AI Value Discovery
“Where should we start?”
Two weeks on site with your team. We map the priority workflows, find the AI opportunities in them and price each one. You get a prioritized list and a quantified value case before anything gets built.
Applied AI in São Paulo
Applied AI for companies in São Paulo, from the first question to production
Extrace is an applied AI company based in São Paulo and Rio de Janeiro. We work with mid-sized and large Brazilian companies to find where artificial intelligence creates measurable value, build it into the systems they already run, and leave the team able to operate it.
Who we work with
AI reached the boardroom.
The front line is still waiting.
The companies we work with are mid-sized and large, usually led by a strong owner, and often have a private equity fund in the capital. Revenue of $20 million and up, decisions concentrated in the CEO or CFO, professionalization under way, and no structured data team yet.
In these companies AI has already arrived, but only at the top. A director uses ChatGPT, the CFO builds a dashboard over a weekend, and the teams keep running on spreadsheets and WhatsApp. When someone inside builds something useful, it does not scale: no security, no integration, no access control.
That gap is where we work. We take AI from the executive layer into daily work: we enter the stack the company already has, build with its team, and it keeps running after we step back.
What we do
Four offers, each answering the question that usually comes next.
AI Delivery
“How do we actually build?”
We design and build the agents, automations and internal tools that solve the problem, integrated with your data and systems. Nothing counts as done until it runs in production and we have measured what changed.
Agentic Transformation Office
“How do we scale beyond projects?”
Guardrails, policies, training and a shared knowledge base, so adoption becomes part of how the company operates instead of a stack of pilots. A continuous engagement, not a project.
AI Literacy
“How does our team learn to use it?”
Hands-on sessions for leadership and frontline teams, on their own work: what AI does well, where it fails and how to use it safely. The team leaves using AI every day, not just knowing about it.
Working from São Paulo
Based in São Paulo and Rio de Janeiro.
Working across Brazil.
Extrace works from São Paulo and Rio de Janeiro. Most of the companies we describe above, and most of the funds that back them, are headquartered in São Paulo, so a good part of our discovery work happens there, on site from the first week. A discovery starts inside the client's business, because the opportunities that matter show up on the floor, not in the org chart. For companies elsewhere in Brazil we travel for the discovery and run delivery together with the team, remotely or in person.
We work in Portuguese with your teams and in English when the board, the fund or the documentation asks for it. The founding team comes from strategy consulting, open finance infrastructure and product engineering.
Case · Healthcare
One healthcare client, from kickoff to running in daily work.
01How fast do we see something working?
2 weeks
from kickoff to a working proof of concept, not a slide about one
02Is it worth the money?
$1M/yr
of opportunity found and priced, most of it in documents still made by hand
03Does it actually reach day-to-day work?
$14k/yr
from the first automation, now live and saving each person about an hour a day
What does an applied AI company actually do?
We find where AI creates measurable economic value in how a company already operates, put a number on each opportunity, build the solution (agents, automations, internal tools) into the systems the company runs on, and stay until it works in production. The capability stays with the client's team when we leave. The difference from a traditional consultancy is that we build; the difference from a software house is that we start from the business case, not from a specification.
How long does an AI Value Discovery take, and what comes out of it?
Two weeks. We spend the first part on site, mapping the priority workflows, then quantify each AI opportunity in dollars. The output is a prioritized list of opportunities and a quantified value case. In a recent two-week discovery for a healthcare company we also shipped a working proof of concept of the highest-impact task, worth around $14 thousand a year on its own, inside a portfolio of about $1 million a year.
Do you only work with companies in São Paulo?
No. We work from São Paulo and Rio de Janeiro, and with companies across Brazil. Discovery usually starts on site, with your team, wherever it is. Delivery runs together with the client's team, remotely or in person. We work in Portuguese and in English.
What kind of company is a good fit for Extrace?
Mid-sized and large Brazilian companies, typically owner-led and often with a private equity fund in the capital, where AI has reached the executive layer but not the front line: leadership uses ChatGPT, the teams still run on spreadsheets and WhatsApp, and anything built internally does not scale for lack of security, integration and access control. We are not a fit for a company that wants a slide deck with no build, or for pure staff augmentation.
Which systems do you integrate AI with?
The ones the company already runs: the CRM, WhatsApp and Teams, e-mail, spreadsheets, documents and internal databases. We do not ask a client to replace its stack; we build into it, with proper access control, so what we ship can run in production and be maintained by the client's team.
How do you handle our data and security?
We sign an NDA before the discovery starts. What we build runs in your own cloud or tenant, so your data does not leave it, and we use the models through APIs under enterprise terms that rule out training on your data. Access is set by role, every step the agent takes is logged, and a person approves before anything goes out. We work within the LGPD, sign a DPA when you ask for one and delete your data from our environments when the engagement ends.
How do we start?
Write to hello@extrace.ai or use the form on our contact page. The first conversation is about your business, not about AI: what holds it back, where the time goes, what a good year looks like. We'll come back with where we'd start and what a discovery would cover.
Contact
Tell us where the hours go. We'll come back with a first step.
The first conversation is about your business, not about AI.