Aniruddha Banerjee, co-founder and CEO of SwitchOn, sits down with Rohit Yadav to explain why factory floors got automated decades ago while quality inspection stayed manual, and what finally changed. He left Nvidia in San Francisco. His co-founder, who is also his first cousin, left a team building India’s moon rover. They started in a Bengaluru apartment, working on a problem their own family’s manufacturing business knew well.

SwitchOn builds AI for quality inspection, aiming at zero defect and zero customer complaint. The core product, Deep Inspect, combines cameras, lighting and compute at the line. Since commercialising in 2021 the company has grown roughly 3x year over year, and now works with 37 clients across about 80 sites and close to 300 deployments, with a team near 100 and around 40 percent of revenue from Europe, Southeast Asia, and North and South America. The system detects defects as small as 150 microns, runs at up to 1,200 parts per minute, holds false positives under 0.5 percent, and lets customers train and deploy their own models in 45 minutes.

Whether you build hardware, invest in physical AI, or run operations in a plant, this one is about the unglamorous layer where most of the value actually sits.Chapters:

00:00 Two cousins, a family manufacturing business, and the quality gap

02:05 Why computer vision and inspection took so long to meet

04:01 What SwitchOn builds, and what came after Deep Inspect

05:27 The numbers: 37 clients, 80 sites, 300 deployments

08:25 Where deployments actually run into trouble inside a plant

11:58 Training on good product instead of defect data

15:23 Why inference runs at the line and not in the cloud

18:29 What stops a pilot from scaling inside a large manufacturer

22:07 Advice for founders: stakeholders, and willingness to pay

24:19 Hardware versus software in physical AI

27:38 The mid-market question, and why they have not gone there yet

31:13 Going global: product fit, referrals, and earning trust

34:50 The moat: 200 million defect images and what compounds

40:54 Closing the loop, and what DeepFix does

43:58 Where the value sits in 2031🔑 Key insights you will walk away with:

➡️ Why rule-based systems needed tens of thousands of images and still only handled a narrow set of checks

➡️ How transformers cut training to under 200 images and 45 minutes, and why that unlocked scale

➡️ The decision to train on good product rather than defects, and what it saves a customer who cannot stop the line

➡️ Why plants in remote locations and confidential product lines made cloud inference a non starter

➡️ What actually goes wrong on deployment, starting with teams who disagree on what good looks like

➡️ Why a pilot dies if value does not land in weeks rather than months

➡️ The founder advice on assessing willingness to pay before building

➡️ Why the hardware versus software split matters less than revenue quality and pricing power

➡️ How 200 million real defect images compound into something synthetic data cannot replicate

➡️ The case for special purpose systems over general purpose humanoids on a factory lineLinks:

Aniruddha Banerjee: https://www.linkedin.com/in/anirudhabanerjee/

SwitchOn: https://switchon.io/

Rohit Yadav: https://www.linkedin.com/in/rohityadav23/

Newsletter: https://yadavrohit.substack.com/

Rukesh Reddy, founder of Deccan AI, and Purva Bhandari sit down with Rohit Yadav to explain where the data layer sits in the model building stack, and why it does not disappear as models get smarter. Every model has three ingredients: compute, the engineers writing the algorithms, and the data. Deccan builds the third one, supplying training and eval data to the handful of labs building frontier models, and now also helping enterprises deploy AI.

Rukesh came to this from finance, working for a bank in New York before what he calls a complete U-turn. The thesis was a shift in what annotation means. The industry scaled around 2014 with self driving, when the work was bounding boxes, accurate but not IQ heavy. Post ChatGPT it needed engineers, PhDs and domain experts who can reason with a model, which he saw as a right to win for a company delivering out of India. The name is deliberate too: after learning that Cisco is a homage to San Francisco, he asked why nobody was claiming the Deccan, India’s own corridor across Hyderabad, Bangalore and Pune. The company bootstrapped for 18 months, raised a small round, then a Series A in March.

Whether you build models, deploy them inside an enterprise, or invest in either, this is a clear look at the layer most people underestimate.

🔑 Key insights you will walk away with:

➡️ Why the data layer survives every change in algorithm, from RLHF through to RL

➡️ The difference between full stack and first party models, and why those lines keep blurring

➡️ Why in house data teams at labs are a shrinking share of the market

➡️ How synthetic and human data fused into a hybrid, and why one or two percent of human input is the kicker

➡️ Why knowing where models break is harder than generating data, and what that demands of a platform

➡️ The case for starting in one country when cycle times can be as short as a day

➡️ Why today’s models have high bookish knowledge and struggle with the exceptions outside codified process

➡️ Why the binding constraint on open source adoption is engineering capacity rather than licensing

➡️ Why ARR is a shaky measure when revenue is far more volatile than previous tech waves

➡️ Why language is a flattened out view of the world, and what world models are actually for

Links:

Rukesh Reddy: https://www.linkedin.com/in/rukeshreddy/

Purva Bhandari: https://www.linkedin.com/in/purvabhandari/

Deccan AI: https://www.deccan.ai/

Rohit Yadav: https://www.linkedin.com/in/rohityadav23/

Newsletter: https://yadavrohit.substack.com/

Anand Jain, founder and CEO of Aerem, sits down with Rohit Yadav to explain why rooftop solar took off almost everywhere except India, and what it took to fix the gap. The economics were never the problem. An MSME pays 8 to 12 rupees a unit and a commercial user 12 to 18, while rooftop solar costs around 3 to 3.5. Payback lands between six months and three years, followed by roughly 25 years of effectively free power. The question Anand kept asking was why a market that obvious was not moving.

Aerem is a full stack platform for distributed solar, combining financing through its own NBFC, B2B commerce, and solar tech. Around 600 crores disbursed, roughly 1,000 crores of annual commerce sales, close to 5,000 installers on the platform, and about 3,500 plants built totalling 1,500 megawatts. Anand brings 25 years to it: Delhi School of Economics, six years in Mumbai across asset management and bank treasury, an MBA at Yale, Wall Street through the financial crisis, then two years in energy at Barclays Capital. He returned to India in 2009 to build a 10 megawatt project under Gujarat's first solar policy, set up KPMG's solar practice, and deployed about a billion dollars in utility scale before starting Aerem five years ago.

Whether you invest in energy, lend against assets, or build in hard physical markets, this one is about why a slam dunk sat untouched for a decade.

Chapters:

00:00 What Aerem is, and the three businesses inside it

05:35 From Delhi School of Economics to Wall Street to Indian solar

14:08 Utility scale versus rooftop, and why both will coexist

19:56 Rooftop was 15 percent of installed capacity. Globally it is 60 to 70.

21:49 Net metering, DISCOMs, and what actually slows adoption down

30:31 The three business lines and how they feed each other

36:09 Why he deliberately avoided the subsidised segment

42:24 Inside the lending book: collateral free, asset financed

45:24 The EPC channel as origination engine, and 100 plus quality checkpoints

49:19 Monitoring data, underwriting, and what it unlocks next

54:14 The competitive map: banks, subsidised lenders, private NBFCs

58:53 Fundraising, and why capital follows a scalable business

1:01:37 What went wrong in US residential solar finance

1:05:54 The next 20 years, and 96 percent of the market still untouched🔑 Key insights you will walk away with:

➡️ Why the savings case for Indian MSMEs is 30 to 80 percent off the power bill, and why that still was not enough on its own

➡️ How rooftop went from 15 percent of installed capacity to around 25, and where he thinks it lands

➡️ Why rooftop obviates transmission and distribution infrastructure entirely, even at a 15 to 20 percent cost premium

➡️ The deliberate decision to build where there is no subsidy, and what subsidies do to market quality

➡️ Why installing a plant costs three to four times more elsewhere, and what higher radiation does to the comparison

➡️ Why lending alone is shooting in the dark when execution takes two to three months

➡️ How 100 plus quality checkpoints during execution feed directly back into underwriting

➡️ Why term loans are collateral free with the financed asset as the only security

➡️ What went wrong in US residential solar: savings that were missold, and tenor mismatches

➡️ Why only 3 to 4 percent of the MSME opportunity has been solarised, in a segment he sizes at 500 billion dollarsLinks:

Anand Jain: https://www.linkedin.com/in/jainanand/

Aerem: https://www.aerem.co/

Rohit Yadav: https://www.linkedin.com/in/rohityadav23/

Newsletter: https://yadavrohit.substack.com/

Garima Bharadwaj, co-founder and CTO of Enlite Research, sits down with Rohit Yadav to explain why the buildings going up across India are beautiful on the outside and running on 30 year old control logic on the inside. Cooling that holds 24 degrees regardless of weather, occupancy or machine age. Systems that cannot talk to each other. And an operating layer that starts degrading from year two of a building designed to last twenty five.

Enlite builds an alternative stack for controlling machines, not buildings. Garima and her co-founders spent the first three and a half years building it: a single universal wireless controller that reads every signal type from every make of machine, paired with a cloud platform that learns from each new asset it onboards. The traditional approach meant 40 to 50 kilometres of cabling even for half a million square feet, with a desktop app at the end telling you what already happened. The first customer took 27 meetings and seven months to convince. By year two they had 5 million square feet. The year after that, 50 million, 70 customers, 12,000 controllers and roughly 3 billion parameters handled daily. They expect to pass 100 million square feet by March 2027, and they bootstrapped for six years before raising.

Whether you build, operate, finance or invest in the physical world, this one will change how you think about the gap between a well specified building and a well run one.

Chapters:

00:00 What Enlite is, and the gap that started it

05:13 Building the stack: cloud native and wireless, against the grain

09:00 Two customers, 27 meetings, and the first lighthouse projects

11:05 India's building boom, from tech parks to tier two towers

17:55 Why deterministic controls break by year two

24:02 Why the hardware had to be built, not bought

28:44 The moat: full stack, and what the patents cover

31:46 Who pays, and why they refuse to lead with savings

36:00 What retrofit buildings actually reveal when you audit them

39:00 The Blackstone building: complaints down, headcount down, energy down

41:53 From passive buildings to responsive grid assets

47:06 Six years of bootstrapping, and why PMF came before capital

53:28 Expanding to the Middle East, and into data centres

58:01 What autonomy looks like five years out

🔑 Key insights you will walk away with:

➡️ Why a control logic set on day one is already wrong by year two, once load, seasonality and machine ageing drift

➡️ What it means to control individual machines rather than a building envelope

➡️ Why cooling plants, occupancy systems and solar generation usually never speak to each other

➡️ How a company sold automation and uptime rather than guaranteed savings, and why that avoids the usual performance contracting fight

➡️ What incumbent automation coverage actually looks like in large premises, and why it cannot be fixed after day one

➡️ The case for owning hardware when you want to replace an incumbent rather than sit on top of one

➡️ Why India's 680 million square feet of retrofit stock is the harder and more interesting opportunity

➡️ How a building becomes a responsive asset rather than a passive load on the grid

➡️ Why building codes alone cannot deliver efficiency if the machines behave inefficiently every day

➡️ The argument for bootstrapping six years rather than finding product market fit on someone else's capital

Links:

Garima Bharadwaj: https://www.linkedin.com/in/garimabharadwaj/

Enlite Research: https://enliteresearch.com/

Rohit Yadav: https://www.linkedin.com/in/rohityadav23/

Newsletter: https://yadavrohit.substack.com/

Ashish Goel, co-founder of Optimist, sits down with Rohit Yadav to explain why the air conditioner in your room is solving the wrong problem. It was invented as a dehumidification device and refined for temperate climates, and on all three variables of thermal comfort it gives an answer built for someone else: air colder than you need, drier than you like, and slower than you enjoy.

Ashish has been building for 18 to 19 years. His first venture was in books, publishing and kids media, sold in 2011. He then co-founded Urban Ladder, which became India's largest furniture brand and was sold to the Reliance Group in 2020. The thread back to energy runs further: a gas engine and gas turbine system he helped design as a consultant in 2006 still sits under Cyber Hub in Gurugram, and he ran a clean energy blog from 2009 to 2011 before years of angel investing in the sector. Optimist builds next generation cooling systems designed in India, with co-founder and CTO Pranav, a mechanical engineer from IIT Madras with thermal sciences and an industrial masters from Cambridge. The company has raised about $12 million, with Accel as largest institutional shareholder.

Whether you build hardware, invest in the physical world, or just want to understand where the grid pressure is actually coming from, this one reframes a device you have stopped noticing.

🔑 Key insights you will walk away with:

➡️ Why the air conditioner was invented as a dehumidification device, and what that still costs you

➡️ How adaptive thermal comfort means two people in the same room genuinely need different settings

➡️ Why fixing the output temperature means redesigning the evaporator, the components, the firmware and the algorithm

➡️ The real share of India's peak load coming from one single device, and where that number is heading

➡️ Why the peak load problem is really millions of local transformers that cannot be upgraded

➡️ Why renewables remain a small share of total energy once you count energy rather than electricity

➡️ The case for an online only launch when you are a new brand in a high trust, high ticket category

➡️ Why service, not product, dominates customer reviews across this entire category

➡️ The gap between manufacturing locally and holding the design IP, and what the auto components industry took to close it

➡️ Why Series B and C capital for hard tech follows breakout companies rather than leading themLinks:

Ashish Goel: https://www.linkedin.com/in/ashish-goel-591572/

Optimist: https://optimist.in

Rohit Yadav: https://www.linkedin.com/in/rohityadav23/

Newsletter: https://yadavrohit.substack.com/

Pranav Pai, Founding Partner of 3one4, sits down with Rohit Yadav to unpack what has actually changed in Indian startup ecosystem over the last decade, and what has not. Ten years of Seed, roughly 10,000 companies pitching in a single year, and a view of deep tech, AI and energy grounded in what actually gets built.

3one4 is an early stage firm based in Bangalore, just past its tenth year. The team backs 30 to 35 companies per fund, and about 70 percent of those are first cheques into businesses that are still two or three people with a laptop. Of a 36 person team, only nine sit on the investment side. The rest work with companies after the money goes in, on finance, governance, revenue and capital markets access, which Pranav argues has quietly become the differentiator in a market where most global VC models stay hands off after the wire.

Pranav is an engineer by training, did his masters at Stanford, was employee number three at a SaaS company that had a great exit, and came back to India to build the firm he wished existed. Whether you are a founder raising a seed round, an LP sizing up India, or an operator trying to read where deep tech and energy go next, this one is worth the hour.

🔑 Key insights you will walk away with:

➡️ Why deal flow growing from 1,000 to 10,000 pitches a year matters more than any single fund's returns

➡️ How the Indian exit picture actually looks

➡️ Why a compounding bank stock is the real competition for every Indian VC raising domestic capital

➡️ The founder side and the investor side of the risk off argument, argued honestly from both

➡️ Why deep tech in India runs on talent first, then patent and grant reform, then a real first customer

➡️ How three decades of global R and D centres built an engineering base that almost nobody counted

➡️ Why fine tuning open models is the practical AI path here, and what that does to cost and time

➡️ Why fixing domestic supply gaps may matter more than chasing the frontier

➡️ Which parts of the energy problem are venture scale and which belong on a corporate balance sheet

➡️ The contrarian view hiding in plain sight: 10,000 chemical factories fighting over local demand instead of selling globally

Links:

Pranav Pai: https://www.linkedin.com/in/pranavpai/

3one4 Capital: https://www.3one4capital.com/

Rohit Yadav: https://www.linkedin.com/in/rohityadav23/

Newsletter: https://yadavrohit.substack.com/

Arjun Gupta, founder and CEO of Smart Joules, sits down with Rohit Yadav on TheOnePoint podcast to unpack why buildings, factories and homes waste roughly 20 to 30 percent of the energy they consume, and what it actually takes to fix that in India: performance contracts, cooling as a service, ten years of real chiller data, AI-run building management, and the unglamorous work of getting a large organisation to say yes.Smart Joules exists to put efficiency first in the built environment. It took a year and a half to sign the first customer, a hospital within walking distance of Arjun's home, and the first cheque came in at three to four times what the team had modelled. The company stayed bootstrapped and profitable for roughly six years, funding its first projects with debt before it ever raised equity, then tripled in size in about a year after signing a single ten-year contract covering 20 Apollo hospitals. Today it runs three businesses: retrofits, cooling as a service, and a building management system built on top of the closed-loop control it had already solved for its own sites. More than 50 hospitals are customers, greenfield is now over half the business, and the company is expanding into hotels, food and beverage, pharma, manufacturing and commercial buildings.Whether you are a founder, an operator, an investor, or simply curious about how AI lands on real physical infrastructure, this episode will change how you think about the demand side of the energy transition.🔑 Key insights you will walk away with:➡️ Why you cannot sell energy efficiency directly in a growth market, and why selling cooling works instead➡️ How a business stayed profitable and bootstrapped for six years by funding projects with debt before equity➡️ What actually changes when you move from one owner-operator building to a large corporate with its own engineering team➡️ Why a performance contract is easy to dream up and very hard to run for a decade➡️ Why over delivering on savings can make a customer unhappy rather than delighted➡️ Why hospitals are better customers than hotels: 24/7 load, commercial tariffs, low humidity theatres, and one decision maker➡️ Why the largest square footage of green rated buildings in the world still leaves 30 percent on the table➡️ How ten years of minute-level chiller data across 28 cities beats a manufacturer's ten-point design curve➡️ How AI turns a complaint in an operating theatre into an already triaged job on a technician's phone➡️ Why 60 chiller plants are not just a consumption story but a future grid assetLinks:Arjun Gupta: https://www.linkedin.com/in/arjun-p-gupta-5746ab10/Smart Joules: https://smartjoules.co.in/Rohit Yadav: https://www.linkedin.com/in/rohityadav23/Newsletter: https://yadavrohit.substack.com/

Jan Lozek, co-founder and Managing Director of Future Energy Ventures (FEV) and a former energy-infrastructure leader at RWE and E.ON, sits down with Rohit Yadav in this "AI Meets the Physical World" series episode to walk through where the energy transition actually stands today as an investment thesis. The conversation spans the shift from climate to sovereignty, the divergence between US and European energy markets, the centralized versus distributed debate, the rise of the intelligence layer, and why the durable venture returns are moving from owning assets to running the software on top of them.

FEV is one of Europe's leading energy-tech venture platforms. Founded in 2016 with a from RWE and E.ON, the firm has backed roughly 50 companies across the intelligence layer of the energy transition, including enspired, GridCare, ev.energy, and Thermondo. FEV has been compounding the same intelligence-layer thesis through an entire cycle, well before AI made grid reliability a mainstream concern.

Whether you are a founder, operator, investor, or simply curious about where physical AI and the energy transition intersect, this episode will sharpen how you think about energy as a sovereignty trade, and about where value accrues in a system where the hardware is increasingly commoditized and the intelligence layer is where the returns live.

Chapters:

00:00 Introduction: The "AI Meets the Physical World" series

00:47 What Future Energy Ventures is, and the intelligence-layer thesis

03:05 From a climate story to geopolitics to AI

06:15 Energy security and Europe's fossil-fuel dependency

07:57 Europe's self-sufficiency versus US AI-driven demand

09:28 The real difference between US and European energy markets

12:01 Centralized versus distributed, and why you need both

16:03 Where the next venture bets come from: data centers and robotics

17:19 Robotics for infrastructure and industrial decarbonization

19:07 Asset-light versus asset-heavy, and bankable assets

22:34 IPOs, exits, and the first ten-billion-euro grid company

23:30 The efficiency hiding inside the grid

25:24 How underwriting changed: team first, then market

27:53 Climate measurement versus energy sovereignty

30:34 China, manufacturing, and Europe's hardware reality

34:28 Owning assets versus privileged access to them

36:02 The future of utilities in ten years

40:01 Closing: an open invitation to founders

🔑 Key Insights You'll Walk Away With:

➡️ Why the energy transition has become a defensive sovereignty strategy rather than a climate story

➡️ Why the US and Europe face the same grid problem wrapped in two opposite narratives

➡️ Why centralized versus distributed is the wrong debate, and why the value sits in the middle

➡️ Why owning infrastructure is not a venture case, but privileged access to it is

➡️ How the same AI driving energy demand is also the tool that has to organize it

➡️ The two overlooked bets: robotics for energy infrastructure and industrial decarbonization

➡️ What China actually taught Europe about hardware, and how to respond

➡️ Why team composition is still the single biggest reason companies fail

➡️ What utilities look like in ten years, and the one thing they will never be good at

Links:

Jan Lozek: https://www.linkedin.com/in/jan-lozek/

Future Energy Ventures: https://www.fev.vc/

Rohit Yadav: https://www.linkedin.com/in/rohityadav23/

Newsletter: https://yadavrohit.substack.com/