12 AI Projects Built by the PwA Community
I Keep Saying You Can Build with AI. Here’s Proof.

You spend hours engineering context for one AI: training its memory, writing system prompts, building project files. Then you switch models and start all over.
Cuey is a free Chrome extension that unlocks that context for cross-model comparisons. When you compare ChatGPT, Claude, and Gemini in Cuey, each model answers with the full picture of your work.
Built for builders who can’t afford to start from scratch.
Try Portable Memory and claim 3 months of Pro free with code ATTITUDE.
You can build with AI.
I’ve been saying this almost every day since I started Product with Attitude last year, and I mean it.
The most common objection I hear is still:
“I’m not sure if I can.’’ You can.
“But I’m not technical.” You can still build with AI.
“But I’m a plumber.” Even better. You understand a niche most software people have never touched. That’s your advantage.
The important word is not AI.
It is build.
AI doesn't magically make every loose idea useful. It doesn't remove the need for judgment, feedback, taste, testing, or patience. It does lower the barrier between noticing a problem and making something that starts solving it. That's the whole promise of vibe coding: describe the thing, ship the thing, fix the thing.
A workflow counts. An automation counts. A prototype counts. An internal tool counts. A tiny product counts. A weird experiment counts if it teaches us something and gets closer to a real user.
Spend a little time inside Product with Attitude and you will see the pattern fast.
Some people were building long before they found this newsletter. Some needed a little push. Some posted one messy thing in the community chat and suddenly had beta testers, feedback, and a reason to keep going.
What we all have in common: we started before we felt ready.
Hey, I’m Karo Zieminski 🤗
AI Product Manager and builder.
I write Product with Attitude, an AI newsletter for thousands of subscribers developing critical AI literacy the only way it sticks: through practice.
We don’t just use AI. We build workflows, automations, and products with it, while studying how AI itself is built, positioned, and woven into our work.
If you’re new here, welcome! Here’s what you might have missed:
What’s Inside
What’s Being Built in the PwA Community Right Now. Twelve community tools, what each one does, and where to find it. Who’s looking for beta testers right now. And how to get your own project into the next builder showcase.
What “Build with AI” Means in 2026
Building with AI means using AI to create, improve, or operate a useful workflow, automation, tool, product, assistant, or system.
The final product can be ambitious or tiny. It just has to solve something real.
That might be a social image generator. A benchmarking system for AI agents. A career coaching loop. A Substack email parser. A private self-knowledge layer that follows us across tools.
None of those require the same stack.
They do require the same builder muscles: noticing friction, naming the job, testing with humans, and fixing what breaks.
AI lowers the coding barrier. It doesn’t lower the thinking bar.
The thinking was the valuable part anyway.
The Wednesday Build Board
Every Wednesday I post the same question in our community chat.
What are you building?
Products, workflows, automations, weird AI experiments. All fair game.
The Build Board exists because building in private is slow and lonely.
Posting what you’re making forces clarity, attracts feedback, and sometimes attracts your first users.
There’s a version of this happening on stages now. Founders showing unfinished work in person, before it’s polished. The Build Board is the async version, open every week, no plane ticket required.
What’s Being Built in the PwA Community Right Now
The 12 Tools at a Glance
Nyozzi:
Interactive Design Prompts You Can Copy, Remix, and Build From
Ileana is scaling Nyozzi, a growing collection of stunning interactive designs, each paired with a prompt you can copy and adapt for your own work.
In true builder spirit, Ileana is sharing these prompts free of charge.
Asaura AI:
ADHD Productivity AI for Getting Unstuck
Hodman Murad built Asaura AI, a productivity system for ADHD and neurodivergent minds. It helps break stuck tasks into one physical step at a time, so “I need to do this massive vague thing” becomes “do this next small action.”
Hodman did something many builders talk about and then avoid. She documented the whole process, every day, for 100 days on LinkedIn.
DraftKit:
The Engine for Creators Who Ship Together
Elena | AI Product Leader continues scaling draftkit.app.
Draftkit gives creators one place to handle the pitch, research, writing, edits, feedback, and approvals, which is exactly where most collaborations usually fall apart.
I had a pleasure of testing it while collaborating on an article with Dinah from Code Like A Girl and it works brilliantly.
TabVerified:
Independent Benchmarking for AI Agents
Rod Miller built tabverified.ai, independent verification for AI agents.
It runs 340+ benchmarks across 26 categories. That kind of systematic testing is exactly where AI compliance work is heading, especially around EU AI Act Article 12 logging requirements, traceability, and broader conformity assessment. If you need to prove how a system behaved, you need tructured evidence.
Latest results live at Rod’s Substack, and he’s offering beta access to anyone who wants to test their own agents.
WonderLead Career AI Mentor:
Turn Career Chaos Into a Growth Loop
Patricia Juarez @ AWS is building WonderLead Career AI Mentor, an AI career coaching system for tech professionals.
The tool helps you capture wins while they’re fresh, practice hard conversations before they happen, and turn those insights into a 30-day action plan. It’s built around the WonderLead Career Growth Loop: Reflection, Conversations, and Action Plan.
High School CS:
A Toolkit for Turning Students Into Technical Orchestrators
The Engineering Dad is building a framework to move high schoolers from “Syllabus Consumers” to “Technical Orchestrators.”
The High School CS Toolkit turns computer science from a list of assignments into a practical path toward building, shipping, and getting real-world experience.
A useful reminder that our tools, frameworks, and teaching materials don’t stop with us. They shape the next generation of builders.
High School CS is available on Gumroad.
The College CS Toolkit, a version for college students, is currently in beta. Feel free to reach out to The Engineering Dad if you’d like to test it.
Drop Kit:
Turn Any URL Into a Social-Ready Image
Paste a blog post, tweet, or product link. Get a polished social image instantly. Free, no AI, no signup. Built by Caroline Vrauwdeunt over a single weekend.
For creators who regularly share articles, launches, or product pages, this solves a real distribution problem: turning a link into something people want to click.
Drop-kit.com is on Product Hunt if you want to give it a boost.
I did.
IdeaGrit:
Validate Your Idea Before You Build the Wrong Thing
IdeaGrit helps founders pressure-test business ideas before building the wrong thing.
Choose SaaS, local business, consulting, or education, and get an opinionated validation report with risks, weak spots, next steps, an actionable roadmap, and a pre-mortem based on 6 similar failed products. See how your validation score changes over time.
It’s also live on Product Hunt if you want to give it a boost. I did :)
Subhook:
Turn Substack Emails Into Automated Workflows
Daniel Rusnok opened beta access to Subhook.
Some of you remember Daniel’s Drippery tool. It works by forwarding the “New Substack Subscriber” email to a dedicated address, where a cloud worker parses the body, pulls out the subscriber’s email, and drops them into a Drippery welcome series.
Subhook takes that further. With Custom Mailhooks, you can respond to any automated Substack email. A visual editor helps you pick exactly what to extract, then send it to an API endpoint of your choice.
Subhook is in early access for subscribers of Digital Craft Product and Product with Attitude. If you want to wire your Substack emails into something useful, give it a try and send Daniel your feedback.
selfActual:
Portable Self-Knowledge
Jeremy is building a self-knowledge layer for AI: a private, portable vault that helps AI tools understand how you work, think, decide, and collaborate.
Instead of forcing you to re-explain yourself every time you move between Claude, ChatGPT, Gemini, or another AI surface, selfActual stores your operator model in a W3C Solid vault and connects it to AI agents through MCP. The point is not more memory for one model. The point is continuity that belongs to you.
Donna MCP:
Connect Your Substack to Your AI Stack
CP is building Donna MCP, a Substack MCP that connects to any AI tool with one click, scans your Substack, finds the gaps, and helps you fix them.
CP has been recruiting beta testers through the PwA community. If a one-click connection between your Substack and your AI stack sounds useful, let CP know you want in.
StackContacts:
Understand Which Readers Are Ready to Pay
Finn Tropy is scaling StackContacts, and I can’t stop recommending it. It’s a local CRM and audience intelligence tool for Substack writers.
It brings subscriber activity, Gumroad purchases, Kit contacts, clicks, comments, and engagement history into one place, so creators can stop guessing which readers are most engaged, which posts drive sales, and who might be close to upgrading.
For writers selling products, subscriptions, or services through Substack, StackContacts solves a very real problem: knowing who your warmest readers are, what they care about, and when it might be worth starting a human conversation.
Tools I Built for Product with Attitude Members
All of the tools I built for Product with Attitude (past, present, and in progress) are included in your Premium Membership, alongside other benefits.
You can use them from day one.
LinkSwap: Trust-based backlink swaps with other writers.
Vault: AI workflows, coding prompts, automation templates shared by the community members.
StackShelf: List your own products, alongside other builders in the community. Most of the builders listed in this article promote their products through StackShelf.
The Pattern: Domain Knowledge Is the Moat
Look across the list and the pattern gets obvious.
These builders are not all building the same thing. They are not using the same tools. They are not aiming at the same buyer.
But they are doing the same kind of work.
They are taking a specific pain they understand and making it easier to handle.
That is the real “you can build with AI” lesson.
A designer can build better design prompts because they understand how visual taste develops.
A neurodivergent builder can design a better unstuck workflow because they understand the shape of task paralysis.
A creator can build collaboration software because they have lived through feedback chaos.
A Substack writer can build audience intelligence because they know how hard it is to connect reader behavior to revenue.
This is the same shift the trade press is finally naming: solo founders are now doing the work of entire teams, because the build cost collapsed and the domain knowledge didn’t (Fortune).
It’s also why non-technical builders shouldn’t apologize for being non-technical. The technical part is more accessible than it used to be. The hard-earned domain knowledge is not.
How to Start Building with AI This Week
Don’t start with a product idea. Start with a repeated pain.
Choose one annoying workflow. Pick something you already do more than twice a week. If it is not repeated, it is harder to improve.
Name the input and output. Good AI workflows usually have a clear before and after. Email in, CRM record out. URL in, social image out. Notes in, 30-day plan out.
Build the smallest useful version. The first version should make one person’s life easier. That person can be you.
Show it before you polish it. Private perfection is a trap. Share the rough version where real people can react.
Write down what broke. Fix it.
That’s building.
Not the glamorous version, the useful one.
How to Get on the Next Build Board
The next builder showcase goes out in about 6 weeks, and the door is open.
If you’re building a product, a workflow, an automation, or a weird AI experiment, post it in the Wednesday Build Board thread.
Looking for beta testers or want to offer something PwA-exclusive? Even better. The community is full of people who will try your thing, break it, and tell you what’s wrong before your first real user does.
FAQ: Building with AI Without Being Technical
Can a non-technical person build with AI?
Yes. A non-technical person can build with AI by starting with a narrow workflow, using AI to create prompts, prototypes, automations, or simple tools, and testing the result with real users. The advantage comes from domain knowledge, not from knowing every layer of the stack.
What can I build with AI besides chatbots?
You can build research workflows, content systems, audience intelligence tools, social asset generators, validation reports, internal dashboards, career coaching loops, education resources, Substack automations, and agent benchmarking systems. A chatbot is only one interface. It is not the whole category.
What is the safest first AI project for beginners?
The safest first AI project is a private workflow with low-risk data and a clear output. Start with something like summarizing research, generating content briefs, organizing notes, creating social assets, or drafting a repeatable checklist. Avoid sensitive personal data, payments, legal decisions, medical claims, and anything users will rely on without review.
Do I need to learn code if I use AI app builders?
You don’t need to become a professional developer to start building with AI. But learning basic technical concepts helps. Inputs, outputs, APIs, databases, permissions, logging, and testing matter even when AI writes the code. The goal is not to know everything. The goal is to understand enough to ask better questions and notice when something is wrong.
How do I know if an AI-built product is ready for users?
An AI-built product is ready for early users when the core workflow works repeatedly, the output is useful enough to test, the risks are understood, and a human can review or override important decisions. It doesn’t need to be perfect. It does need to be honest about what it can and cannot do.
Where do I even start?
Start with the Vibe Coding Hub. It's the map: the tools, the workflows, and the order to learn them in, so you don't waste your first week picking software instead of building. Then pick one repeated pain from your own week and make the smallest version that helps one person. That person can be you.
WHY SUBSCRIBE ・YOUR BENEFITS・ TOOLS I BUILT・CLAUDE HUB・PERPLEXITY HUB ・VIBE CODING HUB
Building in public works.
So tell me what you’re making right now. I’d love to feature it 🤗




































What a beautiful community to be part of and thank you Karo for bringing us together and being such a powerful force behind us builders. So much great stuff being built, 💪😃
Wow! Thank you so much @Karo! Reading everyone's work now, but deeply honoured you included me and SelfActual.ai!!! :D Demos incoming!