Tax season in India is a special kind of chaos.
Every year, between January and March, chartered accountants across the country enter survival mode. Their phones blow up with the same message from 200 different clients: "I'll send the documents tomorrow." Tomorrow never comes. March 31 does.
My CA called me last week with the usual request — Form 16, AIS data, 12 months of bank statements. The standard package. I told him to check his email because my script had already sent everything.
There was a long pause on the line.
"...what script?"
The Problem Every CA Firm Faces
Let me paint you a picture that every chartered accountant in India will recognise.
It is March 15. You have 200 clients. Each client needs to submit roughly 4-6 documents for their ITR filing. That is 800 to 1,200 individual documents that need to be collected, verified, formatted, and filed — all within about two weeks.
Most of these documents live in different places. Form 16 comes from the employer's HR portal. The Annual Information Statement sits on the income tax portal. Bank statements are scattered across 2-3 different bank websites. Investment proofs are buried in email inboxes.
The traditional process looks like this: the CA sends a WhatsApp message asking for documents. The client says they will send them "by evening." Three days later, the CA follows up. The client sends a blurry photo of their Form 16 taken at an angle. The CA asks for a proper PDF. Another two days pass. Multiply this by 200 clients and you understand why CAs age faster than the rest of us during tax season.
The real cost is not just time. It is the cognitive load of tracking who has sent what, what is still pending, and which documents have errors that need to be re-requested. I have spoken with CA firms that hire temporary staff exclusively for document collection during January-March. One firm I worked with was spending roughly ₹80,000 per quarter on a person whose entire job was to chase clients for PDFs.
What I Built: A 40-Line Document Automation Script
Here is what my Python script does every March, completely hands-free:
Step 1: Pull Form 16 data. The script logs into my employer's portal using stored credentials and downloads the Form 16 PDF. For salaried individuals, this is the single most important document and the one that gets delayed the most because people forget their HR portal passwords.
Step 2: Download AIS from the Income Tax Portal. The Annual Information Statement contains every financial transaction the government already knows about — salary, interest income, stock trades, mutual fund transactions. My script pulls this data programmatically and saves it as a formatted PDF.
Step 3: Fetch 12 months of bank statements. This was the trickiest part. Different banks have different portal structures, different session management, and different export formats. I wrote modular functions for the three banks I use — HDFC, ICICI, and SBI. Each function handles login, navigates to the statement section, sets the date range for the full financial year, and downloads the PDF.
Step 4: Package and email. All documents get collected into a single folder, renamed with a consistent naming convention (Form16_ArchitMittal_FY2526.pdf, AIS_ArchitMittal_FY2526.pdf, etc.), zipped, and emailed to my CA with a summary of what is included.
The entire thing runs as a cron job on a ₹500/month VPS. Every March 1, it executes automatically. By the time my CA starts making his annual round of phone calls, my documents are already sitting in his inbox.
Total setup time: one Sunday afternoon. Total recurring cost: ₹4 per execution in compute and API calls.
Why My CA Asked Me to Build This for His Other 200 Clients
When I told my CA about the script, his first reaction was not "that's cool." His exact words were: "Mere baaki 200 clients ke liye bana de." (Build this for my other 200 clients.)
That single sentence told me everything about the automation opportunity in Indian professional services.
CA firms are not short on intelligence. These are some of the sharpest financial minds in the country. What they lack is time — specifically, time that gets consumed by mechanical, repetitive work that has nothing to do with their actual expertise.
Think about what a CA's real value is: tax planning, compliance strategy, audit oversight, financial advisory. Instead, during the busiest quarter of the year, they spend 60-70% of their time on document collection and formatting. That is like hiring a surgeon and having them spend most of their day filling out admission forms.
The numbers get even more interesting at scale. A mid-sized CA firm with 500 clients spends approximately 1,500-2,000 hours per year on document collection alone. At an average billing rate of ₹800/hour, that is ₹12 to ₹16 lakh in opportunity cost — revenue that could have been earned doing actual advisory work instead of chasing PDFs on WhatsApp.
The Bigger Picture: Why Finance Automation is India's Untapped Gold Mine
Here is what most people in the AI and automation space are missing about India.
Everyone is building tools for tech companies. CRM automation for SaaS startups. Code generation for developers. Marketing copy for D2C brands. Meanwhile, the largest chunk of India's professional services sector — accounting, legal, compliance — is still running on WhatsApp messages and Excel files.
India has over 3.5 lakh practising chartered accountants. Most of them serve 100-500 clients each. The vast majority use Tally for accounting, WhatsApp for client communication, and manual processes for everything in between. There is almost no automation layer connecting these tools.
The opportunities are everywhere once you start looking:
GST reconciliation is a monthly nightmare for businesses with high transaction volumes. The GSTN portal data needs to be matched against purchase registers, and mismatches need to be identified and resolved before filing. I have seen businesses spend 3-4 days per month just on GST reconciliation — a process that a well-written script can do in under 10 minutes.
Invoice processing is another massive time sink. Small businesses receive invoices via email, WhatsApp, courier, and sometimes just as verbal agreements. Extracting structured data from these disparate sources and feeding it into accounting software is exactly the kind of work that AI excels at — pattern recognition across unstructured data.
Bank reconciliation for businesses with multiple accounts across multiple banks involves downloading statements, normalising formats, matching transactions, and identifying discrepancies. Every CA firm does this manually. Every single one.
How to Get Started: A Practical Roadmap
If you are a developer looking to build in this space, or a CA firm wanting to start automating, here is what I recommend.
Start with document collection. It is the lowest-hanging fruit with the highest time savings. Build a simple script that can download Form 16, AIS, and bank statements for a single person. Then generalise it. The technical complexity is moderate — mostly web scraping, session management, and PDF handling — but the time savings are immediate and dramatic.
Talk to your CA. Seriously. Have a conversation about their workflow during tax season. Ask them what takes the most time. You will be surprised at how many problems are solvable with basic Python scripting and API integrations. CAs understand the value of automation intuitively because they deal with numbers all day — they just have not had someone translate their problems into code.
Think in workflows, not tools. The mistake most automation builders make is starting with a tool and looking for problems it can solve. Instead, start with a workflow — the exact sequence of steps a human follows — and then figure out which steps can be automated. Some steps genuinely need human judgment (like deciding whether to claim a specific deduction). Others are purely mechanical (like downloading a PDF from a portal). Automate the mechanical ones ruthlessly. Leave the judgment ones to the humans.
Build the "pause and think" step. From my own experience of over-automating my trading routine, I have learned that the best automations always include a checkpoint where a human reviews the output before it moves forward. In the case of ITR automation, this means the CA should review the collected documents before filing — never auto-file without a human in the loop.
The Bottom Line
March 31 is here. Somewhere right now, a CA is manually downloading their 150th Form 16 of the season while WhatsApp notifications pile up with clients asking "bhaiya, meri filing ho gayi kya?" (brother, is my filing done yet?).
Next year, that CA could have a script doing the mechanical work while they focus on what actually matters — advising their clients on how to save more and comply better.
The best time to automate tax season was last year. The second best time is right now.
Archit Mittal is the founder of Automate Algos. He helps businesses automate chaos using AI agents and custom workflows. Connect with him on LinkedIn @automate-archit.
Top comments (0)