Renalto

Making an AI quoting tool feel simple for renovation pros.

Role

Full-Stack AI Engineer

Engineering team

3 people, including CTO

Timeline

May 2025 – Jul 2026

Platforms

Web, iOS, Android

Renalto turns a renovation pro's description of a job into a professional quote. Its AI assistant, Rita, drafts it; the pro reviews and adjusts it. The product follows one path: get started → describe the job → review the quote.

I was one of three engineers at Station F, working directly with the CTO and founders. I designed and shipped onboarding, launched a partner quoting assistant and owned the referral program from scoping through measurement.

A guided path to the first quote.

I designed onboarding around getting a first quote started. A short tour walks the pro through setup, with details prefilled wherever Renalto already knows the answer. It collects the phone number needed for WhatsApp quoting along the way, so the pro finishes ready to describe a job.

Describe the job in your own words.

Type, dictate or attach the client's documents in one input. Dictation appears as the pro speaks, and Rita tidies the transcript before they send it. She asks only what's missing, then shows a summary to review before generating the quote. I built that review step to keep the pro in control of what gets written.

Change the quote. See what changed.

The quote reads trade by trade on a phone. To change it, the pro uses the same text, voice and document input they already know. Rita shows when she's working, then marks the lines she added or changed, so the pro can check the edit without rereading the whole quote.

Giving Rita a character

Rita used to be a static illustration. I rebuilt her in Rive as one state machine. She breathes, blinks and follows your cursor. She waves hello, and while idle she cycles through trade poses. In the quote editor she reacts to what's happening: a chat icon while you type an instruction, working while the AI updates the quote, and a check when it's done. Short messages appear in a speech bubble. Rita reached production; the playground below lets you try her states.

Decisions that shaped it

Build with the partner

Renalto's largest partner channel needed an update to its embedded quoting assistant. The partner's designer supplied the new flow; I built it, demonstrated it to their team and rolled it out behind a feature flag. It reached all of the partner's beta users in March 2026. I also added tracking so we could see how people used the new flow.

Test the growth idea before expanding it

I owned the referral program from scoping through launch. After comparing three approaches, I kept the first version small: attribution, credit rewards and tracking. Attribution survived the sign-in flow; rewards used our shared subscription handling and internal credits, with protection against duplicate grants. That covered different payment providers without a separate rewards platform. The program shipped, but uptake was limited. We reviewed the data and considered clearer placement and invitation copy.

Also shipped

• Shared interface. A design system and AI coding rules adopted by the team, a redesigned workspace, and shared editor and preview controls across desktop and mobile.
• Tenders. A guided flow over the existing DPGF parser, with confidence checks, price sources and document previews. Users could inspect the output and answer questions that reran the analysis.
• Getting quotes out the door. An Excel export with live formulas and a signature block for client approval, a PDF preview on mobile, one-click connection to the accounting tools Tolteck and Batigest, and VAT rules for Belgium, Luxembourg and France's overseas departments.
• Subscriptions and measurement. Most of the in-app Stripe subscription flow, plus onboarding analytics so the team could see where people dropped off.

Built with

Next.js, React, TypeScript, HeroUI, Tailwind CSS, Rive, Supabase and Stripe.