AI automation for stock broking and sub-broker firms in India

Direct answer: which work AI can take over

Most of the daily load in a sub-broker or authorised person office is paperwork that follows a fixed pattern. A new client sends a PAN card, an Aadhaar-linked document, a bank proof, a signature and a photograph. Someone checks them against the account opening form, phones the client when something is missing, uploads the details, waits for the status, and then tells the client the account is live. Repeat that a few hundred times a month and it becomes the largest single task in the office.

AI workers, meaning software agents that read documents, follow rules you set and send messages through email, WhatsApp or SMS, can run most of that loop. They can read incoming files, mark what is missing against your checklist, send reminders in the client's preferred channel, record each step in the client's file, and move the file on when everything is in place. They can also prepare the daily ledger and reconciliation extracts that accounts staff currently build by hand.

What they should not do is decide. Whether a PEP declaration needs enhanced checks, whether a name mismatch between PAN and Aadhaar is a typing error or a red flag, whether a client on hold should be allowed to trade, and how to answer a complaint are decisions for your compliance officer and your senior staff. The office team's job moves from doing every step to reviewing the cases the system could not close.

The whole workflow, stage by stage

The chain below follows a typical client from first enquiry to the monthly statement. Your firm may split the roles differently, but the order of work is usually the same.

1. Enquiry and lead capture

Today, a relationship manager or branch executive takes calls, walk-ins and referrals, often on a notebook or a spreadsheet, and forwards the promising ones to onboarding. Follow-up depends on who remembers to call back. An AI worker can capture enquiries from the website form, WhatsApp and email into one list, send the first reply with the account opening checklist, and remind the relationship manager when a lead has gone quiet for a set number of days. The relationship manager still makes the sales conversation and decides which products to discuss.

2. Account opening and document collection

This is where most hours go. The onboarding executive compares the form with the documents, notes what is missing, and sends the list back to the client, sometimes three or four times. An AI worker can check each upload against the checklist for the account type, flag blurred scans, expired proofs and mismatched fields, and send one clear message that lists everything outstanding. It can also keep a running status for each client so that nobody asks the client for the same document twice. A person still checks the originals where in-person verification applies and signs off the form.

3. KRA and CKYC upload, and status checks

Once the documents are complete, the intermediary must upload the KYC details to a KYC Registration Agency (KRA). The KYC executive then checks the status the KRA returns. An AI worker can prepare the upload file from the form data, submit it through the agency's portal or the broker's back-end where the broker permits it, pull back the status and sort the file into validated, registered, on hold or rejected. Cases that are not validated go to a person with the reason attached. Section 5 covers why this matters more in 2026.

4. Client code creation and account activation

After KYC is accepted, the client receives a Unique Client Code (UCC) and the account is activated on the trading platform. Staff currently copy codes between the back-end, the client master and the messaging tool. An AI worker can create the record in the client master once the file is complete, match the UCC to the right client, send the welcome and activation message, and confirm the activation back to the file. A person should check that the mapping is correct for the first batch before the worker runs unattended.

5. Ongoing client master and records

Client details change: address, mobile number, bank mandate, nominee, income details. Each change must be recorded and, where needed, updated at the KRA. An AI worker can collect changes sent by email or through a form, check them against supporting documents, route the update to the right person and keep a dated log. It can also flag records that are due for periodic review. The person who approves the change stays in the loop.

6. Trading-day support, margin and funds follow-up

On trading days, the back office deals with margin shortfall calls, pay-in and pay-out queries, contract note questions and bank mandate mismatches. Much of this is a message asking for money or a document, followed by a check of whether it arrived. An AI worker can send margin and payment reminders from the approved template, record replies, and escalate anything that touches client money to an accounts executive. It should never move money or change a limit on its own.

7. Ledgers, reconciliations, statements and reports

Accounts staff match bank statements, client ledgers and the exchange and clearing files, then prepare statements and management reports. An AI worker can pull the bank and ledger extracts, match the entries it can match with confidence, and list the breaks for review. Tally or your existing accounting software remains the book of record. The accounts executive investigates each break, and the accountant signs the final numbers.

Staff today vs with AI workers

The table below lists the office roles that most broking and authorised person firms in India use. Titles vary from firm to firm, so match them to your own org chart. The point is the split of work, not the headcount.

The highest-value automations, ranked by effort and impact

Not every task is worth automating first. The list below ranks the work by how much effort it takes to set up against how much office time it returns. Effort is a judgement based on the number of systems involved and the rule complexity, not a measured figure.

Compliance points for India

Three regulatory points shape how far you can automate and how you should document it.

The first is the new rulebook. The SEBI (Stock Brokers) Regulations, 2026 were notified on 7 January 2026 and replaced the 1992 regulations. The 2026 rules list the records a broker must keep, including client account opening forms, client ledgers, margin ledgers and copies of contract notes, and they require brokers to keep them for eight years, up from five. An automated workflow must keep these records in a form you can produce on inspection, with dates and a clear audit trail. Books may be kept electronically, but the location must be notified to the exchange.

The second is the sub-broker label. SEBI discontinued the sub-broker category as a registered intermediary in 2018, and existing sub-brokers moved to authorised person or trading member status. Most firms that still say sub-broker work as authorised persons under a registered broker, and the broker keeps responsibility for what the authorised person does. Check your own registration documents before you describe your set-up to a client or an automation vendor.

The third is KYC. Under the KRA framework, the intermediary uploads client KYC records to a KRA, usually within three working days of receiving the documents, as commonly summarised in SEBI KYC guidance. In February 2025, the exchange circulars said that clients whose KYC was on hold for a particular upload window could not trade or square up open positions until they complied. A file that sits unflagged for a week can therefore stop a client from trading. Your status tracker must show every on-hold record the same day.

The fourth is the change now under consultation. SEBI issued a consultation paper on 16 January 2026 proposing that supplementary KYC data, such as income range, occupation, politically exposed person status, FATCA information, CKYC ID and the expiry date of officially valid documents, be stored at the KRA level and shared across intermediaries. The paper also proposed a periodic KYC review every five years, with alerts to intermediaries when records are stale. Comments closed on 6 February 2026. We could not confirm from the sources available whether these proposals have been finalised, so check SEBI's website before you change your forms. If they are adopted, a sub-broker office would collect less repeated supplementary data at the front desk, but it would need to watch for review alerts and expiring documents, which is a good job for a workflow.

A separate consultation from August 2026 proposes digital KYC from abroad for NRIs and other persons resident outside India, with comments open until 4 September 2026. If you onboard overseas clients, track that paper as well.

Finally, the cyber security and record rules apply to the automation itself. Any AI worker that touches client data needs access controls, logs and a documented vendor arrangement. Where an exchange or SEBI inspects your records, the reviewer will expect to see who or what changed each file.

How the transition happens without disruption

Most firms that move badly try to switch off a manual process on the same day they switch on a worker. A safer order runs over four to eight weeks, depending on file volume and how clean your records are.

1. Map the current process for one account type, such as individual resident accounts, with the people who do it. Write down every document check and every message sent.
2. Run the AI worker in shadow mode for two weeks. It drafts the reminders and gap lists, but the onboarding executive sends them and compares the output with what they did.
3. Turn on the worker for the routine cases only. Send exceptions, such as name mismatches, on-hold KRA responses and rejected files, to the named person with the reason attached.
4. Move the same pattern to the next account type, then to KRA sorting, then to ledger matching. Keep the staff member who owns each step in place until the error rate is stable.
5. Only after several weeks of stable output, reduce the manual checking time. The roles shrink to review. People are redeployed to complaints, client service and reconciliation exceptions, and any headcount decision is made by you on the basis of your own volumes.

Common mistakes

The first mistake is automating a broken process. If your checklist differs from one executive to the next, a worker will repeat the inconsistency faster. Write one checklist per account type and get the compliance officer to approve it before anything is switched on.

The second mistake is letting the worker close exceptions. A file that is on hold, rejected or flagged for PEP or name mismatch needs a named person to decide. Build the escalation path first, and test it with real cases.

The third mistake is ignoring the data. Automation reads whatever the client master says. If the PAN, mobile number or email is wrong in the system, the reminders go to the wrong place. Clean the top 500 records by hand before you start.

The fourth mistake is treating the proposed KYC changes as settled. Build the status tracker so that new fields can be added without rebuilding it, and check SEBI's published outcome before you change client forms.

The fifth mistake is cutting staff before the workflow is proven. The first gain is usually fewer hours spent on chasing, not fewer people. Let the work show the numbers over a full month, including month end and a volatile trading week, before you decide what the team looks like.

Finally, automation is not worth it for a very small volume of accounts that a single person handles comfortably, or where your records are still on paper and no one has time to digitise them. In those cases, a cleaner checklist and a shared tracker in Google Sheets or Excel may give most of the benefit at lower cost and risk.