Questions people ask
Answered the way we would answer them on a call. Where a question depends on your specifics, we say so rather than giving a number that turns out to be wrong.
AI
What can AI actually do in a business like ours?
Four things, reliably. It reads your paperwork — invoices, purchase orders, KYC sets, reports — and enters the data instead of a person retyping it. It answers the questions your team answers all day, on WhatsApp or your website, from your live data. It predicts numbers you currently estimate, like what to stock or which receivable is about to go bad. And it does the coordination work between steps: reading an email and creating the order, matching a payment to an invoice, routing an approval to the right person.
What it does not do is run your business or replace a system you do not have. If your process is still on paper, the first step is usually software, not AI.
How do we know whether our process is a fit for AI?
Three questions. Is the task repetitive and high volume? Is the information already in a digital form, or capturable as one? And can you live with it being right most of the time rather than always, with a person checking the rest?
If all three are yes, it is usually a fit. If the third is a no — anything where a single error is unacceptable and undetectable — then AI belongs in a supporting role at most. We will tell you which of your processes fall on which side before you commit to anything.
What is the sensible way to start with AI?
One narrow process, measured. Pick the task that is highest volume and most repetitive — usually document entry or first-line customer questions — and run it on your real material alongside the existing manual process for a few weeks. You get an accuracy figure from your own data rather than a claim from a vendor, and the cost of finding out is small.
What we would advise against is an open-ended “AI transformation” programme. Those are expensive, slow to show anything, and usually end with a pilot nobody deployed.
How accurate is it, and what happens when it gets something wrong?
Nobody honest will quote you a single accuracy figure before seeing your actual documents and data. What we do instead is measure it on your real material during a pilot, and tell you the number we got.
The design matters more than the number. Every system we build sets a confidence threshold: anything below it goes to a review queue for a person rather than being posted silently. Anything touching money gets checked. You will always be able to see what the system decided, what it was unsure about, and what a person changed.
Do you build the AI yourselves, or resell somebody else’s?
Neither, and the distinction matters. We do not train our own models — almost nobody outside a handful of large labs does, and any Indian software company claiming otherwise is worth questioning. We build on OpenAI’s API.
What is ours is everything around it: how your documents and data are prepared and fed in, how the output is validated before it reaches your system, what happens to the cases the model is unsure about, and how the whole thing connects to the software you actually run. That application layer is where an AI project succeeds or fails, and it is the part we build.
Where does our data go when you process it with AI?
AI processing is performed through OpenAI’s API. Your data is sent to that API for the specific task, and the result comes back into your system.
Before any project involving your data begins, we confirm the current provider data-handling terms in writing — retention, whether anything may be used for model training, and where processing takes place — so that it forms part of your agreement rather than an assurance in a sales call. If those terms do not work for your business, the honest answer is that this is not the right project for you, and we will say so at that stage rather than later.
Can the AI run entirely on our own servers?
Not today. Our AI work runs through a hosted API, so the processing happens outside your infrastructure even when the rest of the system sits on your own servers.
If your policy requires that nothing leaves your network at all, tell us at the first conversation. The non-AI parts of what we build — the ERP, the portal, the automation — can run entirely on your infrastructure, and for many businesses that is the sensible split.
Do we need clean data before any of this works?
For document reading and for agents, no — those work on documents and conversations as they arrive.
For forecasting and scoring, yes, and this is where we would rather lose the project than take it. Prediction needs one to two years of reasonably consistent transaction history. If your data is scattered across systems, full of duplicates, or was entered inconsistently, the honest first step is fixing that. A forecast built on bad history is worse than no forecast, because people act on it.
Can it handle Hindi and Assamese as well as English?
Yes, including messages that switch between them mid-sentence, which is how most customers in this region actually write. Accuracy is highest in English, good in Hindi, and workable in Assamese and Bengali. We test on your own message history before go-live and show you where it struggles, so you know what to expect rather than finding out from a customer.
Working with Arsaviva
What does Arsaviva actually do?
We build custom software for businesses — ERP, CRM, accounting, inventory, HR and operations systems — along with websites, e-commerce platforms and mobile apps, and we build AI into them where it earns its place. Alongside client work we run five SaaS products of our own, which is unusual for a services company and is the main reason we build the way we do.
We started in 2017 and have delivered more than 650 projects for over 300 businesses, from Guwahati.
How do you price a project?
Against a written scope, agreed before any code is written. We do not publish indicative figures, because the same-sounding requirement can differ several times over in cost depending on the specifics.
What drives the number: how many distinct user roles there are, how many screens and reports, how much data has to be migrated from existing systems, how many other systems it must talk to, and whether your process is already documented or has to be worked out first. A first conversation is usually enough for us to tell you the rough order of magnitude.
How long does a project take?
It depends on how much of the system you want live at once, which is why we usually suggest starting with one module rather than everything.
What holds true regardless is the shape of it: you see something working every two weeks from the start of the build. You are never waiting three months to find out the project went wrong in week two.
What happens if our requirements change halfway through?
They will, and that is normal. What matters is that it is handled openly instead of quietly absorbed until somebody runs out of goodwill.
Anything that adds real work gets written down as a change request with its own cost and its own effect on the date, and you decide whether it is worth it before we do it. Nothing gets built because somebody mentioned it on a call, and nothing gets billed that you did not approve.
Can we start with one module and expand later?
Yes, and for most businesses it is the better way in. A first module that is live and working tells you more about whether we are the right people than any proposal does, and it limits what you risk finding out.
The one condition is that the whole system is designed at the start even if it is built in stages, so the second module does not require the first to be rebuilt.
Do you work with businesses outside Assam and the Northeast?
Yes. We are based in Guwahati and a good share of our work is regional, but we build for clients across India and our products are used well beyond the Northeast. Projects run over video calls and shared environments, with on-site visits at the stages that genuinely need somebody in the room — usually discovery, go-live and training.
How do you stop our staff from going back to Excel?
By building it around what they already do, and by talking to them during the scope stage rather than only to the person signing the cheque. Most systems are rejected because they add work for the people entering the data while the benefit lands somewhere else.
Practically: fewer screens, fewer mandatory fields, capture what can be captured automatically, and make the daily task faster than the spreadsheet it replaces. Then train in small groups and stay close for the first month, which is when the real questions arrive.
Ownership and risk
Who owns the source code?
You do, on final payment. Source code, database, documentation and intellectual property in the work are assigned to you in the contract. There is no licence that expires and no situation in year three where you discover the software is not yours.
We also hand over in a usable state: documented, commented, and written so another developer can pick it up.
Who owns our data, and where is it stored?
The data is yours at all times. Where it is stored is your choice — your own server, your own cloud account, or hosting we arrange for you. If you want the system inside your own infrastructure, that is a normal request and we build for it.
You can export your data in a standard format at any point.
Can it run on our own servers rather than the cloud?
Yes. We build to be deployable either way and the decision is yours. On your own infrastructure you carry the backups, uptime and patching, or you pay somebody to; on cloud you get easier scaling and remote access. Regulated and data-sensitive operations often have good reasons to keep everything in-house, and we build for that without argument. The exception is AI processing, which currently runs through a hosted API.
Do you sign NDAs?
Yes, as a matter of course, before any detailed discussion of your process or data. We will also sign yours rather than insisting on ours.
Custom vs off-the-shelf
Why not just buy a ready-made ERP or CRM?
Sometimes you should, and we will say so. If your processes are standard, a well-configured packaged product is faster to deploy and cheaper to start.
Custom becomes the right answer when your process is the thing that makes you competitive, when the packaged product needs so much customisation that upgrades break it, or when you have already bought one and half the team has quietly stopped using it. The comparison that matters is not licence against build. It is the licence, plus the customisation, plus the work still being done by hand, against the build.
Can custom software work alongside Tally or Busy?
Yes, and for most businesses that is the better answer. Your accountant knows Tally, your auditor expects it, and replacing it adds risk for no gain.
We build a two-way integration instead: masters, vouchers, stock and ledgers move between your operations system and Tally automatically, so the entry is made once. The operations software does what Tally was never meant to do, and the accounts stay where everybody expects them.
Can you take over software somebody else built?
Often, yes, and it is a more common request than anybody admits. We start with a technical audit: what it is built on, what state the code is in, what is missing, and what it would cost to keep it running versus rebuild it. You own the findings whether or not you continue with us.
Sometimes the honest conclusion is that the system should be replaced. We will say that plainly, and we will also say when it should not be.
Can we use one of your products instead of building something custom?
Frequently, yes, and we would rather tell you that than sell you a build. If you need verified field attendance, Trackrr already does it. If you run a vehicle service centre and want customers coming back, Serviciing already does that. If you need WhatsApp campaigns and automation, that is WhatEdge. If you need a CRM, Relead may fit without a line of custom code.
Custom is the right answer when nothing existing fits your process — not when nobody checked.
Websites
How long before a new website starts bringing enquiries?
Paid traffic works the day you launch. Organic search does not. Realistically you are looking at three to six months before new pages rank for anything competitive, and longer for the most valuable phrases.
Anybody promising first-page rankings in thirty days is either buying ads or lying. What we can do is build the structure correctly from day one, so the clock starts immediately rather than after a rebuild.
Not answered here?
Ask it directly. We would rather have the conversation than have you guess from a website.