Context
It was 2023/2024. The AI market was exploding, but most consultants — financial, marketing, sales — had no idea how to apply these tools in their daily work. The company wanted to create AI solutions for this audience but had no product, only intention.
I ran the project with high autonomy: ideation, validation, roadmap, and part of the technical development in N8N/Dify alongside the tech team.
The Problem
I started with direct conversations with consultants. Before building anything, I tested two hypotheses:
- Free classes — consultants liked them but felt the content was too generic for their specific context
- Hands-on immersion building agents in N8N — some left with working flows, but the process was slow and required a lot of effort
The central insight came from these conversations: they didn't want to learn to build — they wanted to show up and use. They wanted ready-to-go agents from day one, with no learning curve.
The Solution
I repurposed the e-commerce infrastructure the company already had. I adapted the model to work like a community: each forum topic was a different AI agent. Users would pick an agent, have a conversation, and the history was preserved.
Decisions that defined the product:
- Existing infrastructure instead of building from scratch — dramatically reduced time-to-market. Not the ideal stack, but viable and fast.
- 50+ agents from day one — enough coverage for the most common consultant use cases. If the right agent exists, the user finds value on first use and comes back.
- Clear freemium model — free tier to capture interest, monthly or annual subscription to convert. Simple, no friction.
Results
In under two months: 1,000 signups and 60 paying subscribers. The product covered its costs in the first months of operation. The 80% DAU among subscribers was the most meaningful number — high engagement for any SaaS product, a clear signal the product was solving the problem.
The project validated that consultants wanted to consume AI, not learn to build it — an insight that shaped the company's strategic direction in the months that followed, even after the product was discontinued due to a strategic pivot.
What I'd do differently
Build a faster feedback loop with subscribers. With 80% DAU, there was a clear engagement signal, but I didn't systematize the qualitative learning. Monthly interviews with top users would have accelerated prioritization of the next agents.
Negotiate a minimum runway before launching. The product was discontinued due to a strategic decision, not lack of traction. An explicit agreement for at least 6 months of operation would have allowed time to test annual plan conversion at scale.
Document product learnings more thoroughly. When the project ended, much of the knowledge about consultant behavior stayed implicit — a structured repository would have preserved that asset.