As a tax attorney in Brazil, Adriana has just signed on a new client for a huge tax compliance lawsuit. After the thrill of winning the big client fades, reality sets in. What Adriana dreads most about starting her case is the drudgery of initiating and performing the intensive required research. She knows she will spend hundreds of hours searching and analyzing prior legal case precedents and tax rules in order to prepare her client’s case.
As noted in the NYTimes, Brazil is known to have the most complicated tax system in the world and legal compliance in Brazil is highly costly and very inefficient. Tax rules change constantly — more than 377,000 new rules have been added since the late 1980s — and currently, there are more than 28,000 tax rules in effect. Taxpayers and revenue agencies are constantly disputing in the courts and the estimated accumulated amount under litigation is at least 12% of Brazil’s GDP (1).
To prepare for tax litigation, lawyers in Brazil spend hundreds of hours on complex legal research performing analysis of case precedents and studying their outcomes in order to decide the most effective litigation or compliance strategy. The process is not only laborious from a research point of view — since the tax codes are constantly changing and new court decisions appear almost on a weekly basis — but lawyers must collaborate and coordinate between their own teams and with their clients. For a large and complex case, a lot of time and energy is spent maintaining communication among the legal teams collaborating on the case. Since attorneys rarely use tools to collect, archive and share research results, they usually start every case from scratch — losing out on the opportunity to build upon previous work.
A goal of improving research productivity of attorneys working on Brazilian tax cases
Sao Paulo-based startup Turivius is working on how to improve the productivity of tax attorneys in Brazil and their capacity to leverage data analytics to refine their compliance and litigation decisions. The founding team is developing a machine learning platform that integrates legal research, data analytics, an archive-and-share tool and automated production of documents in the same workflow.
Machine learning enables the automation of applying heuristics, or rules. The Turivius platform “learns” the best results for each search, improving results in each user interaction. For example, consider the guidance Adriana might give to her intern tasked with searching for relevant case precedents. Directions to guide the intern’s efforts could be expressed in rules such as “Only find court decisions from the last 10 years that deal with transfer pricing and in which the judge ruled in favor of the taxpayer”. Adriana can provide feedback to guide Turivius’ search results to focus on the precedent cases she is most interested in, similar to the way she might give guidance to an intern who is performing searches. Like a good intern, the platform quickly learns the preferences of its supervisor. But different from an intern, Turivius learning process is augmented by its interaction it has with several other lawyers guiding it. After several iterations in which lawyers seamlessly ‘crowdsource’ their search and result preferences, Turivius’ artificial intelligence encapsulates the optimal research patterns of hundreds of lawyers.
The result is a system that can accurately predict the best case and norms for each situation, saving a huge amount of time and labor for Adriana and her law firm. Instead of hundreds of hours of research to prepare for her case and coordinate aspects of the case with her legal team, Adriana can perform better quality research in far fewer hours and with less effort.
As Adriana advances in her search for case precedents, Turivius also understands and recommends the most relevant data analytics for the topic she is working on and delivers it to her without interrupting her workflow. There is no need to split time and attention between a research tool and an analytics tool since both are integrated in her workflow.
Adriana continues her search and research efforts. As she views the resulting recommendations and data analytics from Turivius, she pins relevant ones to a board similar to the way she would pin images to a Pinterest board. By saving the relevant search results into a Turivius case folder shared in the cloud with her entire team, Adriana’s research is always up to date and available to everyone on her team.
The tax code’s fast-changing rules and the high volume of new court decisions make it nearly impossible to keep up with the latest updates. To solve this challenge, Turivius continually monitors these updates to keep abreast of changes to relevant precedents and rules and incorporates the updates into relevant case folders. Machine learning applied to the tax code enables Adriana and her legal team to always be up to date with case precedents, with very little effort.
For an attorney, the final step in her workflow includes the production of a legal document such as a petition to file in court or a legal opinion to be sent to a client. Turivius is working on supporting Adriana’s “end to end” legal workflow by automating this final step of legal document production. Instead of trying to reinventing the wheel in this area, the team has decided to focus on integrating its own platform with existing solutions for automating document production that already exist in Brazil. By doing so, the platform will enable Adriana to only fill out a brief form that will transform the results of her search into a legal document of her choice in a matter of seconds. She will save extra time by not having to switch to a text editing application and initiating the time-consuming process of brainstorming, structuring and writing the legal document
In summary, integrating legal research, analytics, knowledge management and automated document production into a unified experience helps relieve lawyers of repetitive and intellectually undemanding tasks so that they can focus their efforts on high-priority matters for their clients. Additionally, adding analytics will empower them to make data-driven decisions on litigation and compliance strategy as well as facilitate how they communicate and justify these decisions to clients. The potential gains in productivity and quality of decisions for Adriana and her team can be transformational.
Towards a collaborative experience for tax compliance
The Turivius team is focusing on key areas to ensure business success and a major part involves prioritizing the platform’s user experience. The founding team is actively working with early customers to understand their pain points and incorporate feedback to provide a solution that enhances users’ workflow and streamlines their productivity.
Here are three key principles the Turivius team is employing for finding a way to combine the best digital product design methods with pragmatic applications of machine learning focused on legal compliance.
A relentless focus on user adoption by solving pain points early
Successful digital products combine “pain killers” — solutions to customers’ pain — with “vitamins”, which are additional services or insights that leverage the solution and help drive engagement and retention. Driving adoption is easier in cases where the product is solving a pain point as compared the optimizing customer retention.
The Turivius team is focused on encouraging user adoption by targeting a solution to the unique and high priority pain points of tax attorneys in Brazil, its initial target market. A key pain point the team identified early on in market research is the hundreds of hours tax lawyers spend searching case precedents found among numerous court websites. As an early part of the user journey, using Turivius starts with a “Welcome Experience” and then offers an experience centered on “informed search” — where users can search all case precedents across many court databases with a single query. Nothing gets in the user’s way in initiating and quickly completing searches and reviewing results.
Centering the platform experience on informed search solves the tax attorneys’ critical pain point early on in the user journey, showing accurate results without a lot of effort while concurrently building the user’s trust. Search results can be narrowed down by using filters and tagging can be used to provide guidance on search result accuracy. Doing this puts the lawyer in control of guiding the system to achieve highly accurate results in a collaborative way.
Setting a high bar for usability
The Turivius team is setting a high bar for usability, placing it at or above the level of consumer online services in the belief that searching for relevant case precedents should be as simple and easy as finding a place to stay on Airbnb.com.
As Adriana progresses with her case research and sees search results, the platform delivers customized ‘vitamins’, i.e., data analytics with insights related specifically to that search input. By combining a high degree of usability with accurate results derived across a wide variety of data sources, it’s a goal of the user experience to drive “habitual” use and deliver a service that the tax attorneys will look forward to using daily.
Success for the Turivius UX includes both qualitative and quantitative metrics. If successful, combining pain-killers and vitamins in a seamless user experience should cause Adriana to positively anticipate having to perform a legal search. As a qualitative measure, the Turivius team aspires to have Adriana saying a year from now that she can’t imagine doing her job without Turivius — and in turn encouraging her attorney colleagues to try the service. To gather quantitative UX metrics the team is exploring ways to allow users to be able to self-assess their own productivity gains while using the platform in addition to monitoring activities.
Balancing user needs and business goals while leveraging automation to deliver a great user experience.
Turivius’ approach is inspired by the notion of augmenting lawyers’ expertise — not replacing it — moving towards improving their productivity and decreasing drudgery. Through enabling this kind of collaborative “everyday automation”, high-skilled labor and machines can collaborate in a way that the human component of this symbiotic relationship can focus on things that only humans can do: generate concepts for a new legal thesis, probing evidence, negotiating deals or discovering new clients. Lawyers should feel that the platform reduces their workload and frees up their time.
As the Turivius platform takes care of the most tedious part of the case research, attorneys can engage less and less with it. This leads to a paradigm that departs from current practices in the development of digital products emphasizing the need to drive up user engagement to the limits of addiction — hands constantly on a keyboard or tapping on a phone screen. We understand that these are new engagement metrics that are not typical for SaaS platforms, where success is measured in fewer — not more — key inputs, clicks or time spent on a page. We are also aware of the challenges of delivering technology to an audience that includes segments that are tech-averse, hence the focus on prioritizing the delivery of ‘pain killers’ through a user experience that delivered at the level of modern and successful consumer online services.
For the Turivius team, the mission is to build a future where technology can enrich people’s lives by automatically performing tedious work so that users spend less time in front of a computer screen and more time engaged in value-added work. The appropriate value metric ceases to be continued use, but how much the user is willing to pay for such time-saving convenience.
We understand that machine learning is pointing towards a future where the perception is that human users will be replaced by computer software. The Turivius team is focused on automating tasks that are tedious and require a large amount of manual labor by tax lawyers who are under pressure to complete their legal case research and serve their clients. We are trying to streamline legal case workflows to make lawyers’ interactions with technology during case research more natural and valuable, in turn facilitating better collaboration among legal teams and resulting in a better case outcome.
By staying focused on early customers’ needs and pain points, Turivius is creating business value in the area of legal compliance through automation and seamless integration with business processes centered on delivering a great user experience.
Karen Donoghue, Principal Interaction Architect at HumanLogic
Karen is a consultant focused on product planning and user experience architecture for enterprise platforms. She is an Advisor to Turivius and is working to define and enhance the Turivius user experience. A graduate of MIT, she advises companies ranging from the Fortune 100 to startups about how they can most effectively leverage strong user experience architecture and design to empower their customers and employees.
Danilo Limoeiro, PhD, CEO and Co-Founder, Turivius
Danilo holds a Ph.D. from MIT and a Master of Philosophy from Oxford University. He is a former policy advisor and manager in the Brazilian federal government and co-founded Turivius’ in 2018 with the goal of developing and deploying new technologies to decrease bureaucracy.
Guilherme Kenzo, CTO and Co-Founder of Turivius
Guilherme has a law degree and is a software developer finishing his Master’s degree at the University of São Paulo. He is the founder of Hackerspace in his hometown and a finalist of Hack Brazil, a startup competition organized by MIT and Harvard students. He has expertise in artificial intelligence and natural language processing.
For more information on Turivius’ work on automating legal compliance, please visit the Turivius web site.
(1) The estimate is based on data published by Brazil’s Public Account Court (Tribunal de Contas da União) Report TC 015.289/2018–4. According to the Report, the total of accumulated litigation in the federal administrative fiscal court (Conselho Administrativo de Recursos Fiscais) is BRL 600 billion and the accumulated disputes in Brazil’s local branches of the Federal Revenue Service is BRL 160 billion. This total still does not account for the hundreds of thousands of Judiciary Branch tax lawsuits and disputes in the many state-level administrative courts. Accounting for these sums could take the total close to 15% of Brazil’s BRL 6,8 trillion GDP.