AI Is Not Only a Productivity Tool. It’s a Survival Investment.
Five founders show what AI makes possible. Ilaiyaraaja and Vairamuthu show how to get there.
I often hear people debate how much AI improves productivity. Some say 20%. Some say 50%. I think we are asking the wrong question.
AI is not a tool that makes our old work a little faster. It lets us reimagine the work itself. The founders I work with at Mudhal have redesigned their processes, automated workflows, removed tasks and bureaucracy, and in some cases made work fully autonomous.
Nvidia CEO Jensen Huang put it well: “You’re not going to lose your job to an AI, but you’re going to lose your job to someone who uses AI.” I believe the same is true for businesses of every size. A business will not lose to AI. It will lose to a competitor who uses AI well.
This article has two parts. First, five stories from Mudhal portfolio companies that show what is possible. Second, what I have learned about adopting AI and getting the most out of it, with help from two legends of Tamil cinema.
Part 1: What is possible
The founder who built his own ERP
Rajan of Sirimiri Health surprised me the most. In January 2026, he told me none of the ERPs in the market met all his needs. In April, he told me he had started vibe coding his own. At the end of July, he showed it to me, live in production.
Rajan started on February 12 and worked on it a few hours every day. In about five months, single-handedly, he built procurement, goods receipt, stores, production and finished goods inventory, production planning and management, field sales, sales orders, invoicing, distributor management, and integrations with e-commerce and his accounting software. He went live on July 15.
The whole system took him about 500 hours. In my experience, most business owners spend more than 500 hours just evaluating, selecting and implementing an ERP. Rajan built one that fits his business exactly.
The ERP has automated all his operations by capturing every piece of data about his business resulting in a single source of truth. It gives him real time visibility into production, sales and inventory, helping him make better quality data driven decisions.
Rajan and Vijaya, the co-founders of Sirimiri, are very particular about cleanliness and deeply committed to quality. Rajan is now using AI to automate quality control at the factory. Every machine must be cleaned at scheduled intervals. Asking operators to tick a box saying “cleaned” proves nothing. So Rajan stored a high-resolution image of each part when it is clean, such as the chain in a packing machine. After cleaning, the operator simply takes a photo. The system compares it with the clean image and judges how well the cleaning was done. For the operator, it is just one photo. For the business, it is quality that can be verified every time.
As Rajan puts it, you can now imagine a process and actually implement it, at a pace that was unthinkable before AI. Earlier, an idea like this would have needed a lot of time, effort and money. All of this has given Sirimiri’s employees a healthier work environment, and brought happier customers and faster business growth. Most importantly, Rajan and Vijaya have the right perspective on every aspect of their business, and with it, less anxiety and more peace of mind.
A team whose goal is to forget coding
I am a co-founder and investor in Delium.ai. Simon Roy is one of the most brilliant engineers I have worked with in my career.
When I met Simon in April 2026, he told me he had not written a line of code in six months. He was learning from the code Claude Code wrote. Then he said something that shocked me. The goal for his four-member engineering team in 2026 is to forget coding.
When I met him again in late September, I realised what that meant. Delium had built, in a little under 100 person-months, what took us more than 100 person-years of effort at Gofrugal. The Delium product is AI-native and more advanced, yet priced comparably to legacy solutions. One reason is that the team carefully picks the right model for each job: open source, small or large language models.
Delium also completed its cybersecurity audit, including the Vulnerability Assessment and Penetration Test (VAPT), in 2 to 3 days. The auditors did not find a single flaw. Just before the audit, the team’s own internal AI-assisted testing had already found four open issues. Simon shared with the auditors how the AI tools are able to spot these issues much faster. The auditors were taken by surprise. Simon and team fixed those issues after the audit.
How did they do it? They set up their development environment with complete context. And they kept refining the functional requirements as they built.
When I met Simon again in October, he walked me through his development lifecycle. Claude is now the only interface for his entire engineering and product team. Everything happens from there: managing the development pipeline, taking a requirement all the way to production, engineering management and audits. I was also impressed by the information security built into the development environment, including web application penetration testing (WAPT), the traceability of every issue and the clear accountability of each team member. What I found most interesting is how the system keeps helping team members learn and become more productive.
With Claude reimagined as the interface for all work, the pace of development is accelerating every month. In the last 10 days alone, the four-member team shipped 42 features and bug fixes per day. That is three times the pace of the same period last year, and about eight times the pace of 2024.
Happier customers, and a business you can talk to
Hari, co-founder of Pickyourtrail (PYT), pushes AI adoption across the company. PYT was founded in 2014 and has been building automation tools for a long time. Hari has always been obsessed with customers. He has tracked Net Promoter Score (NPS) for as long as I can remember, and for years it stayed around 30.
Hari used AI tools to bring customer conversations from every channel into their CRM. Now they track response time, quality of engagement and resolution time. Within six months, NPS jumped to over 50 and repeat business went up. In the same period, AI resolved more than half of all customer queries without any human help. That gave his team more time for the queries that needed a person.
What impressed me most is the conversational MIS Hari has built. He and his team simply ask questions about the business and get clear, easy-to-read answers. No more raising a request and waiting for a report. Often, by the time a report used to arrive, the need for it had passed. Information on demand has improved understanding, sped up decisions and made the team more agile. In an operations-heavy business, Hari says this has meant higher growth, happier customers and more repeat business.
Hari has also built AI agents to manage his own work. This has not only helped him multi-task more and run more projects in parallel, it has also helped him spend more time on his fitness, sleep and family. The result is faster, sustainable growth on stronger foundations, and a healthier, happier founder.
Turning employees into creators
I met Bert, the founder of California Burrito, in September 2026. He walked me through their technology roadmap. Using AI tools, he had designed the complete user interface for the entire operation, with screens for every role and every person who uses the system.
He shared these wireframes with ERP consultants to evaluate and choose an ERP. His plan is to configure and customise the ERP backend for their needs, while the interface stays their own, so that everyone finds the system easy to use. They are going live one function at a time.
California Burrito has also used AI to build an applicant tracking system, a learning management system for employees and many highly effective training videos. AI has turned many of their employees into creators, and the solutions are both effective and frugal.
In short, AI has transformed California Burrito into a more consistent, efficient, engaged, happy and innovative organisation.
A company that runs on AI
Alphageek Enterprises makes Zerowatt, a smart energy solution that helps industries track, analyse and optimise power consumption. Its core product, the Zerowatt Digital Brain, is itself an AI system. It learns how a large plant runs and cuts its energy consumption. For Alphageek, AI is not a side project. It is the core strength of the company, and the team now uses that strength to run the company itself.
The last time I met Ansha, Alphageek’s founder and CEO, he showed me the AI tools they built for hiring. They had more than 1,000 applicants for four field sales positions. Their AI agents:
- read every resume
- email candidates to schedule an AI screening interview, with full disclosure and the option to opt out
- talk to each candidate in the language they are comfortable with
- shortlist and rank candidates, so the hiring team meets the best fits first
- learn from each interview, improving the shortlisting and revising the shortlist already made
Alphageek now hires the right people faster and at lower cost.
Get Kumar Vembu’s stories in your inbox
Join Medium for free to get updates from this writer.
Hiring is only one example. Most of Alphageek’s internal systems are now AI-native:
- A company brain. Every team chat, email and document goes into one knowledge base. Anyone can ask about a customer, a deal or a past decision and get an answer with its source.
- Their own CRM, marketing automation and task system, all built in-house.
- AI agents that handle routine work from start to finish.
- Proposals, presentations and explainer videos drafted with AI and refined by the team.
- Finance and compliance work, like accounting reconciliation and security questionnaires, handled by AI using the company’s own records.
All these tools are connected to each other and to the company’s data through MCP (Model Context Protocol), an open standard that lets AI agents read from and act on every system a company uses.
AI has accelerated Alphageek’s journey to excellence. It has helped the team learn faster, practise better, work independently and improve the customer experience at every touchpoint.
Ansha’s goal is simple. Turn every repetitive task into an AI-native workflow. Raise the whole team’s productivity several times over. And keep people focused on work that truly needs human judgement.
All these initiatives have significantly improved every team member’s contribution. They have removed friction by giving each person a lane to run at their own pace, and they have made everyone feel a lot more valued.
Part 2: How to adopt AI and get the most from it
Set the right expectations
Here is what AI has done for me:
- It lets me attempt, and complete, things I never imagined I could do. It also gets me started on things I knew I could do but kept postponing because of the time and effort.
- In work I am already skilled at, it lets me work more independently, much faster and much better. The gain is not 25 to 50%. It is 5 to 10 times.
- This is why I think measuring AI in percentage gains undersells it. Don’t adopt AI to save a little time. Adopt it to change what you are able to do.
Treat it as a capital investment
Adopting AI takes time, effort and money upfront. You have to rethink how you learn, create, perform and improve. Treat that effort like a capital investment. Once it is done, the payback continues for years, at a much lower operating cost.
Learn from Ilaiyaraaja and Vairamuthu
The results you get from AI depend on how you work with it. Two legends of Tamil cinema taught me this.
Vairamuthu and “Chinna Chinna Aasai”
For Roja (1992), A.R. Rahman played the melody of “Chinna Chinna Aasai” and Vairamuthu wrote a draft fairly quickly. Mani Ratnam was not satisfied. He felt the lines did not capture the innocence and joy of a village girl who was about to be uprooted from her world.
Mani Ratnam kept rejecting conventional film lyrics. He pushed Vairamuthu to go deeper into a young girl’s smallest, purest wishes. As Vairamuthu has recalled, he spent nearly five days rewriting. Out of that came images like a girl who wants to catch a fish, only to let it go back into the river. Vairamuthu has credited Mani Ratnam for drawing that out of him. A lyricist gives his best when the director refuses to accept less.
Ilaiyaraaja and the directors who got his best
Isaignani (the musical genius) Ilaiyaraaja was often asked why directors like Bharathiraja, Mani Ratnam, Balu Mahendra and Mahendran got more hit songs from him than others did.
His answer has always been that he does not change his effort based on the budget, the producer or the star. When he composes, he sees only the situation and the emotion in front of him. Raaja, as Ilaiyaraaja is fondly called, has also said a director need not understand music. The director’s job is to explain the scene: the emotion, the character’s state of mind, and what is happening.
The directors who got his best work did exactly that. They did not ask for specific instruments or ragas. They explained the story, the characters and the mood in rich detail. With that context, Raaja often composed remarkably fast. Film critic Baradwaj Rangan has made the same point: the quality of creative output depends on how clear the director is about what he wants.
Iterate, and know when to start over
Both of them iterate. Both also know when to stop polishing and start over. When something isn’t working, they don’t keep patching it. They listen to the feedback and come back with a completely new idea. Knowing which one to do, and when, is a skill in itself.
My three guidelines for working with AI
These stories map directly to how I work with AI:
- Provide comprehensive context. Be like the directors who got Raaja’s best. Explain the situation, the goal, the constraints and what good looks like. AI gives its best when it understands the full picture. Delium did exactly this when it set up its development environment.
- Don’t settle for mediocrity. Be like Mani Ratnam. Push back until you get what you want.
- Know when to iterate and when to start over. Sometimes refining works. Sometimes the faster path is to scrap the work and begin again with a better brief. Don’t be afraid to throw work away. With AI, starting over costs very little.
In the end, it is in your hands
Ilaiyaraaja brought the same commitment to every film. What each director took away depended on how they engaged with him. AI is the same. It is available to every business, big or small. What you get from it is in your hands.
For a sole proprietor, a small business or a startup, AI lets you achieve far more with limited resources. It can also make you a better leader and a more effective manager, as Hari shows. The businesses that adopt AI will move ahead. Those that don’t will find it very hard to compete.
Thiruvalluvar gives a warning in his Kural 535.
முன்னுறக் காவாது இழுக்கியான் தன்பிழை
பின்னூறு இரங்கி விடும்
It means, one who fails to guard against danger beforehand will regret his mistake later.
Your business’s survival depends on how well it adopts AI. And that starts with you.
Start now. Give it context. Push back. And don’t be afraid to start over.
My sincere thanks to Rajan, Simon, Hari, Bert and Anshu for sharing their experiences, reviewing this draft and making it richer.
