Schematise's year in review
Schematise had a quiet but constructive financial year 2025-26. Amongst other endeavours, it built its data annotation capabilities, a process that with the help of Aditi Kamble (currently engaged as a Legal Data and Research Consultant), Aswin Pradeep (proprietor of Lamsta) and the wonderful team at Essentia.dev (more details on our collaboration coming soon) was fruitful in determining our scope of operations.
Also, our old friend Daksh Jain did us a solid one with the logo design (see the top of the post for the image.
The process mentioned above involved working on the aspect of determining the scope with which we would employ Large Language Models in our data annotation pipelines. We think it reasonable to concur with most other guidelines on responsible AI usage that using LLM’s in a field like analysis of laws must involve human review, and should only be utilised for work where the risk of hallucination is low.
Our strategy for deploying applications and how it's going
As an organisation, Schematise has placed value on a few core philosophies - amongst them, the commitment towards building upon the capabilities of current technology that serve the legal domain is one.
Naturally this would entail the avoidance of duplication of efforts. Hence, a large part of the work done this year was an exploration of practicability of this motive - we now have our own instance of the Indigo Platform which we are using to generate and publish laws.
We are also exploring managed self-hosting options which we use to test our data connectors on applications such as LibreChat and protocols such as RSS. While this is still very much an idea in development, the key factor has been using this internally first to create legal applications that are interoperable.
Schematise's contributions to AI policy
We have also contributed to a joint report with Extensity AI and Indo Pacific Legal Research LLP titled Artificial Intelligence, Market Power and India in a Multipolar World. Although this work was authored only by Sankalp Srivastava (Founder, Schematise LDA), he used his learnings from AI engineering he had gathered. To this effect, he also participated in a joint discussion with Extensity AI where he summed up his contribution from the report.
However, with the aim of writing more acutely and working within the realm of AI policy with a clear benefit to Schematise’s interests, mostly of those aligned with “copyleft” licenses and the surrounding political economy of AI, Sankalp decided to part ways with the ISAIL Advisory Council. It is foreseeable that using the capabilities of data annotation done thus far, Schematise will contribute further to AI policy by its own merit.
Plans for the future
- Publishing ICAT v2 - This could involve creation of an update version of our dataset for Indian Contracts in Adjudicated Texts using the repository at GitHub
- Using data annotated using our Indigo Platform instance and possibly selling them as low-cost versions of Indian laws.
- Working further on extracting data and representing it as an ontology on the platform "Mutaabik Wiki"
- We also have plans to use datasets that are in public and provide an API connection to their "Schematic" inference using Cypher queries - while this is very much a project in development, it requires some support, so we hope to do it with the help of our clients.
