back
workers.ai
We built Brazil's first AI SDR in 2024, before the agent wave. The design problem was getting companies to trust it with their pipeline.
Role
Product Designer
Product Manager
Team
1 product owner
3 developers
Period
04/2024 to 05/2025
About the company
Workers.AI built Brazil's first AI SDR platform, launched in 2024, before AI agents became the crowded default. Autonomous agents that qualify sales leads, with human-in-the-loop supervision, a knowledge base, and granular behavior configuration. Bootstrapped, and at break-even within a year.
—
Tags
I joined Workers.AI as Product Designer and Product Manager and turned a working prototype into Brazil's first AI SDR platform, built before agents became the crowded default they are now. When I came in there was a POC but no product: no backlog, no MVP, no flows, no specs. For a long stretch I carried both the design and the product management, working as PM and PO while the company found its footing.
Logo I designed to temporarily give the project an identity, one that ended up sticking.
The problem
By 2023, language models could do work that used to take a team. But sales still ran on human SDRs and basic chatbots, and both were breaking
The SDR role was expensive and unstable
A junior SDR cost R$6,000 to R$9,000 a month before benefits and tools, took around 90 days to reach full productivity, and churned at roughly 40%, by the founder's read of the market.
Companies lost institutional knowledge every time one walked out.
The tools that existed weren't the answer. Most were chatbots: they answered questions but didn't do the work, had no governance, no limits, and were built for engineers, not for the sales managers who'd actually run them. Underneath it all sat the real barrier: no company hands its sales pipeline to an AI on day one.
How I got to the product
Most of my understanding came from interviews, and it came fast, because the company had strong commercial people who knew the work cold. I learned the pattern from them: there is no single SDR, every company qualifies, talks, and sells differently, so the first job was understanding that variability rather than pretending it away.
Out of that, one design method emerged and shaped every major decision: take a process the user already understands from the human world, and translate it to the agent. I saw the pattern, and we built on it, first for the interface, then for supervision, then for onboarding.
What I built
We shipped incrementally, starting with a focused MVP, lead qualification, basic personalization, and the knowledge base, then layering in configuration, supervision, and omnichannel as trust and usage grew. Three moves mattered most.
Design it like managing a team, not debugging a model
I could have exposed the models, parameters, and tokens. Technically honest, useless to the buyer. Sales managers think in hiring, training, and managing people, so the whole product used that model: you hire a Worker, train it on your business, set its goals, and manage it. Adoption moved fast because nothing about it felt like operating a model.
01.
Using familiar flows from other tools, the user can configure the cadence the AI agent will follow during outreach, as well as monitor all interactions in real time.
02.
Performance dashboard
Build trust in layers
Companies don't trust AI overnight, so I designed supervision as a ladder, not a switch.
Supervision: three levels
Level 1
Every message is reviewed before it sends.
Level 2
Only selected agents or high-value leads need approval.
Level 3
The agent runs on its own and the manager just monitors. Most clients started at Level 1. After a week or two of watching the quality hold, they moved up. Supervisor mode was the reason they said yes.
Managers could also see why the agent made each call, not just what it did. That visible reasoning turned skepticism into permission.
The user can deeply configure each agent's scope of work and personality, ensuring that the language aligns with their business tone and the profile of the leads being engaged.
Make it say "I don't know"
An AI that invents answers under pressure is worse than no AI in sales. One confidently wrong claim about pricing or terms can kill a deal. So I designed the knowledge base to ground every answer in the company's own materials, and to admit the gap when the answer wasn't there. We trained it on 1,500+ high-performance sales emails from early clients, so it learned from what actually worked, not from generic text.
Grounded, not guessing
The agent answers from ingested documents, sites, and sales materials. When something isn't there, it says "I don't know" instead of inventing. It sounds like someone who studied the material, because it did.
Study of how we could deliver better responses using a multi-agent structure.
Knowledge base whre all agents draws information.
How we worked
We were a small, remote team. It started as me and one developer, and as it grew I helped structure the company, hiring a PO and a front-end developer and keeping continuity through the developer turnover we went through. Everything ran documentation-first, with asynchronous updates and weekly reviews of real client conversations that fed straight into the next sprint. No "AI solves everything," and no waiting for a quarterly review to fix what wasn't working.
Results
LIDE Paraná, a major business association, put the agent on its member base and booked meetings across it in the first week, with no human SDR touching the keyboard. Across clients, reps stopped prospecting and moved to closing, receiving qualified opportunities instead of noise. Nobody fired their team; they kept their best people and pointed them at the work that closes.
By the numbers, clients contracted around 100 agents, at roughly 60% the cost of a junior SDR, and the company, bootstrapped, reached break-even in under a year. For a first-of-its-kind product in a market that didn't exist yet, that was the real proof: companies didn't just try it, they paid for it and kept it.
First
AI SDR platform in Brazil
1,500+
Sales emails as proprietary training data
~100
Agents contracted by clients
~60%
Cheaper than a junior SDR
What I learned
Trust in AI isn't granted, it's earned in layers. The instinct with a capable model is to show it off and let it run. What actually made companies adopt was the opposite: visible reasoning, a supervision ladder they controlled, and an agent honest enough to say "I don't know." Years later, that oversight is exactly what the industry is scrambling to add. The design job was never to make the AI look capable but to make people comfortable handing it something they cared about, one step at a time.
