The Journey
How a simple question — "what if AI could actually help run a business?" — turned into a full-stack platform built in weeks. This is the honest story.
The spark
It started with frustration. We'd use ChatGPT for business advice — market analysis, strategy brainstorming, competitor research — and it was genuinely helpful. For about 20 minutes. Then you'd close the tab, come back the next day, and it had forgotten everything. Your market. Your budget. Your constraints. The decision you made yesterday. Gone.
We thought: what if the AI actually remembered? Not just the last few messages, but the full picture — your business context, every decision, every piece of research. What if each session built on the last? What if it could research your competitors, build your plan, track your tasks, and tell you when your assumptions were wrong?
So we started building. Not as a startup — as a portfolio project. Something to demonstrate that we could design, architect, and ship a complex AI-native product at speed. The goal was simple: push the boundaries and see what's possible.
Key realizations
Context is everything
Every AI tool we used reset to zero each session. You'd explain your business, get advice, close the tab, and start over next time. The realization was simple: if the AI could remember — really remember — your market, your constraints, your decisions, your progress — it would get exponentially more useful over time. That became the foundation.
A chatbot isn't enough
Conversations are great for exploration but terrible for accountability. You can't track progress in a chat log. You can't assign tasks. You can't enforce a decision framework. We needed structure — plans, tasks, decisions, workspaces — and the AI needed to operate within that structure, not just talk about it.
Governance matters more than intelligence
The smartest AI in the world is useless if you can't trust it. So we built guardrails: a three-tier decision system where the AI can handle routine operations autonomously, recommends tactical moves for your review, and blocks on strategic choices until you explicitly approve. And it can never mark its own work as done. Trust is earned through constraints.
Proactivity changes everything
Most AI tools sit idle until you type. We gave Solmyr the ability to set its own triggers — time-based reminders, condition checks, inactivity nudges. It follows up. It flags risks. It tells you when your plan has a hole. A good co-founder doesn't wait to be asked, and neither should an AI.
People wanted to use it together
We built it for solo founders. Then someone asked: "Can my co-founder use this too?" We refactored to multi-tenant — companies, roles, invitations, shared workspaces. What started as a single-player tool became collaborative infrastructure. The architecture had to grow, and it did.
The portfolio piece became the product
We set out to build a CV project — a demonstration of what's technically possible. We'd show it in interviews, explain the architecture, move on. But then people started signing up. They gave feedback. They asked for features. They came back. Somewhere along the way, the demo stopped being a demo.
The surprise
Here's the thing nobody expected, including us: it actually worked. Not just as a technical demo — as a useful tool. The compounding context made conversations dramatically better over time. The decision framework built genuine trust. The proactive triggers caught things we would have missed.
We built Solmyr to show what we could do. But people started using it for real businesses, real decisions, real plans. They gave feedback. They asked for features. They came back day after day.
Is it a product? Maybe. Is it a research project? Definitely. Is it a CV? Absolutely. But most importantly — it's proof that with modern AI tooling and solid engineering, two people can build something in weeks that would have taken a team of ten months to deliver a few years ago.
Try it. See what it does. That's the best pitch we have.