David Prati

Hi, I’m David Prati, a product designer with 15 years of experience on complex products, using AI to build and ship what I design.

Selected work

ZefirFirst product hire2021–2026

AcquisitionPortfolioActions requisesTous les biensTraining pricingAdmin
Amaya R.
Portfolio
Adresse
Zone géographique
Account executive
Agent
À lister 5En vente 130Offre acceptée 4Mandat perdu
AdresseActionsSigné il y aPrix de reventeIntro callAgentMandat
30 Chemin de la Terrasse 92170 Chaville7 jours447 000 €22/06/23–Nadia Z.–Envoyé le 24/08
8 Avenue Salvador Allende 93000 Bobigny6 jours527 000 €À faire––
9 Rue Guynemer 92130 Issy-les-Moulineaux3 jours256 000 €À faire––
2 Rue Véronèse 75013 Paris1 jour390 000 €En attente d’infos client––
Proposer le mandat
Répartition de la commission
60/40
50/30/20
Agent
Nom de l’agentNadia Z.Nadia
Nadia Z.
L’agent sera notifié par SMS et vous serez alerté quand il aura accepté le mandat.
Proposer le mandatProposer le mandat
Mandat proposé !Vous serez prévenu sur Slack lorsque l’agent aura accepté le mandat, et lorsqu’il aura signé la délégation.
Souhaitez-vous prendre en charge ce mandat ?
Nous préparons votre offre initiale. Elle sera prête d’ici 24h, et nous reviendrons vers vous par email.
36 Rue Etienne Marey, 75020 Paris320 000 € (7 800 € / m²)Appartement - 52m²DPE : E
Accepter le mandat
Non, refuser le mandat
Mandat accepté ! Planifiez votre visite de courtoisie grâce aux infos ci-dessous.
3 personnes seront sur place
M. DispaVendeur
M. CasemiroLocataire
M. BayrouGardien
Planifier la visite
Zefirmaintenantun nouveau mandat est disponible
Nadia a accepté de gérer votre vente !Elle sera votre agent référent : elle présentera votre bien sous son meilleur jour et négociera avec les futurs acquéreurs.
À propos de Nadia
2 ansAvec Zefir8Biens gérés
Avez-vous trouvé cette information utile ?
ZefirmaintenantNadia a accepté de travailler sur votre bien !
Account managerAgentSeller
300 → 3.2K homes in portfolio without hiring.0 → 1 design of a real-estate marketplace · 3 products
25% → 50% of homes listed at the right price.Design · B2B and B2C pedagogy
0 → 50K monthly visitors on an AI buyer platform.Design and code · A search that says why each home fits
+15 pts conversion on lead acquisition.Design and code · Turned a lead form into an AI assistant

Before Zefir2011–2021

  • A back-office table of customers waiting for an answer, with the time left for each, next to the return instructions on a customer’s phone.A back-office table of customers waiting for an answer, with the time left for each, next to the return instructions on a customer’s phone.

    Life after the purchase made simpler for customers and refurbishers.Back Market · 2020

  • Four phone screens from the Alan app: choosing a type of care, the documents to send, the confirmation that the request was sent, and its tracking.Four phone screens from the Alan app: choosing a type of care, the documents to send, the confirmation that the request was sent, and its tracking.

    Reduce support contacts by automating reimbursements.Alan · 2021

  • A Deezer artist page with its discography, top tracks and playlists, next to an album page on a phone.A Deezer artist page with its discography, top tracks and playlists, next to an album page on a phone.

    Redesign profile screen and album pageDeezer · 2017

  • Two views of the Château de Versailles website: the home page with an aerial view of the gardens, and further down its exhibitions and masterpieces.Two views of the Château de Versailles website: the home page with an aerial view of the gardens, and further down its exhibitions and masterpieces.

    Lead the Château de Versailles website redesign.Château de Versailles · 2016

  • The City of Paris home page, with a search bar and shortcuts over a photo of the Seine, above a grid of practical services.The City of Paris home page, with a search bar and shortcuts over a photo of the Seine, above a grid of practical services.

    Lead the Paris website redesign.City of Paris · 2015

How I work

AI changed how I work as a designer: I check my own data and build my own prototypes, in code. Here’s where it comes in, step by step.

  1. A chat with Claude that queries BigQuery, next to the dashboard it built on listing creation adoption.A chat with Claude that queries BigQuery, next to the dashboard it built on listing creation adoption.

    I dig into the data myself with AI and MCP tools like BigQuery, to monitor and check intuitions.

  2. A chat with Claude that read support tickets and app reviews, next to a report of recurring words and the main problems ranked by mentions.A chat with Claude that read support tickets and app reviews, next to a report of recurring words and the main problems ranked by mentions.

    I go through user reviews and support tickets with AI to find what keeps coming up.

  3. A code editor with the prototype’s source and a Claude Code session in the terminal, next to a live preview of the new listing flow.A code editor with the prototype’s source and a Claude Code session in the terminal, next to a live preview of the new listing flow.

    I prototype with AI coding tools (Claude, Cursor, Lovable, v0) to align the team faster.

  4. A code repository with a contribution graph and a list of merged commits, the latest one deployed to production.A code repository with a contribution graph and a list of merged commits, the latest one deployed to production.

    I ship to production myself, from small improvements to whole prototypes.

Beyond design

I also build teams: I’ve hired or helped hire 15+ PMs and product designers across 5 companies, mentored them, and helped set how we worked: OKRs, roadmaps, prioritization and product framing.