Public tenders · government sales (Brazil)

Fourteen companies that stopped hunting for tenders by hand

A tender-monitoring system already sold to fourteen companies across four industries. The bot searches Brazil's national procurement portal for each client's own keywords, every tender it finds goes through an LLM that reads and summarizes it, and the result lands in a custom platform that also hosts a tender-analysis agent and a pricing agent.

Fourteen companies that stopped hunting for tenders by hand
n8nOpenAIPNCP public APIs

The project.

Companies that sell to the government in Brazil live on the PNCP, the national public procurement portal. It's where tenders are published, and where opportunities appear and disappear. Finding the ones that matter is heavy work: log in, filter, open each tender, read the full documentation and decide whether it fits what the company sells.

We built a system that does that digging on its own. It sweeps the PNCP for each client's keywords, and every client has their own, because the industry changes everything. A stationery supplier looks for one thing, an IT company for another, construction and food companies for others.

Today fourteen companies use it, across four industries. Each one has its own platform, with the relevant tenders already filtered and summarized, and with other agents on hand to help analyze the tender and the pricing.

The numbers.

14companies using it today
4industries served
3agents per platform
0manual portal searches
The challenge

Before, it was all manual. Someone logged into the portal, checked what had been published, picked what made sense for the business, opened and read each set of documents, and organized everything by hand. It eats the day, and it has to be redone constantly, because new tenders appear all the time.

The cost wasn't only the hours. It showed up in the opportunities that slipped by because nobody had time to look that day, or because the right tender was buried among dozens that had nothing to do with the business.

The solution

The workflow runs on n8n, pulling data straight from the PNCP's public APIs. The bot searches with the keywords configured for each client and brings back only what matches their industry.

Every tender found goes through an LLM that reads and summarizes it. The team gets it already understanding what it's about, instead of opening a huge PDF to find out whether it's worth their time.

The result lands in a custom platform, one per client. Inside, they also get other agents: one to analyze the tender in more depth and one to analyze pricing.

What changed.

  • Fourteen companies using the system, across four industries
  • Keyword search on the PNCP, configured by each client's industry
  • Every tender read and summarized by an LLM before it reaches a person
  • A dedicated platform per client, with tender-analysis and pricing agents
  • More tenders found, and far less manual digging

Built with.

  • n8n
  • OpenAI
  • PNCP public APIs

Free diagnostic

Your AI Diagnostic

Answer a few quick questions and get a diagnostic made for you, showing where AI should go first in your business.

  • Takes under 5 minutes
  • A reply from Camilly within 1 business day
  • 100% free
What does your business do?