactly what needed automating. The AI just executed what he already knew needed doing.
That's the bit most people skip. AI is brilliant at following instructions. It's less good at deciding what instructions matter.
Andrew knew what mattered. AI did the rest.
Watch Andrew's talk from our recent meetup here.
{% if subscriber.tags contains "Create With Pro Member" %} {% else %} {% endif %}
VIDEO: From No-Code to AI Code - Julian's Journey
Think you need years of computer science to build real apps Julian proves otherwise. Called "lazy" as a kid, he skipped college, taught himself design, and built a 40-person agency using tools like Webflow.
Now he's taking it further with AI-powered coding. In this candid conversation, Julian breaks down why no-code builders are perfectly positioned for the AI coding revolution, sharing his journey from simple Webflow scripts to building full production web apps with Claude. His advice for skeptics Stop overthinking and start building.
Key takeaways:
- Why being "lazy" can actually make you more efficient at building
- How AI coding is just another abstraction layer (like no-code tools)
- Why no-code experience gives you a huge advantage in AI development
Final Thoughts
Min Choi
Claude Code started as Boris' s side project at Anthropic And in ~6 months it reportedly hit $1B ARR and changed how devs code. His best-practices thread is pure gold. Bookmark this 👇 https://t.co/VgIhBw9CS2
— @minchoi
From side project to billion-dollar ARR in six months The hype may very well be real. Probably time to read Boris's thread.
See you next week!
With good vibes ✨