Case study on structuring an artificial intelligence platform for legal petitions

Understanding the problem

The context

Legal systems organize the application of law, varying in basis, which can be civil law (codified law, typical of Brazil, continental Europe) or common law (jurisprudence prevails, typical of USA, United Kingdom). In Brazil, the system is based on civil law, with a strong focus on written codes and norms, where the judicial process is documentary and formal.

Petitions are initial or intermediate procedural documents that formalize requests to the Judiciary. Examples: initial petition (for process initiation), intermediate petition (to express oneself in the process), appeals, manifestations, etc. They are the main form of communication between lawyers and the Judiciary, and their correct drafting is essential for the processing and decision of processes.

Current scenario

Volume de Petições: Volume of Petitions: According to data from CNJ (National Council of Justice) and TRT (Regional Labor Court), Brazil processes millions of annual cases, and each case can contain several petitions. For example, CNJ reports that in 2023, the Brazilian Judiciary recorded more than 20 million active cases, with an estimate of tens of millions of petitions filed annually.

Custos e Tempo: Costs and Time: Most costs and delays in the Judiciary are linked to the volume and complexity of petitions. Studies indicate that poorly drafted petitions generate dismissals, rework, and delays. A poorly founded petition can increase judgment time by weeks or months.

Digitalização e Processo Eletrônico: Digitalization and Electronic Process: The implementation of PJe (Electronic Judicial Process) has increased speed and control over petitions, but also generated high demand for software capable of managing large document volume.

Identified opportunity

Given the scenario faced and the rise of artificial intelligences, we have possible problems to be solved:

• Generate personalized petitions based on specific case information;
• Reduction in preparation time (from days to hours/minutes);
• Reduction of human errors;
• Time optimization for strategic lawyer activities;
• Greater efficiency in managing large document volume;

Desk research

Use of Artificial Intelligence (AI) in Law and Petitions

AI Landscape in Law: AI has been applied to optimize document analysis, jurisprudence research, predictive analysis of decisions, contract review, and automation of legal documents. Tools like IBM Watson Legal, Luminance, and local startups focus on the digital transformation of law.

IA em Petições: AI in Petitions: Automation of drafting: Platforms use AI to generate petition drafts based on data and legal models. Ex: Brazilian legal techs like JusBrasil and Neoway have solutions for this. Analysis and review: AI can review petitions, detecting inconsistencies, formal errors and weak legal points, reducing risk of dismissal. Research and jurisprudence: AI legal assistants help lawyers find relevant precedents to support petitions. Outcome prediction: Predictive analysis can guide the best strategy for petitions, indicating success chances.

Dados e Estudos: Data and Studies: According to Thomson Reuters research, 64% of large law firms already use some type of AI to optimize repetitive tasks. In Brazil, research indicates 20-30% annual growth in the use of legal techs that use AI for document automation.

Challenges and Limitations

Adaptação Cultural: Cultural Adaptation: Initial resistance of legal operators to AI adoption.

Regulamentação: Regulation: Ethical use of AI, data privacy, legal responsibility for errors generated by automated systems.

Qualidade e Treinamento dos Dados: Data Quality and Training: AI depends on well-structured and updated databases to function correctly.

Sources and References

• - CNJ — Relatórios estatísticos anuais do Poder Judiciário (disponível em: cnj.jus.br);
• - Thomson Reuters — Legal Tech Report 2023;
• - Pesquisa FGV/Direito Rio sobre IA e Direito;
• - Relatórios de startups brasileiras de legal tech (ex: JusBrasil, Neoway);
• - "Impacto da inteligência artificial na prática jurídica" (Revista de Direito e Tecnologia, 2024);
• - IBM Watson Legal Overview;
• - Dados públicos do Tribunal Regional do Trabalho (TRT);

Competitive benchmarking

Through competitive benchmarking, it was possible to analyze the main artificial intelligence platforms focused on Law and petitions, which are:

Quantitative research

A survey was conducted by the team involved in the project. Analyzing the results, the following points stand out:

of people use some type of template for creating petitions

78%

of people have already used artificial intelligence to create petitions

57,7%%

of people would use a petition creation tool several times a week

65%

Analyzing results

The impact on society

Petition automation platforms, like Ticio.ai, are transforming the legal sector and generating relevant social impacts:

Acesso à Justiça: Access to Justice: Making legal production faster and cheaper allows more people, especially low-income people, to have access to quality legal services.

Agilidade no Judiciário: Judiciary Agility: Well-structured documents with fewer errors reduce rework and contribute to unclogging the judicial system, making processes faster.

Novo papel para o advogado: New role for the lawyer: By automating repetitive tasks, the lawyer can focus on strategy, analysis and client relationship, elevating the value of their work.

Inovação na educação:Education innovation: These tools also help better train Law students, focusing on real practice and correct structuring of documents.

Responsabilidade e ética: Responsibility and ethics: It is essential to ensure that technology is used with transparency, human supervision and respect for legal principles to avoid abuses or automated errors.

Legal automation can expand access to Justice and improve the efficiency of Law in Brazil, as long as it is used responsibly and focused on social impact.

Mapping users

From the data collected in the research, personas and a user journey map were developed, with the objective of identifying opportunities and directing the next stages of the process:

Idealizing the solution

Brainstorming

Using unstructured brainstorming, possible points to be considered in idealizing the solution for the previously identified problems were raised:

• Facilitate collaboration between team members;
• Track the history of each created petition;
• Save and organize templates and documents;

After brainstorming, it was possible to identify that the solution should be a responsive web platform. Creating an app in the initial version would not be viable due to the complexity of the project and limited time for development.

Mind map

The practice was used to organize the solution structure in a faster and more efficient way:

User flow

After the mapping and organization created in the Mind map, at this stage the user navigation flowchart was created:

Low fidelity prototype

With the flow defined, the prototyping stage was initiated. At this first moment, low fidelity screens were created, allowing rapid validation of initial ideas and concept adjustment before advancing to more detailed versions.

Usability tests

Some usability tests were executed on the low fidelity prototype, from which some flow and writing errors were observed. Subsequently corrected in the high fidelity prototype.

Test planning:

• Definition of activities to be tested;
• Recruitment of users who resembled the personas;
• Creation of the script detailing what would be the "success" of the selected activities for testing.

Metrics analyzed in the tests:

• Completion rate;
• Complexity rate;
• Execution time

Moodboard

To contextualize the entire visual identity of the application, a board was created to be followed as an aesthetic/visual model for creating the styleguide.

UI Kit

To contextualize the entire visual identity of the application, a board was created to be followed as an aesthetic/visual model for creating the UI Kit.

High fidelity prototype

Taking into consideration the design system and insights from usability tests, it was possible to structure the high fidelity prototype:

Mobile:

Desktop:

Conclusion and learnings

Throughout the project, we mapped real automation opportunities, deepened understanding of lawyers' and legal teams' pain points, and transformed this knowledge into a functional, accessible platform with high added value.

Legal automation has growing space, but still requires market education for broad and conscious adoption.

The integration between AI and traditional legal knowledge is not about replacing the lawyer, but rather expanding their productive capacity.

The user experience needs to be fluid, even when dealing with a technical and sensitive topic like Law.