REPORT No. 001 — AI IN PRODUCTION

AI that answers, sells and automates — with proof.

Inovais ships 24/7 WhatsApp AI assistants, process automation and tailor-made platforms — with engineering, testing and evidence before the pretty slides. Try the demo: see how the briefing becomes a report.

client projects live
19
product bets in the portfolio
15
production-grade technical services
03

INVISIBLE BRIEFING — DEMO

name .............. you

02 ············

03 ············

Who you are and where to reach you.

Step 1 of 3

(01)  FOUNDATION

Five commitments before any product

Vision, mission, values, legacy and positioning, declared up front because they hold for everything that follows on this page: the services, the portfolio and every report issued.

ART. 01

Vision

Accounting firms drowning in manual obligations, spreadsheet-based calculation and tight deadlines, with the biggest tax reform in decades on the way: that is the niche Inovais observed and chose. Well-built AI agents give accountants back the time to advise.

ART. 02

Mission

We put AI agents in production that sell, serve and cut manual work, with proof: honest diagnosis, delivery in weeks, automated testing and LGPD compliance from day one. Everything already delivered is listed in Annex B, with own product kept separate from client work.

ART. 03

Values

03.1

Clarity

The problem well understood before the first line of code.

03.2

Antifragility

Every test, failure and fix makes the system stronger.

03.3

Resilience

Production that survives real clients: fallback, circuit breakers and governance live.

03.4

Union

Client and builder on the same side of the table.

ART. 04

Legacy

To be remembered as the firm that brought banking rigor to small-business AI, leaving every client more autonomous than they arrived: documented method, readable code and a live, measured agent.

ART. 05

Positioning of excellence

Delegate. Follow. Verify.

Client-proximity excellence: you speak directly with whoever builds it, from briefing to production.

APPLIES TO: SERVICES · PORTFOLIO · ANNEX B · BRIEFING

(02)  RAFAVIOLINISTA METHOD

From problem to proof

Every project at Inovais follows the Rafaviolinista Method, the same one taught in the Orbita course on building AI projects. This is the path that turns a problem into a delivery with proof.

01

The problem

Most AI projects do not die from a lack of models. They die at the first hard question: what problem does this solve, and where is your proof? The symptoms are familiar:

  • ·Tools chosen for hype, with no clear problem behind them
  • ·An agent in production with no metric, so nobody knows whether it improved
  • ·The eternal pilot that never becomes production

02

The thesis

An AI project that survives the first hard question is a project with a method: an understood problem, disciplined construction, validation with proof.

Tool choice, scope and proof for every delivery come from this sentence.

03

The method, in three phases

The same ones from the Orbita course:

  1. 1.

    Understand the problem

    Written hypothesis, audience, metric and a stop condition before the code.

  2. 2.

    Build with AI

    Context, RAG, agents and architecture with clear limits.

  3. 3.

    Validate with confidence

    Evaluation, security and governance before scaling.

It is the path of the course's 51 lessons: from product before the model to governance.

04

Doesn't it slow things down?

The expensive part is redoing. Without a written hypothesis, you iterate without knowing what to measure; without validation, the defect shows up at the client instead of in the test. Each phase exists to kill rework early.

05

The proof

Own products live, pilots with real clients and client projects in production, all with a public address in Annex B. None of them started as a loose bet: each followed this path.

06

Why Rafaviolinista

On the violin, nobody takes the stage without scales: daily technique before the repertoire. AI engineering is the same: fundamentals, training and base technique before the concerto. It is the discipline Inovais holds to in projects and teaches in the course.

07

The next step

Bring your problem and see the method applied to it, from diagnosis to proof.

(03)  ABOUT US

Artificial Intelligence that solves real problems

Inovais is an AI engineering firm founded by Rafael Novaes. We build Artificial Intelligence solutions for companies that want to automate, scale and sell more. Modern software engineering with the most advanced AI tech, and everything leaving here goes through testing and evidence before going live.

19
Client projects live
15
Product bets
03
Services in production

HOW WE START

Invisible Briefing

  1. 01

    Contact

    Who you are and where to reach you.

  2. 02

    Goal

    The problem that needs to disappear.

  3. 03

    Materials

    What you already have to start.

01

Cutting-edge AI

LangGraph, RAG and multi-agent architectures in solutions that truly understand your business.

02

Agile delivery

From concept to MVP in weeks, not months. A methodology focused on fast, measurable results.

03

Proven quality

Automated tests, prompt injection detection and LGPD compliance in every project shipped here.

04

Tailor-made

Each solution is designed for the client's specific context, from automotive to real estate.

(04)  SERVICES

How Inovais ships AI

SVC-01

01

(01)AI Chatbots

Multi-agent conversational assistants for WhatsApp and web. Sales, lead qualification and 24/7 support with LangGraph and RAG.

WhatsApp / Multi-agente / LangGraph / RAG

LIVE19 CLIENT PROJECTS LIVE15 PRODUCTS · 4 LIVE · 2 PILOTS WITH A REAL CLIENTLGPD IN THE FORM3-STEP BRIEFING41 ITEMS IN EXHIBIT B

(05)  PORTFOLIO

Projects that deliver results

Meet some of the projects Inovais has delivered, each focused on solving real problems with cutting-edge technology.

01

ClarIA

AI for accounting · Real client

A tax-reform agent for accounting firms: answers based on official sources, with deterministic evaluation and an audit trail on every answer. In pilot with a real client.

43 deterministic evaluations · HMAC audit trail on answers · Search over controlled official sources

TRY THE SIMULATION →

Agentes / Busca oficial / Eval determinístico / Auditoria HMAC / LGPD

02

GeraBoleto

Accounting SaaS · Real client

Invoicing and billing management for accounting firms, in production and evolving in vertical slices, with ClarIA integrated into the tax-query flow.

In production with a real client · Native ClarIA integration · Continuous evolution in slices

TRY THE SIMULATION →

Node.js / Neon Postgres / Fatias verticais / CI

03

CarInsight

Sales agent · Live

A WhatsApp sales assistant for used-car dealerships: a 7-node LangGraph flow, semantic search with RAG and over 1,028 automated tests, with prompt injection detection and LGPD compliance.

1,028+ automated tests · 7-node LangGraph flow · Prompt injection detection and LGPD

TRY THE SIMULATION →

TypeScript / LangGraph / RAG / WhatsApp API

04

FacilIAuto

B2B SaaS

Complete SaaS platform for dealerships with a recommendation engine, WhatsApp chatbot and admin panel. Multi-tenant architecture with 12 specialized AI agents.

12 AI agents · TCO calculator · Automatic inventory scraping

TRY THE SIMULATION →

Python / FastAPI / React / LangGraph / Docker

05

Inovais

Multi-Industry AI

Conversational sales assistant over WhatsApp with a multi-agent architecture adaptable to any industry. 7-node LangGraph workflow, semantic search with RAG and multi-LLM routing with circuit breaker.

1,028+ automated tests · Prompt injection detection · LGPD/GDPR compliance

TRY THE SIMULATION →

TypeScript / LangGraph / Multi-LLM / WhatsApp API

(06)  TECHNOLOGIES

Production-tested stack

Languages

TypeScript, Python, JavaScript

Frontend

React, Next.js, Tailwind CSS, Framer Motion, Chakra UI

Backend

Node.js, NestJS, FastAPI, Express

AI & LLMs

LangGraph, LangChain, OpenAI, Google ADK, Groq, RAG, pgvector

Data

PostgreSQL, Prisma, Sequelize, Redis, DuckDB, SQLite

Infra & DevOps

Docker, Railway, Vercel, GitHub Actions, Nginx

Testing

Vitest, Jest, Cypress, Playwright, pytest

Integrations

WhatsApp API, Stripe, Firecrawl, Google APIs

(07)  Contact

Let's build something amazing?

Tell us about your project and discover how AI can transform your business.

Talk to Inovais

Your data will be used to evaluate the project and reply to the contact, for up to 90 days. Don't send passwords or documents. Privacy.