Image used for representational purposes only
| Photo Credit: Getty Images
The Madras High Court has decided to seek the assistance of Artificial Intelligence (AI) to help it save time by retrieving relevant information from voluminous documents in a commercial dispute between Mumbai-based Gammon-OJSC Mosmetrostory JV – which had constructed some of the underground Metro stations in Chennai – and Chennai Metro Rail Limited (CMRL).
Justice N. Anand Venkatesh has zeroed in on the algorithm Superlaw Courts, with the concurrence of the counsel representing both parties, after its working method was demonstrated in the court. The court was informed that the algorithm was designed to help legal professionals locate, organise, and understand information contained strictly within the documents placed before it for a particular matter.
The algorithm was intended to reduce the time and effort involved in manual searching and cross-referencing of voluminous documents. It was, however, not intended to replace legal reasoning, judicial determination, or a counsel’s professional judgement in a given matter. “Superlaw Courts operates under controls analogous to those expected in court proceedings,” the judge was told.
It was also brought to the notice of the judge that Superlaw Courts works exclusively on documents provided for the matter and that it does not consult external sources, general knowledge, or materials outside the record. If the record does not contain information in a traceable form, the system states that such information was not found in the documents provided, rather than generating an unsupported response.
“The system does not draw conclusions, assess credibility, interpret intent, or express legal views. It only presents what the documents state. Further, the outputs are based on identifiable portions of the record, allowing lawyers and the court to independently verify the context,” Justice Venkatesh said and recorded that the counsel, too, were satisfied with the working method of the algorithm.
He further took note that when documents get uploaded on Superlaw Courts, the system creates a dedicated digital workspace, comparable to a sealed record room in judicial chambers, and then the documents get processed to ensure they were workable. Scanned or image-based documents get converted into searchable text using Optical Character Recognition (OCR) so they could be searched.
Thereafter, the records get organised by grouping all similar materials, identifying duplicates and maintaining a consistent internal order before they get broken into meaningful excerpts through a process known as chunking. Instead of splitting documents mechanically by page or length, chunking preserves logical units such as clauses, paragraphs, or coherent passages, similar to preparing reliable extracts from a brief.
“An internal index is created over these excerpts, functioning like a combined subject index, list of names, and date guide, enabling rapid location of relevant material. When a legal professional asks a question, the system follows a disciplined process of understanding the query in the relevant context and retrieving the data from the most relevant excerpts. This limits error and ensures traceability,” the judge said.
The AI partner’s role was limited to re-expressing or summarising the retrieved excerpts in clear language. It would not introduce new facts, make assumptions, or apply legal reasoning. If the retrieved material does not support an answer, the system states so explicitly. In essence, Superlaw Courts functions as an exceptionally organised and cautious record assistant, the judge added.
“This is the first case where the assistance of Artificial Intelligence is going to be used by the court. Hence, it was agreed that a draft order will be prepared containing the facts of the case and arguments put forth by both sides, which will cover the pleadings, evidences, and findings of the arbitral tribunal in the commercial dispute. Till this stage, the draft order will be circulated to both sides. Upon such circulation, the reliance placed on Artificial Intelligence will also come to an end,” Justice Venkatesh said.
He asked the counsel representing both sides to work on the algorithm for about 10 days and decided to hear the case next on February 12, 2026.
Published – January 31, 2026 03:44 pm IST
