Context
Qualaro is an AI-powered B2B lead generation platform, aiming at the space Apollo occupies but with enrichment, outreach and conversational AI in a single workflow. When I came in they had finished UI designs in Figma and had early beta users waiting. What they did not have was a product.
This is also the engagement that became a seat rather than a project. I stayed on as fractional CTO through to April 2026.
The problem
A prospecting platform is deceptively hard to build quickly, because almost none of the value is in the interface. It sits in the enrichment sources, the AI layer, the credit accounting, and the integrations, all of which have to work before a single user sees anything useful.
The constraint was time. Beta users who have been promised a product and then wait six months are not beta users any more, they are a lapsed list.
What I decided
I split the stack rather than picking a side. Bubble on the front end, because the Figma designs were finished and matching them pixel for pixel was faster there than anywhere else. Supabase underneath for data, auth and edge functions, because a lead generation platform is a data product and putting that data in a no-code database would have created a ceiling within months.
That hybrid is the whole decision. It bought a twelve-week MVP without mortgaging the backend.
The second call was the credit system, which I insisted on building into the MVP rather than deferring. Every AI feature in the product costs real money per call, across OpenAI, voice, and enrichment providers. A platform that cannot meter its own usage is a platform that cannot price itself, and retrofitting metering after launch means reconciling usage you never recorded.
What I traded away was front-end control. Bubble has a ceiling, and for a SaaS product intending to reach enterprise customers that ceiling will eventually be reached. The right way to read that is as a deliberately staged decision: get to market and to investors on the hybrid, then replace the front end when there is revenue to justify it. The wrong way is to pretend the ceiling is not there.
What shipped
Lead discovery with filtered search, AI outreach covering generated email and voice calls with live transcription and sentiment tracking, enrichment through Clearbit and People Data Labs, a credit system tied to plans and feature access, a real-time analytics dashboard, and an admin console for users, credits and plan configuration.
The result
A functional MVP in twelve weeks, with manual prospecting effort reduced by up to 70% for users, and an architecture modular enough to add AI and CRM integrations without structural work. The platform was investor-ready at the end of it, which for a company at that stage was the outcome that actually mattered.