Fiakto

Inspiration

Finding a trustworthy plumber, electrician, technician, or other tradesperson is still stressful for millions of people. Customers often depend on informal referrals, cannot compare clear quotes, and have little protection when the scope changes. Skilled workers face the opposite problem: irregular demand, weak digital reputation, and time wasted on poorly described leads.

Fiakto was inspired by a simple idea: a person should be able to show a problem, understand what kind of help is needed, and safely hire a verified professional without already knowing the right technical words.

What it does

Fiakto turns photos, video, audio, and a short description of a real-world problem into a structured service request. Gemini analyzes the evidence, asks focused follow-up questions, identifies the likely trade, highlights urgent safety risks, and creates a non-binding reference range when enough information exists.

Verified local professionals receive matching opportunities and submit private quotes. The customer compares price, experience, verification badges, availability, and reputation. After a quote is accepted, payment remains inside the regulated payment flow, additional work requires an approved change order, and completion requires confirmation from both parties. If they disagree, the evidence is organized for human review rather than allowing an AI system to make an irreversible financial decision.

How Gemini is used

• Multimodal triage of images, video, audio, and text
• Structured classification and professional matching
• Context-aware follow-up questions
• Safety-risk detection and emergency guidance
• Neutral quote summaries without choosing for the customer
• Evidence organization and dispute summaries
• Auditable workflow assistance through function calling

Gemini assists with judgment-heavy information tasks, while deterministic rules control authorization and money movement.

How we are building it

Fiakto is being built as a responsive web application for Argentina, optimized for low- and mid-range phones. The planned architecture uses Next.js and TypeScript, Firebase Authentication, Firestore and Cloud Storage, Firebase App Hosting on Google Cloud, and the Gemini API with schema-validated structured output. Payment and identity providers are isolated behind server-side interfaces so regulated services can be integrated without coupling them to the product's core workflow.

Challenges

The main challenge is not simply matching supply and demand. It is creating trust between strangers while protecting addresses, identity data, evidence, and payments. Another challenge is using AI where it adds real value without presenting uncertain analysis as fact. Fiakto addresses both through explicit confidence, separate verification badges, private quotes, audit logs, double confirmation, and human dispute resolution.

What we learned

AI becomes more useful when it is connected to a complete real-world workflow. A generic assistant can suggest how to fix something, but it cannot independently provide a verified professional, a comparable quote, a protected transaction, an evidence trail, and accountable completion. The defensible value is the trusted system around the model.

What's next

The first pilot targets CABA and Greater Buenos Aires while keeping a nationwide data model. The validation goal is 20 verified professionals, 50 genuine requests, 10 paid and completed jobs, at least five independent paying customers, testimonials, auditable Gemini activity, and documented transaction economics.

Built With

• cloud-firestore
• cloud-storage
• firebase
• firebase-app-hosting
• firebase-authentication
• function-calling
• gemini-api
• google-cloud
• mercado-pago
• multimodal-ai
• next.js
• node.js
• react
• typescript
• zod