2025
Baker Tilly México Transforms Tax Litigation with SentenciA powered by Karini AI on AWS
Learn how Baker Tilly uses generative AI to analyze 800,000+ legal precedents, reduce research time by 60%, and deliver data-driven litigation strategies.
reduction in legal research time
legal precedents analyzed in real-time
Overview
SentenciA is an LLM-driven analytics platform developed by Baker Tilly México and deployed on AWS to bring evidence-based, quantitative risk assessments to administrative litigation before the Mexican Federal Administrative Justice Court (TFJA).
By analyzing in real time more than 800,000 public TFJA precedents, research teams can now estimate the probability of obtaining a favorable decision, understand recurring patterns of particular chambers and magistrates, and design stronger defense strategies aligned with TFJA requirements.
Built on Karini AI's Agentic AI Foundation platform, SentenciA leverages Amazon Bedrock, Amazon Textract, Amazon Neptune, and Amazon OpenSearch Service to transform unstructured judicial information into a rich knowledge graph. This has led to a 60% reduction in legal research time.
About Baker Tilly México
Baker Tilly México is part of Baker Tilly International, a global network providing audit, tax, legal, and advisory services. Its tax litigation practice handles high-stakes disputes with tax liabilities reaching hundreds of millions or billions of pesos.
To meet growing client expectations, the firm sought to move from intuition-based advice to evidence-based, quantitative risk assessments grounded in TFJA's precedent history.
"We needed to transform tax litigation from intuition-driven practice into an evidence-based discipline. Our clients deserve risk assessments grounded in comprehensive precedent analysis, not limited samples."
— Adrián Bueno Osorno, Socio de Legal, Baker Tilly México
Opportunity | Using Amazon Bedrock and Karini AI to Transform Legal Research
Baker Tilly México's tax litigation team faced a challenge: manually reviewing hundreds of thousands of TFJA precedents was slow, subjective, and impossible to scale. Lawyers spent days searching for relevant cases based on limited samples and institutional memory.
With billions of pesos at stake, the firm needed a data-driven approach to analyze the entire corpus of federal precedents and deliver quantitative risk assessments in real time.
Outcome | Measurable Impact and Future Applications
60% Reduction in Research Time
From 8-10 hours to just 1-2 hours per assignment, freeing attorneys to focus on strategy.
Expanded Analytical Scope
From reviewing 15-20 precedents to analyzing thousands instantaneously with quantitative indicators.
Enhanced Case Capacity
Accept more complex matters and optimize resources with insights into magistrate patterns.
Future Applications
Expanding to Supreme Court jurisprudence and other practice areas like labor law.
"SentenciA has fundamentally changed how we approach administrative litigation. We can deliver evidence-based risk assessments backed by comprehensive analysis of hundreds of thousands of precedents."
— Legal Research Team, Baker Tilly International
AWS Services Used
Amazon Bedrock
Foundation models for natural language understanding through Anthropic's Claude.
Learn more →Amazon OpenSearch
Fast semantic search for instant retrieval of relevant legal decisions.
Learn more →