Skip to content

tjsasakifln/SmartLic

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3,123 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

SmartLic — Public Procurement Intelligence for Brazil

Live Backend Tests Frontend Tests CodeQL

Live product: smartlic.tech · Status: Production · Paid trials · Stripe live

AI-powered intelligence layer on top of Brazil's $500B/year public procurement market. We crawl, classify and rank ~10,000 daily tenders so B2G suppliers find winnable contracts in minutes — not days.

For B2G suppliers, procurement consultancies and government sales teams who need to monitor 10,000+ daily tenders across 27 states without drowning in noise.


The Problem

Brazil's government buys $500B+/year in goods and services. Most suppliers still find out about opportunities through fragmented portals, outdated PDFs, regional newsletters and WhatsApp groups.

The official source — PNCP — publishes ~10,000 tenders per day across 5,000+ agencies. Without automated classification, a supplier in a specific sector has to manually scan hundreds of irrelevant listings to find the two or three that actually matter to them. The ones who do it systematically win more contracts. Most don't.

Why Now

Three things converged recently:

  • Lei 14.133/2021 replaced a 30-year-old procurement law and mandated that every federal contract flow through a single digital portal (PNCP) — the first time this data has ever been fully structured and accessible
  • PNCP's public API (launched 2023) made it possible to ingest the full stream programmatically for the first time
  • LLM inference costs dropped enough to make per-tender AI classification economically viable at scale — classifying a tender now costs a fraction of a cent

The window is early. Most govtech in Brazil is still manual or spreadsheet-based.

The Market

Annual government procurement (Brazil) $500B+/year
Procurement software market (Brazil, 2025) $298M → $746M by 2035
Agencies issuing tenders 5,000+ across 27 states
New tenders daily via PNCP ~10,000

What We've Built

A production procurement intelligence layer on top of PNCP, ComprasGov and PCP v2 — with a proprietary DataLake that didn't exist before.

DataLake 3.5M+ records (1.5M tenders + 2M historical contracts)
AI sector classification 20 sectors · precision ≥ 85% · recall ≥ 70%
Full-text search latency < 100ms at p95
Test suite 5,131+ passing · 0 failures
Organic reach 10,000+ programmatic SEO pages (ISR, Google-indexed)
Production since v0.5 · Stripe billing live · paid trials active

Traction

  • Paid trials: Active — Stripe billing live, 14-day trial, no credit card required
  • Organic reach: 10,000+ programmatic SEO pages indexed by Google, driving inbound from suppliers searching by sector and geography
  • Test coverage: 5,131+ passing tests · 0 failures · CI green on every commit
  • Data compounding: 3.5M+ records growing daily — 1.5M tenders (400-day window) + 2M+ historical contracts

For MRR, conversion rate and growth metrics, request the investor data room: tiago.sasaki@confenge.com.br

Architecture

flowchart LR
    subgraph Sources
        A[PNCP]
        B[ComprasGov]
        C[PCP v2]
    end

    subgraph ETL
        D[Daily ETL - ARQ + Redis]
    end

    subgraph DataLake
        F[Tenders - 1.5M rows]
        G[Contracts - 2M rows]
    end

    subgraph AI
        H[Sector Classifier - GPT-4.1-nano]
        I[Viability Scorer]
        J[AI Summaries]
    end

    subgraph API
        K[FastAPI - Search Pipeline]
    end

    subgraph Delivery
        L[Web App]
        M[Email Alerts]
        N[Excel and Sheets]
        O[SEO Pages]
    end

    A --> D
    B --> D
    C --> D
    D --> F
    D --> G
    F --> H
    F --> K
    G --> I
    H --> K
    I --> K
    J --> K
    K --> L
    K --> M
    K --> N
    K --> O
Loading

Data Moat

The DataLake is the core defensible asset. This dataset does not exist anywhere else in a clean, normalized, searchable form:

  • 1.5M+ tenders with 400-day rolling retention — full-text search in Portuguese, < 100ms at p95, and growing daily
  • 2M+ historical contracts — price benchmarking, supplier win-rate analysis, agency spending patterns by CNPJ
  • 20-sector AI classification using keyword density + GPT-4.1-nano arbiter for edge cases
  • Daily ETL across all 27 states and 6 procurement modalities, incremental refresh 3×/day
  • Programmatic SEO — 10,000+ pages indexed by Google, driving organic inbound from suppliers searching by sector and geography

Business Model

SaaS, 14-day free trial, no credit card required.

Plan BRL/mo USD/mo (approx)
Pro (monthly) R$ 397 ~$80
Pro (annual) R$ 297 ~$60
Consultoria (monthly) R$ 997 ~$200
Consultoria (annual) R$ 797 ~$160

Stack

FastAPI · Python 3.12 · Next.js 16 · Supabase (PostgreSQL 17) · Redis · ARQ · GPT-4.1-nano · Stripe · Railway

Team

Tiago Sasaki — Founder & CEO GitHub · tiago.sasaki@confenge.com.br · smartlic.tech

Founder of CONFENGE Avaliações e Inteligência Artificial LTDA. Built SmartLic from the PNCP crawler through Stripe billing as a solo technical founder — DataLake architecture, AI classification pipeline, 187 API endpoints, 10k+ programmatic SEO pages.

Backed by CONFENGE Avaliações e Inteligência Artificial LTDA — CNPJ 52.407.089/0001-09.

Where We're Going

Next 6 months: Scale paying base in 5 priority sectors (engineering, IT, healthcare, cleaning, food services). Launch supplier intelligence module — CNPJ-level win-rate analytics and competitor spend patterns.

12–24 months: Expand to municipal tier (5,500+ city governments). Launch consultancy/partner program. Become the default procurement OS for Brazilian B2G suppliers.

Vision: The intelligence layer for every B2G transaction in LATAM. Brazil first — then Mexico, Colombia, Chile (combined $400B+ annual procurement).

For Investors & Partners

Talking with seed-stage funds focused on LATAM B2B SaaS, govtech and vertical AI.

If you invest in govtech, public-sector SaaS, or vertical AI on regulated data:

tiago.sasaki@confenge.com.br

Data licensing and enterprise partnerships: same address.


Tags: govtech · b2g-saas · pncp · comprasgov · public-procurement · gpt-4-classification · brazil · latam


Technical Docs


Operational Status

Live SLO surface for SmartLic backend (target: p95 < 2s, /health/ready uptime > 99.5%, MTTR < 30min). Detailed runbook and error budget live in _reversa_sdd/operational-reliability-2026-05.md.


License

© 2024–2026 CONFENGE AVALIAÇÕES E INTELIGÊNCIA ARTIFICIAL LTDA — All rights reserved.

Proprietary software. Unauthorized use, copying, or distribution is prohibited.

Contact: tiago.sasaki@confenge.com.br

About

AI procurement intelligence for Brazil's $500B/year public tender market. DataLake 3.5M+ tenders & contracts, 20-sector GPT classification, <100ms search. Production SaaS · Lei 14.133 · PNCP.

Topics

Resources

License

Contributing

Security policy

Stars

Watchers

Forks

Packages

 
 
 

Contributors