Artificial intelligence
Generative AI, retrieval over organisational knowledge, and machine learning where the data supports it.
Technology
Technology is a means. The problem is the point.
Two decades of databases, telecom systems, business software, companies and AI have left Rohit with a practical view: technology earns its place only when it solves a problem that a business, an organisation or a person genuinely has.
Technology philosophy
Start with people and process, end with technology
Quick answer
What is Rohit's approach to technology?
Rohit treats technology as the last layer of a solution rather than the first. He begins with the people who do the work and the process as it really runs, then examines the data and systems, and only then decides whether automation or AI will help. The aim is a solution people will use every day, not an impressive demonstration.
Much of this view was formed in telecom back offices, where a fraction of a percent of wrongly rated calls could mean a large billing dispute, and in small businesses, where a counter clerk would abandon any software that slowed down a queue. Both environments punish technology that ignores how people actually work.
It is also why hype words rarely appear in Rohit's proposals. A plain description of what a system will do, what it will not do and what it depends on is more useful to a decision maker than any adjective.
Areas of expertise
Where the technology work sits
A broad range, held together by one habit: understanding the operation before changing it.
Agentic AI and AI agents
Multi-step AI systems that plan and use tools, designed with limits, logs and human approval.
Intelligent automation
Rules, workflows, document handling and AI combined, with people handling the exceptions.
Software engineering and architecture
Application design, data models, integration and the unglamorous work of keeping systems maintainable.
Enterprise applications
Billing, inventory, operations and administration systems that carry an organisation's daily work.
Digital transformation
Changing how work gets done, with technology in service of a redesigned process.
Technology strategy and consulting
What to build, what to buy, what to leave alone, and in which order.
Data, cloud and security awareness
Data quality, sensible cloud choices and security basics that small organisations often skip.
Technology education
Teaching practical AI and IT skills, guiding projects and mentoring early career engineers.
AI and intelligent systems
Useful AI is mostly careful engineering
Rohit's first model was a neural network trained by backpropagation to predict sunspot activity, during a 2004 internship at IIT Roorkee. The tools have changed beyond recognition since then. The core discipline has not: know what the model has learned, know where it will fail, and design the system around both.
Generative AI has made it possible to read, summarise, classify and draft at a scale that was impractical a few years ago. In organisations that translates into real gains: faster handling of documents and emails, better search across internal knowledge, first drafts that a skilled person can correct in minutes rather than write in hours.
The same systems also produce fluent mistakes. So the practical questions are always about design. Which tasks tolerate an occasional error that someone will catch? Where does the model get its facts? Who reviews the output, and how is that review made easy rather than a formality? What is logged, and who can see it?
Answering those questions well is less dramatic than a demo, and far more valuable.
Agentic AI
How much should an AI agent be allowed to do?
AI agents plan steps, call tools and act. Rohit's work on agentic systems, including the implementation of Workologix for one of the leading logistics companies in the United States, uses a simple ladder to decide how much autonomy each task gets.
Inform
The agent gathers and summarises information. A person decides everything.
Suggest
The agent proposes an action with its reasoning. A person chooses.
Prepare
The agent drafts the action in full. A person approves before anything happens.
Act and report
The agent acts on low risk, reversible tasks and reports every action for review.
Act within limits
The agent acts alone inside tight, monitored boundaries, with automatic escalation.
The default rule
Agents can prepare, draft and check. Anything that moves money, changes customer records, communicates externally or affects someone's health needs a named person to approve it, at least until the system has earned a higher level with evidence.
Software, enterprise systems and transformation
The foundations under every intelligent system
AI sits on top of software, data and processes. When those are weak, AI makes the weakness faster. Most of Rohit's career has been spent on the foundations.
Software engineering and architecture
Rohit's engineering background runs from Oracle schema design and PL/SQL through Unix automation to application architecture for multi-location businesses. The lessons are old and still ignored: model the data carefully, because it outlives every screen; design for the exception, because that is where support calls come from; and keep systems understandable by the next person who has to change them.
Since 2011 his teams have built software for distributors, multi-location clinics, hotels and restaurants, schools and an international franchise network. Each domain looked different on the surface and turned out to share the same underlying problems: identity, inventory, money, schedules and permissions.
- Data models that match how the business actually works
- Integration planned from the start, not bolted on
- Clear audit trails for anything involving money or records
- Performance at the moment of use: the counter, the front desk, the ward
- Documentation and handover that survive staff changes
Enterprise applications and operations
Three years implementing operations and business support systems for MTNL at Xalted, Bengaluru, covered process mediation, interconnect billing, fraud management and reconciliation. Those systems handle huge volumes of records, and the discipline they demand (reconciling, validating, tracing every figure back to its source) applies to any organisation that depends on data being right.
For smaller organisations, enterprise technology means something humbler but just as important: a single reliable source for customers, stock and accounts, and reports that people trust enough to act on.
- Mediation and validation of incoming data
- Reconciliation between systems that disagree
- Controls against fraud and leakage
- Reporting that management can rely on
Digital transformation, done in the right order
Transformation projects fail less often because of software and more often because the process was never redesigned. Automating a confused process produces faster confusion. Rohit's approach is to draw the current process honestly, including the workarounds, agree on the target process with the people who do the work, and only then choose tools.
The essay on drawing the process first explains the method in more detail.
Map the real process
Including informal steps, spreadsheets and phone calls.
Find the friction
Delays, rework, errors and waiting.
Redesign with the team
Remove steps before automating any.
Choose and build
Simple tools first, AI where it clearly helps.
Measure and adjust
Compare against the baseline, not the plan.
Business process and intelligent automation
The first automation in Rohit's career was a set of Unix shell scripts that removed a few hours of repetitive work each week. The principle carried forward: automate the dull, repeated, well-understood task first, and leave judgement to people.
Intelligent automation extends that principle with AI: reading documents, classifying requests, extracting details and drafting responses. It works best as a partnership in which the system handles the predictable majority and routes the unusual minority to a person with the context already assembled.
- Rules and workflows where the logic is clear
- AI for reading, classifying and drafting
- Human review for exceptions and consequential steps
- Logs that show what happened and why
- A way to switch automation off safely
Full page: software and digital transformation Full page: business and intelligent automation
Technology consulting
Advice that includes when not to build
Rohit advises organisations on AI adoption, agentic systems, automation, software architecture and technology strategy, directly and through RCode Intelligence.
Good consulting often saves money by stopping a project, narrowing it, or replacing a custom build with an existing tool. The questions below usually decide more than any technology comparison. Details of previous work and selected references are available on request.
Discuss a technology questionFull page: technology consulting
Questions asked early in a consulting conversation
What problem costs you the most today?
In time, money, errors or customer frustration, and how you know.
What happens if nothing changes?
Sometimes the honest answer is that nothing urgent does.
Who will use the solution every day?
And what they currently do instead.
What data exists, and how reliable is it?
AI and automation inherit every flaw in the data.
What would a wrong output cost?
This decides how much human review a system needs.
Entrepreneurship and product thinking
Building companies from Roorkee since 2011
Starting a technology business in a small Indian city teaches product thinking quickly, because every customer is close enough to tell you exactly what is wrong.
Rohit co-founded Acmez Business Solutions with Dr. Aarzoo Saini on 6 April 2011, and the business was incorporated as Acmez Technologies Pvt. Ltd. in 2017. AARHIT Systems followed the same year, then Webyfied Global in 2024 and Sciematics Insights and AIIT Roorkee in 2025. Each was started to meet a specific need rather than simply to add another company.
Product thinking, in this experience, is mostly about restraint: shipping the smallest thing that solves a real problem, watching how it is used, and resisting features that only one loud customer requested.
- Sell the problem you solve, not the technology you use
- Support calls are the best product research available
- Small, steady customers build durable businesses
- Hire people who can learn, then teach them well
- Keep the founder close to the code and the customer
Co-Founder, Director and Chief Technology Officer
The software and technology company where Rohit's entrepreneurial work began.
Co-Founder, Director and Chief Technology Officer
An AI, intelligent automation and applied research company.
Founder
A digital services company working on websites, applications, commerce and search.
Founder and Chief Technology Officer
An artificial intelligence, data and cybersecurity company based in Roorkee.
Founder
A technology education institute for AI, IT skills, research and projects.
Principal Consultant
Rohit's independent practice for AI and data science consulting.
Emerging technologies
Following new technology without chasing it
New tools arrive every month. The useful filter is whether a tool changes what a real organisation can do, at a cost and risk it can carry.
Cloud and web
Sensible hosting, modern web applications and the discipline to avoid needless complexity.
Data and analytics
Clean, owned data first. Dashboards and models only make sense once the numbers can be trusted.
Cybersecurity awareness
Access control, backups, updates and staff awareness: the basics that prevent most incidents.
What comes next
Smaller specialised models, AI on local devices and better evaluation methods, watched with interest and tested before trusted.
Teaching and mentoring
Teaching technology by building it
Rohit founded AIIT Roorkee in 2025 to teach practical AI and IT skills, guide student research and projects, and help learners prepare for work.
The teaching approach mirrors the consulting approach. Students start from a problem, build something that runs, find out where it breaks and fix it. Theory matters, and it sticks better when it arrives as the explanation for something a student has just seen fail.
Mentoring conversations with early career engineers and founders tend to return to the same themes: learning fundamentals that outlast frameworks, writing clearly, and understanding the business around the code.
Rohit also gives talks and workshops on these subjects; topics and formats are on the Media page.
For students
What should a technology student learn first?
Programming fundamentals, data structures, databases and networking, followed by one real project taken all the way to deployment. AI tools are most useful to people who already understand what the generated code is doing.
For professionals
How should an experienced engineer approach AI?
As a new layer on existing skills rather than a replacement for them. Retrieval, evaluation, prompt design and agent limits are learnable engineering topics, and system design experience makes them easier to do well.
AI and intelligent systemsGenerative AI can read, summarise, classify and draft at a scale that was impractical a few years ago.Agentic AI and autonomyAI agents plan steps, call tools and act.Software and digital transformationAI sits on top of software, data and processes.Business and intelligent automationThe first automation in Rohit's career was a set of Unix shell scripts that removed a few hours of repetitive work each week.Technology consultingRohit advises organisations on AI adoption, agentic systems, automation, software architecture and technology strategy, directly and through RCode Intelligence..Teaching and mentoringRohit founded AIIT Roorkee in 2025 to teach practical AI and IT skills, guide student research and projects, and help learners prepare for work..
Insights
Technology insights
Essays on AI, software, transformation and building companies.
Where AI Agents Help, and Where They Should Stop
A practical framework for agentic AI: which tasks suit AI agents, how to set levels of autonomy, and why people should approve consequential actions.
Before You Automate, Draw the Process on Paper
A practical method for digital transformation: map the real process, find the friction, redesign with the team, then choose automation or AI.
What Telecom Billing Taught Me About Data Quality
Lessons from implementing telecom OSS/BSS systems for MTNL on mediation, reconciliation and data quality, and why they matter for AI and analytics.
Questions
Questions about Rohit's technology work
What areas of technology does Rohit work in?
Rohit works in software engineering and architecture, enterprise applications, business process automation, digital transformation, artificial intelligence including generative AI and agentic AI, technology consulting and technology education. His background includes Oracle database development, Unix scripting and telecom OSS/BSS implementation.
What is agentic AI?
Agentic AI describes AI systems that pursue a goal through several steps: planning, using tools such as search, databases or other software, checking intermediate results and deciding what to do next. Unlike a single question and answer exchange, an AI agent acts. That makes clear limits, logging and human approval for consequential steps essential.
Does Rohit offer technology consulting?
Yes. Rohit advises organisations on AI adoption, agentic systems, intelligent automation, software architecture and technology strategy, mainly through RCode Intelligence and the companies he co-founded. Details of work and selected references are available on request.
How does Rohit decide whether AI is the right tool?
He starts with the process rather than the technology: who does the work, where errors and delays happen, what data exists and how costly a wrong answer would be. AI is recommended where the task tolerates occasional mistakes that can be caught, where there is enough reliable data, and where a person can review consequential outputs. Where rules or simple software will do, those are preferred.
Conversations
Working through a technology decision?
Share the situation in a few lines: the organisation, the problem and what has been tried. Rohit reads every message personally.
- AI adoption
- Agentic AI
- Automation
- Architecture
- Technology strategy
- Training