Enter × Unicamp / September 2026
Enter Decision Platform.
A team-built prototype that helps legal teams consider litigation and settlement, while keeping the final decision with a lawyer.
The problem
Legal teams managing large volumes of cases must decide which cases to defend and when to propose a settlement. Each decision involves uncertainty about the outcome, potential financial loss and the amount of a reasonable offer. A useful decision-support tool needs to make these considerations understandable to its users.
At the Enter × Unicamp Hackathon, held with SECOMP in September 2026, our team worked on this problem over approximately 17 hours. We built a prototype combining case analysis, recommendations and dashboards, and placed first in the event.
The approach
The prototype analyzes case documents and models the probability of an unfavorable outcome alongside projected financial loss. It then recommends considering a defense or settlement. When recommending settlement, it proposes an offer informed by the company's policy.
The recommendation includes an explanation of the factors behind it. This is central to the workflow: the lawyer should be able to inspect the reasoning and retain responsibility for the decision, rather than act on an unexplained output.
The team also built dashboards to examine model adherence and outcome performance. The interface brings the analysis and recommendation into a single workflow that can be explored through the demonstration.
My contribution
My work focused on the AI Core component of the prototype. I worked as part of the team building the decision-support system.
Working as a team
I participated alongside Andrei Quaresma Pinto, Bruno Amaral, Sofia Valverde and Tainá Almeida. The project brought together different parts of a software solution under a short deadline: understanding the problem, building the decision-support flow and presenting the result.
The first-place result belongs to the team. The event provided a setting for applying AI to a business problem and making the resulting recommendations inspectable by a human user.
What the result shows
A functional hackathon prototype and a first-place team result at Enter × Unicamp.
The demo illustrates a proposed decision-support workflow. It is not evidence of production adoption, realized financial savings or validated legal outcomes. Those would require deployment and evaluation beyond the hackathon.
The important design choice is the relationship between an AI recommendation and its user: the output provides supporting information, while a lawyer reviews the context and makes the final decision.