CAMPAIGN
Campaign
Paste the ad, landing page, or deck. Must flags promised returns, missing statutory warnings, and cherry-picked performance.
AI-native compliance
What you must do.
What you must not.
Must reads the live SEBI corpus and reviews your product launch, branding campaign, or codebase. Answers come with citations — not last year’s blog post.
must · sebi check
Input
Verdict
Paste what you are about to ship. Must retrieves the matching circulars and regulations, then answers MUST / MUST NOT / MAY with citations.
4,343
SEBI documents indexed
190k
Searchable chunks
2,974
Circulars, including 2026
Cited
Title, date, circular, URL
Three surfaces. One corpus.
v1 is the index. v2 is the product check — the same retrieval, pointed at what you are about to ship.
LAUNCH
Launch
Drop the product memo. Must maps each feature to a current obligation — LODR, intermediaries, MF, FPI — before you ship.
CODE
Code
Point it at copy in the repo. The investor sees the string in production, so that string is the advertisement.
Why this is not ChatGPT with a prompt.
01
Official corpus
We scrape sebi.gov.in/legal.html — acts, rules, regulations, master circulars, gazette, guidance — and keep PDFs as source of truth.
02
Retrieve, don’t memorize
A generic model will sound sure and still miss the 2026 Advertisement Code. Must searches the live index, then answers.
03
MUST / MUST NOT / MAY
Every claim carries the instrument: title, date, circular number, URL. If the corpus is silent, Must says so.
| Generic model | Must | |
|---|---|---|
| 2026 MF Advertisement Code | “I haven’t read the schedule.” | Fifth Schedule, cited. |
| Performance in an ad | “Use standardized periods.” | CAGR 1 / 3 / 5 / inception + ₹10,000. |
| Source | News and firm blogs | sebi.gov.in PDFs |
Early access
Bring a campaign, a launch, or a repo.
We are validating with teams who will actually run a check. Not legal advice — a research layer over public SEBI publications, built to sit in front of counsel.