Vidhan uses advanced math for legal reasoning
to uncover the most effective strategies and solutions.
Designed by
Princeton math Ph.D. IIT alumniGuided by
experienced lawyers and judges
The lost argument
Every senior lawyer has experienced the pain of discovering a stronger argument after pleadings were filed, when there is little you can do with it
Unfortunately, that is also when partner attention is most fragmented across multiple matters. So only two or three theories get explored where ten might exist.
Counsel must still decide if the material supports a theory, what counters will be raised and where it falls short, instead of exploring all possible alternatives to find the strongest argument. The result is a weaker case.
Pure Generative AI can be confidently wrong especially in familiar looking yet genuinely new scenarios. So counsel must check the output's accuracy - a hard task if the reasoning process is not exposed.
The lost hours
Current legal AI seem useful. Summaries. Annotations. Drafts. But the law is still organized as tables and text—not as the relationships, constraints and implications you need to reason through a case. So the lawyer is still left to connect the dots before the next move becomes clear.
The inherited burden
“When you raise the dead, they bring their baggage.”
Databases inherited the library’s filing logic: year, court, citation, subject and keyword. They can find the pieces. They do not, by themselves, show how those pieces form an argument.
Existing legal databases organise law for retrieval — keyword search, classification, citation lists. Atlas organises law for reasoning — by ratio, by argument structure, by logical chains and their higher dimensional cousins.
Powered by Mathematical Reasoning Algorithms
Inside Vidhan Atlas
Vidhan Atlas represents legal principles and argument structures mathematically. The reasoning core explores those structures; it is not a language model generating the next plausible sentence.
Conceptual workflow — not a product demonstration.
Translate facts and legal principles into mathematical representations.
Apply mathematical reasoning algorithms to examine argument paths, counters, and risks.
Inspect the premises, supporting authorities, and conclusions before deciding what to use.
Language models have two restricted roles: translating small, bounded pieces of language into mathematical representations and orchestrating the mathematical algorithms. Retrieval supplies authorities; the strategy is built by the mathematical reasoning core.
Vidhan exposes the premises, reasoning steps, supporting authorities, and conclusions so they can be inspected against current legal sources.
Develop argument paths connecting facts, legal principles, and supporting authorities.
Explore counterarguments and examine where an argument may fail.
Examine factual and legal vulnerabilities that could change your strategy.
Consider next actions in the context of your argument and its risks.
Planned: distinguish a judgment’s ratio from facts, obiter, and procedural detail.
Planned: identify judgments that may conflict with the authorities supporting an argument.
Builds the chain that connects your facts to your desired outcome — with paragraph-level citations at every step.
Stress-tests your argument against the strongest counters — and tells you where the chain breaks.
Surfaces the threats your argument is exposed to — a document the opposition could produce, a witness who could shift the bench.
Recommends the next move — what to file, what to discover, who to prepare — ranked by impact on your case.
Extracts the binding ratio from any judgment — separated from obiter, facts, and procedural detail.
Flags overruling, distinguishing, or conflicting judgments your authorities depend on — before opposing counsel does.
The team
Our founders bring mathematical research and systems engineering together, with legal advisers helping connect the work to practice. Their credentials explain how the reasoning architecture is built.
Co-Founder (CEO)
Full-stack systems builder across automotive, media, medical imaging, and defence.
Math PhD, Princeton · IIT Bombay
Co-Founder (CSO)
Mathematician, ML engineer, and entrepreneur.
Math PhD, IISER Pune · IIT Madras · Cambridge ELLIS · TIFR postdoc
Founder
Math PhD, Princeton · IIT Bombay · INSA Young Scientist · Former Mathematics Department Chair, IISER Pune · Fellow of the Indian Academy of Sciences
Former District Judge, CJJD & JMFC
Senior legal adviser and mentor
LLM graduate and practising advocate
Legal adviser
Early access
Apply to explore argument and strategy generation with Vidhan. We review each application for fit.
Free until MVP testing ends or the agreed usage limit is reached, whichever comes first. Limits are communicated during onboarding, before you begin. No long-term commitment.
Apply for Early AccessDescribe your work generally. Do not submit client names, party details, confidential or privileged information.
We're testing the interface with a small cohort of corporate disputes counsel. Apply to join them and see Vidhan work on a real fact pattern.
Vidhan is built for Indian corporate and commercial counsel exploring arguments and shaping legal strategy. It keeps professional judgment with the lawyer and makes the reasoning available for inspection.
Vidhan Atlas represents legal principles and argument structures mathematically. Mathematical algorithms explore how those structures connect with the facts of a matter. Language models have two restricted roles: converting small, bounded pieces of language into mathematical representations and orchestrating Vidhan's mathematical algorithms. Retrieval supplies authorities; the strategy is built by the mathematical reasoning core.
No. Vidhan's core reasoning uses mathematical algorithms. LLMs are restricted to converting small, bounded pieces of language into mathematical representations and orchestrating those algorithms; they are not asked to generate open-ended legal answers. This makes hallucinated outcomes highly unlikely, but “virtually zero” is not guaranteed zero. The practical issue is responsibility: Vidhan makes every material claim, source, and reasoning step easy to inspect, while the legal professional remains responsible for the work. See the Terms of Use.
Describe your practice and use case in general terms. Do not include client names, party details, privileged material, confidential facts, or documents. The application is not a place to submit a matter for analysis. Read the Privacy Policy before applying; questions about data handling can be sent to privacy@vidhan.ai.
The current focus is Indian corporate and commercial law. The coverage page lists the available legislation and the current boundaries for judgments and practice areas. Tell us your area of work in general terms so we can assess fit.
Early access is free during MVP testing, until testing ends or the agreed usage limit is reached, whichever comes first. Usage limits are communicated during onboarding, before you begin. There is no long-term commitment.
We ask you to be available for a 30-minute onboarding call within two weeks of being accepted, try the available capabilities, and share candid feedback. Your experience helps inform what we build next. We review each application for fit.
We review your application for fit and contact you about next steps. If you have a question before applying, contact founders@vidhan.ai.
Questions before applying? Contact the founders.
Bring the strategic question that matters before pleadings harden around a weaker theory. Vidhan explores the argument space and exposes the reasoning, so your judgment can focus on which path to take.
Describe your work generally. Do not submit client names, party details, confidential or privileged information.
