How Abbiamo replaced its entire support team and turned Duckie into a revenue engine

Abbiamo is building the post-purchase infrastructure layer for e-commerce in Brazil and Latin America. That means orchestrating everything from checkout through final delivery — tracking, disputes, returns, exchanges, carrier integrations, and customer notifications — across a continental geography with hundreds of carriers and wildly different data standards.

The scale is real. More than three million deliveries move through Abbiamo every month. Around 20% of those generate disputes: packages that never arrived, wrong items, damaged goods. Every one of those disputes creates a support interaction, and those interactions arrive from multiple directions at once — from end customers, from carriers, from brand operators managing their own logistics.

For a while, Abbiamo's answer to this complexity was headcount. A head of customer support, two reps beneath them, a triage process, and a Freshdesk queue. The system handled basic tickets adequately, but response times stretched longer with each passing week, and the quality of answers became extremely variable as the team struggled to keep up with growing volume.

"We were hiring people and the knowledge base was staying in people's heads," said Aurélio de Pádua Gandra, founder and CEO. "It would take two or three months for a person to be well-versed in Abbiamo. That's just not scalable."

The quality problem was as sharp as the cost problem. Aurélio is a technical founder who spends significant time with customers — someone who cares deeply about tone, quality, detail, and style in every customer interaction, because those elements are crucial for his business. When he reviewed tickets being sent under Abbiamo's name, he was uncomfortable with what he saw. The team was oscillating between two failure modes: technically accurate but cold, or warm but imprecise. Getting both right at the same time proved nearly impossible — the technical reps delivered precise but cold responses, while the warmer reps lacked the technical depth to resolve complex issues.

"I always had to make this choice," he said. "Super technical, great answers, but bad customer experience. Or very good customer touch, but bad on the technical side. We were living between those two worlds and getting it right in neither."

The support problem at Abbiamo wasn't just a representative consistency problem. It was an AI quality problem.

When Aurélio and his team evaluated other AI support tools, they hit a fundamental wall: the systems either sounded robotic and rigid, or they went completely off-script. Some promised capabilities that didn't exist. Others confidently answered questions they had no business answering. For a B2B logistics platform handling millions of deliveries, those failure modes were unacceptable.

"The hardest part is getting an AI to answer exactly what you want, the way you want — without sounding robotic," Aurélio said. "If you loosen the rails too much, it goes bananas. And as an enterprise B2B, we cannot let that happen in any case."

That's where Duckie was different. The guardrails were real. The fine-tuning actually worked. Duckie could maintain the precise balance Aurélio needed — technically accurate answers delivered with human warmth, taking actions for customers that actually resolved their issues, without veering into dangerous territory or making promises the platform couldn't keep.

Abbiamo replaced Freshdesk entirely. Where there was once a ticketing platform, a triage process, and a three-person team, there is now a single Duckie widget embedded in the dashboard. Nothing else.

The head of customer support is gone. So are the two reps. Aurélio now manages Duckie himself, not as a stopgap, but deliberately, because he believes customer support and product shouldn't just be the same department. He thinks customers should actually be part of the product itself.

The speed gains were immediate. What used to take three to four days — a ticket getting triaged, passed around, re-read by someone new, finally answered — now takes minutes. In the rare cases when Duckie can't handle something autonomously, it escalates directly to Linear with full context — pulling actual customer data, order history, and relevant logs — so engineers can resolve the issue quickly and effortlessly with Duckie's help.

"From end to end, we took three days off of our support ticket life cycle," Aurélio said.

The second use case emerged unexpectedly. Abbiamo regularly onboards new carriers into its platform — companies that need to integrate via API, run tests, validate behavior, and start shipping orders. Historically this meant a lot of back-and-forth email: "Is this the expected behavior?" "What should I do next?" "Is this field required?" Customers would get confused by the documentation, and questions often required manual human assistance.

When the Duckie widget went live, the majority of early questions weren't from existing customers at all. They were from people in the middle of onboarding.

Duckie now knows Abbiamo's full API documentation. It can answer integration questions the way a specialist would, and help customers fully resolve issues by taking actions on their behalf. Duckie does all this with the precision, context, tone, and technical detail that Abbiamo requires. Customer onboarding time dropped from over two months to just over four weeks — a ~2x improvement in time-to-revenue.

"Faster onboarding is revenue," Aurélio said. "A customer integrates, they start running orders, then you start charging them. We got weeks off our average onboarding cycle. That was not expected."

Faster onboarding doesn't just accelerate revenue — it amplifies it. When customers integrate in weeks instead of months, they're not only paying sooner, they're actually spending more. The momentum of a smooth onboarding carries through: they ship more orders, expand to more carriers, and commit deeper to the platform before doubt or friction can set in.

What made the difference was visibility. With human agents, Abbiamo was flying blind. Mistakes were happening constantly and they couldn't even see them. Quality was a black box, measured through weak proxies like customer satisfaction scores that arrived weeks too late to matter.

Now they can see everything. Every interaction, every response, every escalation is tracked with precision metrics that actually mean something. When something needs adjusting, it's a configuration change in the playground, not months of retraining humans who might interpret the feedback differently than intended. Batch testing ensures changes work across hundreds of scenarios before they ever touch a real customer.

The improvement curve has been exponential.

"We can see everything that's being done well and everything that's failing," he said. "That's actually an improvement over the human version, where mistakes were happening and we couldn't see them. Now we know exactly where to make changes."

Abbiamo went from three support employees to zero. Ticket volume increased 30% over the same period. Duckie costs the equivalent of one full-time employee. On a steady-state basis, that's a 66% reduction in support cost against a workload that is actively growing.

Two of Brazil's largest pharmaceutical companies are coming onto the Abbiamo platform later this year — a move expected to increase order volume by 80%. Daily ticket volume is projected at 10x November. The plan is to handle all of it with Duckie.

Time-to-revenue for new customers is now 2x faster — onboarding that used to take two months now takes four weeks. Carrier onboarding, once an unstructured back-and-forth, is now a guided experience that accelerates integration and gets customers to their first order faster. This isn't just operational efficiency — the faster onboarding directly translates to $400K in additional revenue this year from customers who would have otherwise churned during lengthy integrations.

The qualitative shift matters as well. Answers are now consistent and have the technical depth necessary while maintaining the warmth of an excellent customer support representative. They sound the way Aurélio would sound. The tone and technical depth can be adjusted with simple and effective configurations.

Abbiamo's situation is common to any platform that operates at the intersection of technical complexity and customer-facing operations. You need support that sounds human and answers correctly. You need onboarding that moves fast without creating confusion. And as you grow, you need both of those things to scale without the headcount climbing in lockstep.

The traditional answer has always been to throw more people at the problem — hire more agents, build bigger teams, accept that quality will vary with whoever's on shift. But that model breaks when you're trying to scale across multiple countries and languages while keeping your burn rate under control. What worked for Abbiamo was acknowledging that AI isn't about replacing human judgment — it's about amplifying the founder's expertise across every customer interaction, 24/7, without the overhead of managing a distributed support team.

"Duckie costs the same as one full-time employee I had," Aurélio said. "And I got rid of three. But more than the savings — I don't need to deal with hiring people, training people, watching the knowledge walk out the door when they leave. That's the real value."

The Abbiamo story is a volume story. Three million deliveries a month now. Eighty percent more by year's end, with two major customers coming online. The support infrastructure that took years to build and still underperformed is being replaced by a system that gets better as usage grows.

The onboarding use case, which wasn't part of the original plan, has become a source of competitive advantage. Faster time-to-revenue, less friction for carriers entering the ecosystem, a specialist available at any hour who knows every field in the API documentation. And now Duckie is building agentic frameworks that guide users to the exact tools they need, integrating directly into the product flow to solve their specific challenges and proactively drive revenue.

"It's not only saving money," Aurélio said. "It's actually growing. New revenue is king."