Productivity · News · 🇧🇷 Brazil
CNP Seguradora brings documents, calls and requests into one AI platform and claims analysis drops from weeks to hours
ÍSIS, built with GFT and AWS in four months, picks the AI model for each task by cost and complexity and already has more than 330 users at the insurer.
- Profession
- Technology Manager, Financial Analyst
- Tools
- ÍSIS, AWS
- Source
- segs.com.br ↗
- Country
- 🇧🇷 Brazil
- Report date
- November 3, 2026
- Verified on
- November 4, 2026
What they did
CNP Seguradora, part of the French group CNP Assurances, built ÍSIS with GFT Technologies and AWS, a platform that brings the processing of documents, calls and requests into a single entry point, with sign in and a record of every interaction. The insurer's departments use the platform through an API, and it picks on its own which AI model to use for each task, based on complexity and token cost. The platform connects to systems the company already had, such as Confluence and its data lake, and went from design to production in four months, running since November 2025.
How it went
According to GFT and the insurer, operating costs fell up to 10 times in specific processes, more than 4,000 hours of work a year were saved and claims analyses that took weeks now take hours, always with people supervising. In policy member changes the lead time fell from 7 days to 1; in accounts payable, average handling time went from 80 to 5 minutes; in purchasing, from 93 to 15 days. Today there are more than 330 active users and 33 use cases being rolled out.
Good to know
The text was sent by GFT itself, the project partner, and the gains apply to specific processes, not the whole company. The point they highlight is the architecture: a single auditable entry point and services reused in each new use case.
The prompt
I work in the [department] team at an insurer and the [process name] process currently takes [time] per request. Here are the steps: [list the steps, who does them and which documents come in]. Show where an AI assistant can help (reading documents, summarizing, checking data, building a checklist), where a person needs to decide, what sensitive data is exposed at each step and how I would measure time before and after to know whether it was worth it.
Replace what is in [brackets]. Prompt written by us from the account, not a literal copy of the source. Do not paste other people’s personal data (privacy laws such as LGPD and GDPR apply).
Summary in our own words. Results as stated in the original account; we have not tested or audited them. Read the source ↗ Found a mistake?
Similar cases
01Setting up a monthly bank reconciliation routine with AI
A finance professional describes how to use ChatGPT to cross-check a bank statement against the general ledger and flag differences.
02An AI chatbot in Teams cuts time spent logging hours by 72% at dti
The Brazilian tech company built Crafty, which answers HR questions and runs internal routines for nearly a thousand employees.
0360-year-old accounting firm uses AI to review financial statements and a bot for FGTS
Gomide Contabilidade (Brasília) is testing AI that cross-checks the trial balance, income statement and balance sheet; an RPA bot cut FGTS control from 10 to 15 days to about 36 minutes.
04AI agent checks invoices and frees up 660 hours a month at an accounting firm
Syhus, an accounting firm for startups, built an agent that compares thousands of invoices across two systems; a person reviews and approves the final spreadsheet.
05Accounting firm in Ji-Paraná posts 30,000 journal entries with AI and saves 298 hours
An academic study shows a firm in Rondônia, Brazil, using AI to read bank statements and reports and turn them into journal entries.
06A journalist lets Claude triage his Gmail every morning
A PCWorld writer set up a routine in Claude Cowork that sorts out what matters, archives the rest, files receipts and prepares reply drafts.
07Vibra says it saves 38,000 hours a year with AI in everyday work
The Brazilian fuel distributor brought Microsoft 365 Copilot to its business teams and set up an internal AI academy.
08AI agent contacts rental properties on behalf of apartment hunters
At Kaleno, renters describe what they want and, if they opt in, the AI messages properties, chases missing details and requests tours.