In most mid-sized practices, return season runs on article assistants and long evenings. Clients arrive in waves, each one a folder of bank statements, and someone has to turn all of it into numbers that will hold up if the assessing officer asks a question two years later. The bottleneck is almost never the computation itself. It's everything that has to happen before you get there.
That part of the workflow has barely moved in fifteen years. It's finally starting to, and it's worth understanding why, and what to actually check if you're looking at new software for the firm.
The two problems that don't go away
Manual data entry doesn't scale. However tight your process is, turning a client's bank statement into categorised transactions by hand is the single largest use of staff time in the whole engagement. It's also close to the least valuable use. A junior spending two days tagging entries is doing work a machine finishes in a minute, and not learning much along the way. Across forty or sixty clients, that's weeks of the season gone.
Client data has quietly become a liability question. A firm holds financial records for dozens of people. The moment that data goes up to a cloud tool, it lives on someone else's server, under someone else's security policy, exposed to someone else's breach. Clients have started asking where their statements actually go. "On a server somewhere abroad" is not the answer most of them want, and after the DPDP Act it isn't an answer a firm should be relaxed about giving either.
What offline-first means in practice
ITRSimple keeps statement import and transaction classification on the local machine. Nothing uploads. There's no network round-trip, so there's no waiting on a slow line in the last week before a deadline, and no outage that stops the whole team at once.
It also closes off a whole category of client conversation. "Where does our data go?" now has a one-word answer: nowhere. It stays on the machine you're working on.
AI classification changes where the time goes
The point of an AI engine that classifies every transaction in a statement isn't only that it beats manual entry on speed. It moves your team's hours somewhere more useful.
Instead of a junior tagging every line as income, expense, transfer, loan or tax paid, the engine does that first pass and sends only the entries it's unsure about to a review queue. Your staff reviews exceptions. A statement with 600 transactions might surface fifteen that genuinely need a person to decide. If you want it, there's a fuller explanation of how the AI classification actually works.
What's left for the team is the work that actually needs a CA's judgment: confirming the classification is right, checking the computation, catching the thing the client forgot to mention. That's a better use of a trainee's time and a better use of a partner's review.
Built for a book of clients, not one return
Software written for a single filer doesn't stretch to a practice. A firm tool has to assume many clients and many years from the start. ITRSimple is put together that way:
- A separate workspace per client, so two engagements never bleed into each other
- Financial-year-wise computation, with each client's history kept year on year
- ITR-1, ITR-2 and ITR-4 (44ADA): salaried clients, clients with more than one property, and professionals on presumptive taxation, with capital gains on the roadmap
- Client-ready output: P&L and computation statements that export straight to PDF or Excel for the file
Licensing is per client-year, not per feature and not per seat. Every plan, free or paid, carries the full engine. You add capacity as the client list grows; you never buy functionality back. The pricing page lays out the per-client-year model, with a note for firms running larger client counts.
What switching actually looks like
The honest version: you don't migrate the whole practice in a weekend. Most firms start with a handful of clients on the new tool, usually the messier ones with multiple accounts, and run them alongside the existing process for a season. If the classification holds up and the review queue stays short, the rest follow the next year. The free plan is enough to run this exact trial for one client.
Questions worth putting to any vendor
If you're weighing a move off spreadsheets or an older desktop tool, ask the same four questions of whatever you look at:
- Does client data ever leave the local machine? If it does, when, and to where?
- Does the classifier genuinely cut down review, or does it just rearrange the same manual work?
- Can it hold many clients and many years without charging more per feature?
- Does it cover the exact ITR forms your clients file, not a generic subset?
Where this is heading
Tools built for individual filers were never going to scale to a firm's volume. Cloud tools built for convenience were never going to satisfy a client base that has started paying attention to data. The practices switching this year are the ones that found something that answers both at once: speed from the AI, control from staying local. There's more on how the desktop and online versions work on the ITRSimple home page, and on who is building the product on the about page.