The MCA portal is down again, and your diligence is due Monday

There is a category of legal work that nobody puts on a pitch deck and everybody spends real hours on.

Pulling company documents. Checking charges. Finding the right entity when three of them have almost the same name. Downloading filings one at a time, in a session that times out, on a portal that is having a difficult evening. Nothing about this requires a lawyer. All of it blocks one.

This is easy to overlook when we talk about legal AI. The conversation tends to focus on the parts that sound impressive: reasoning, drafting, agents, autonomous workflows. But a large part of a lawyer's day is still spent getting the information needed to do those things in the first place.

And retrieval is not a trivial problem.

If a diligence team needs forty MCA filings, the work is not simply finding forty documents. It is identifying the right company, making sure the right entity has been selected, retrieving the documents, checking that the downloads are complete, organising them and then getting them into whatever system the team is using for analysis.

Each step is small. Together, they can consume a day.

This is one of the places where we think legal technology should be doing more of the work and we solved it.

MCA V3 documents and reports can be retrieved directly by company, with the entity matched against its CIN before the documents are pulled. The important part, however, is what happens next: A surprising amount of legal technology still works by asking the lawyer to move information from one system to another. Download it here. Upload it there.

The same issue appears in legal research. A corporate question may require a judgment, a provision of legislation, an MCA filing and a recent regulatory action from SEBI or the RBI. The answer does not live neatly inside one database. It lives across sources.

The useful system is therefore not necessarily the one with the most information. It is the one that can bring the relevant information together, show you where it came from and return it in a form you can actually use.

That is increasingly what we think legal AI should be measured against. How many unnecessary steps disappear between a lawyer having a question and having something useful to act on.

If, in 2026, a lawyer still has to stop doing legal work because a website is having a bad afternoon, there is an issue.