Submission type
Generic use case
Name
LegalMitra-YourLegalCompanion
Industry category in which use case would best fit in (Select up to 2 industries)
Legal
Complexity level
Advanced
Summary (abstract)
LegalMitra is an AI-powered consumer dispute assistant built on UiPath Maestro BPMN that helps everyday consumers in India understand their legal rights and navigate the consumer forum process — without needing to first find and afford a lawyer. When a victim emails their complaint, a UiPath Coded Agent (built with LangGraph and grounded via UiPath Context Grounding) analyses it against 100+ real National Consumer Disputes Redressal Commission (NCDRC) judgements, producing a structured legal assessment: applicable law, relevant precedent, case strengths and weaknesses, a weighted confidence score, and a recommended course of action. The victim receives this in plain language via email and multi-language WhatsApp (English, Hindi, and their regional language via Twilio), while the assigned lawyer receives the full report and makes the final call — proceed to the consumer forum or pursue settlement — through a human-in-the-loop review in a UiPath App. The entire case lifecycle, from intake to closure, is tracked in UiPath Data Fabric. What previously took a lawyer approximately 3 hours of research and drafting per case is reduced to roughly 8 minutes of automated analysis, with human judgment preserved at the one point that matters most.
Detailed problem statement
India’s consumer protection ecosystem processes approximately 5 million complaints annually, and the Consumer Protection Act 2019 (CPA 2019) provides a clear, well-established legal framework for resolving them through District, State, and National Consumer Disputes Redressal Commissions. In principle, the system works. In practice, access to it is deeply unequal.
Approximately 75% of complainants appear before consumer forums without legal representation. This is not because their cases lack merit — it is because the process of preparing a case is itself a barrier. A complainant needs to know which section of the CPA 2019 applies to their situation, what a consumer forum will expect to see as evidence, what similar cases have resulted in, and how to structure a formal complaint or legal notice. None of this is intuitive to someone without legal training, and none of it is available to them in a form they can use quickly.
This barrier is not evenly distributed. Victims in Tier-2 and Tier-3 cities — where the majority of India’s population lives — have the least access to affordable legal counsel and the least exposure to English-language legal documentation. A consumer in Nagpur or Coimbatore facing a defective product dispute has functionally less access to justice than one in a metro city, purely because of geography and language, not because their case is any weaker.
On the supply side, the constraint is different but equally real. Lawyers and legal aid practitioners who do serve this population spend a disproportionate share of their time — roughly 2 to 3 hours per case by our estimate — on repetitive research and first-draft work: searching for relevant precedent, identifying applicable statutory provisions, and drafting an initial complaint or notice. This is necessary work, but it does not require a lawyer’s judgment — it requires their time, which is the scarcest and most expensive resource in the system. Because that time is expensive, affordable legal aid for consumer disputes remains difficult to scale, and the backlog of self-represented, under-prepared complainants continues to grow.
The result is a system with strong law and legitimate institutions that remains inaccessible to the majority of the people it was built to serve — not due to any deficiency in the legal framework, but due to a structural gap in case preparation, language accessibility, and research capacity at the point of intake.
Detailed solution
LegalMitra addresses this gap by automating the research- and preparation-intensive portion of the consumer dispute process while deliberately preserving human legal judgment at the decision point where it is legally and ethically required.
Intake. A victim emails a description of their dispute to a designated inbox. A Gmail Integration Service trigger initiates a UiPath Maestro BPMN process, which extracts the case details and creates a persistent case record in UiPath Data Fabric — the single source of truth for the case throughout its lifecycle.
AI-powered legal analysis. The core of the solution is a UiPath Coded Agent built with LangGraph. The agent first queries a UiPath Context Grounding index — CPACaseJudgements — containing over 100 real NCDRC and State Consumer Commission judgements sourced from the public legal database indiankanoon.org, spanning ten major dispute categories including property, e-commerce, banking, insurance, medical services, education, and more. This Retrieval-Augmented Generation (RAG) step retrieves the most semantically relevant precedents for the specific complaint. The agent then calls an LLM (Claude, via UiPath’s LLM Gateway) with the case description and retrieved precedents to produce a structured output: a case summary, the applicable CPA 2019 provisions explained in plain language, the relevant precedents and how they apply, specific strengths and weaknesses of the case, a recommended course of action (filing directly with the consumer forum, or sending a formal lawyer’s notice first), and a transparent, weighted confidence score built from five explicit criteria — narration quality, precedent applicability, legal clarity, claim quantifiability, and procedural compliance.
Multi-language, multi-channel communication. The victim receives the full analysis by email and a concise, plain-language summary via WhatsApp (through Twilio Integration Service) — delivered simultaneously in English, Hindi, and their regional language, automatically selected based on their city. This directly addresses the language accessibility barrier that excludes most Tier-2 and Tier-3 complainants from engaging with legal processes conducted in English.
Human-in-the-loop review. The full structured analysis is also sent to the assigned lawyer, who reviews it in a UiPath App deployed to Action Center. The lawyer — not the AI — makes the final decision: proceed toward filing with the consumer forum, or pursue a settlement outside court. This decision drives the remaining branches of the BPMN process, each of which notifies the victim of the outcome and updates the case record in Data Fabric through to closure.
A conversational companion for the victim. A second, separate UiPath Conversational Coded Agent — grounded on the same Context Grounding index — allows the victim to interact directly and ask follow-up questions in natural language, receiving concise, precise answers along with a dynamic “resolution likelihood” indicator that updates as the conversation reveals more detail. Where a case looks weak, the agent does not simply reject it — it actively offers to connect the victim with LegalMitra’s legal team, including the option of pursuing an out-of-court settlement, so no one is left with a dead end.
The result is a system where an AI agent does the research and first-pass legal reasoning that used to consume hours of a lawyer’s time, while every consequential legal action — a notice being sent, a case being filed — still passes through a qualified human before it happens.
Narrated video link (sample: https://bit.ly/4pvuNEL)
Expected impact of this automation
Time savings. Legal research and precedent identification that previously took approximately 90 minutes per case is completed by the RAG pipeline in under 30 seconds. Case analysis and initial drafting that took roughly 60 minutes is produced by the agent in under a minute. Combined with automated multi-channel victim communication, the total time from complaint receipt to a fully analysed, communicated case drops from approximately 3 hours to roughly 8 minutes — a reduction of approximately 95%, with the lawyer’s remaining time spent entirely on judgment rather than research.
Capacity multiplication for legal aid. Because lawyer time is the binding constraint on how many consumer disputes a legal aid organisation or independent practitioner can take on, reclaiming the majority of research and drafting time translates directly into more cases served with the same staff. A practitioner or organisation that previously handled a fixed caseload could realistically take on several times that volume without adding legal headcount. At the population level, applied across India’s approximately 5 million annual consumer complaints, even partial adoption represents millions of hours of legal research capacity returned to the system.
Language and geographic accessibility. By delivering case analysis in the victim’s own language over WhatsApp — the dominant communication channel in Tier-2 and Tier-3 India — LegalMitra removes the language barrier that today silently excludes a large share of complainants from understanding their own legal position, regardless of the merits of their case.
Improved case quality at the point of filing. Every case that proceeds is grounded in a systematic review of real, relevant precedent rather than whatever an individual lawyer happens to recall or has time to search for. This raises the quality floor of case preparation across the board, particularly benefiting practitioners handling high case volumes who cannot manually research every matter to the same depth.
Preserved accountability. Because no legal action is taken without explicit lawyer approval, and because the confidence scoring is transparent and auditable rather than an opaque probability, the system improves throughput without displacing professional responsibility — the human remains accountable for every consequential decision, informed by better and faster information than they previously had.
A replicable model. The architecture — a Coded Agent grounded on a UiPath Context Grounding index of domain precedent, orchestrated through Maestro, with human review at the point of consequential action — is directly extensible to other legal domains beyond consumer protection (tenancy disputes, employment claims, motor accident claims), and to other public-access-to-services problems more broadly, wherever the barrier to a citizen exercising their rights is research capacity and language, not the underlying law itself.
UiPath products used (select up to 4 items)
UiPath Action Center
UiPath Agent Builder
UiPath Coded Agents
UiPath Data Service
UiPath Integration Service
UiPath Maestro
Integration with external technologies
Gmail, Twilio
