Notes from the work, written to be useful.

Custom software is priced as hours times a rate. Typical hour ranges for websites, web apps, mobile apps, SaaS products, AI agents and integrations, what moves the number, and how to compare quotes.

A Global Capability Centre is a team in another country that does technology, operations or analytics work for its parent company. How the captive, build operate transfer and partner models compare, and when a mid-sized company should start with a partner.

The DPDP Act sets penalties of up to INR 250 crore per breach. The full schedule of penalties, how the Data Protection Board inquires and decides, the factors it weighs, appeals, and the steps that lower your risk.

What the WhatsApp Business Platform allows, the 24 hour service window and template rules, which tasks are worth automating, how to hand over to staff, and how consent works under the DPDP Act.

How software connects to TallyPrime, which entries are worth automating, how to map ledgers and avoid duplicates, and how to read invoices into Tally with document AI and a person approving the exceptions.

RPA repeats fixed steps on screens. AI agents read messy inputs and choose the next step. How to tell which one a process needs, when to combine them, and how to measure either one.

Plenty of AI pilots impress in a demo and then go nowhere. Four steps, with a clear gate after each, give yours a better chance of reaching daily use.

How overseas companies work with an Indian software partner: engagement models, time zones, contracts and IP, data protection, invoicing and tax, and how to test a partner with one block of work.

Consent Manager provisions start in November 2026 and most duties apply from May 2027, as currently notified. A realistic plan for your first three months of work.

Most dashboards are built, admired for a week and forgotten. A dashboard an owner reads every morning has a handful of numbers, each tied to a decision, and data that is right.

Most DPDP obligations apply from May 2027, as notified. A phased plan that lets a small or mid sized business get ready in steps it can afford, starting with knowing what personal data it holds.

DPDP compliance is not a one time project. Consent, rights requests, breaches and audit evidence happen every day. How TrustOS, our DPDP compliance platform, turns those into routine work.

The forward deployed role is turning up in more and more AI teams. What it means day to day, how it differs from a developer or a consultant, and when it is the right choice.

No model is best at everything. Choose by the task, how sensitive the data is, what it will cost at volume and where the system has to run.

Most service contracts put the risk on the client. Ours moves it to us. Why we work this way, and how it stays fair without turning into a dispute.

The rules every Code Colonies team works to, from the first commit to production. For clients who want to know what to expect, and for engineers considering joining.

How a strategist turns a vague request like 'use AI in finance' into a measured use case that engineers can build and leaders can believe.

Most MSMEs outgrow spreadsheets around Tally long before they outgrow Tally itself. How to tell which problem you actually have, and which of the four routes fits it.

Most AI pilots that stall do so for reasons that were visible before the first line of code. A five gate model makes those reasons explicit and decides when to stop.

Most plants can measure overall equipment effectiveness on the machines they already own. How to connect older equipment, what to measure first and where projects go wrong.

Prompt injection, data leaks and agents with too much power are the usual ways AI applications go wrong. What each risk means and how to reduce it.

Retrieval augmented generation is easy to demo and hard to get right. Whether answers can be trusted depends on unglamorous details.

Most implementation disputes trace back to a contract that was vague about scope, data, acceptance or exit. A checklist to work through with your vendor before you sign.

Not every AI application needs to be written from scratch. When an application builder such as AksharAI is the faster, safer route, and when custom code still wins.

A practical layout for adding AI features to a modern web application, from the interface down to evaluation and monitoring, and the reasons for each layer.

Both put outside people in your team. The difference is what they are accountable for, and that shapes the result more than anything else.

Many Indian companies already work to GDPR. How much of that effort carries over to the DPDP Act, and where you will need to do more.

An agent takes actions, so a mistake costs more than a wrong answer. A practical way to measure agents before launch and keep measuring after.

When stores, a website and marketplaces each keep their own stock count, you oversell online and sit on dead stock in branches. How to get to one number everyone trusts.

Many agent evaluations measure the wrong thing, on the wrong cases, once. A framework for building evaluations whose results actually predict how an agent will behave in production.

Customers expect a live update on WhatsApp and a photo when their goods arrive. How transporters and distributors can add electronic proof of delivery, better route planning and customer messages without a big platform change.

You do not need a data science team to get value from AI. Five practical starting points for small and mid sized businesses in India.

The security basics every software team should have in place before it handles customer data at any real scale.

How we train the engineers and strategists who work on client projects, and why we chose on the job training over classroom courses.

Consulting and deal teams lose time and context across email, spreadsheets and shared drives. How Synerger keeps engagements and transactions in one place from the first call to the closed deal.

If your software or its AI output reaches users in Europe, the EU AI Act may apply to you. A short guide to roles, risk levels and first steps.

Model usage bills can creep up month after month. A short checklist to keep them predictable without hurting quality.

Indian sales teams sell mostly on WhatsApp and the phone. A CRM only works if it captures those conversations without asking salespeople to type everything twice.

Time and materials contracts put delivery risk on the buyer. Output based contracts move it to the supplier, but only work when acceptance is objective. A framework for writing them.

Paper permits and incident forms slow work down and hide risk. A practical guide for oil and gas operators on digitising permit to work and HSE reporting without weakening control.

A short checklist for reviewing pull requests, including code written with the help of AI assistants.

A spreadsheet can run a small godown well. The signs that you have outgrown it, what a warehouse management system really adds, and how to roll one out without stopping dispatch.

The RBI's Digital Lending Directions 2025 set clear rules for loan apps, lending service providers and the banks and NBFCs behind them. What your software must do to stay within them.

AI business cases are often inflated. A plain, defensible way to estimate the value of an AI use case before you build, and to prove it afterwards.

Most B2B customer portals are launched with an email and then ignored. The ones that get used answer the three or four questions customers otherwise phone you about.

Good integrations start with good APIs. A checklist for APIs that partners, internal teams and AI agents can depend on.

As catalogues grow into thousands of variants across warehouses and channels, many online stores start to strain. How ShopenUp handles storefronts, inventory, payments and marketing as you scale.

How to prepare so a penetration test spends its days finding real risks.

Indian payroll mixes central and state rules that change every year. What an HRMS must handle, where payroll usually goes wrong and how to switch systems without a bad month.

AI systems depend on context that lives inside the client's teams and systems. An operating model for embedding engineers and strategists where the work is, and handing over cleanly.

If your new system touches GxP data, it has to be validated. What 21 CFR Part 11, EU Annex 11 and a risk based approach mean in practice for pharma teams building or buying software.

Enterprise buyers send long questionnaires. The documents and evidence to prepare so you can answer quickly and honestly.

Developers juggle RERA disclosures, booking paperwork, demand letters and buyer questions across spreadsheets and phone calls. How one system for sales, collections and customer updates can work.

If your invoices come from an ERP, a billing app or a custom system, it must talk to the IRP and the e-way bill portal reliably. How the integration works and what to watch.

Packaged software, low code and custom builds each fit different problems. A decision path that helps owners and CTOs choose, and recognise when composing the three is the better answer.

Ayushman Bharat Digital Mission integration lets your HMIS or clinic app link ABHA, share records with consent and register with the national registries. What the work involves and how long it takes.

When technicians get jobs on WhatsApp and close them with a phone call, you lose track of time, spares and customer promises. What field service software should do and how to introduce it.

Schools and coaching institutes often run fees in one tool, attendance in another and parent messages from teachers' phones. How to bring them into one system that staff and parents will use.