Visualizee · B2B · AI rendering for architects
Swap a complex editor for a chat assistant that writes the prompt for users
Revenue ~$150/month -> ~$10K/month in 6 months; pricing changed too
Workedproduct-activationonboarding
What they did
V1 was a node-based editor like ComfyUI (upload image -> node, connect nodes, write prompt, many parameters). After users said it was too hard, rebuilt as a ChatGPT-style chat: user uploads sketch/model and types a plain request; an AI assistant ('Vizzy') asks follow-up questions, presents a plan with selectable options (materials, lighting e.g. 'raw concrete', 'blue hour'), builds the prompt, and sends it to image-generation services. Also produces videos.
What happened
Revenue went from ~$150/month to ~$10K/month (8.2K MRR) within ~6 months; before, only a couple of sign-ups, after, users started rendering
- Stage
- Scale (2 years post-launch, stuck at ~$100-150/month)
This shipped together with other changes, so the result can't be credited to this alone.
In their words
Starter Story · I was stuck at $100 for 2 years until I made one change. Now my app makes $10K/month“I had a direct feedback from people, "Okay, it's too hard to understand for us." So I decided to pivot and to simplify the product to chat-based solution with the a- prompt assistant”
“That helped me go from $150 per month to nearly $10,000 per month.”
“Before, I had like only couple of sign ups, but after this change, people started rendering.”
“For 2 years, I blamed slow growth on marketing when the real problem was that my product was too hard to use.”
Related topics
Get tactics like this every Monday
New tactics from the week's founder interviews, each linked to where it was said.
More from this episode
- Visualizee: Switch from one-time payments to $35 or $80 monthly plans with a 7-day trial~300 subscribers, $8.2K MRR, 25% trial-to-paid; the app was also redesigned
- Visualizee: Aim AI image generation at architects and interior designers
- Visualizee: Keep going when competitors have customers; blame your execution, not demandStuck with it until a pivot found traction