Running Two Experiments with AI and Early‑Stage Capital

At Pentathlon Day 2025 in Pune, we put a founder and a fund on stage together and turned the next hour into a live debrief on how both sides are responding to AI, regulation, and a very different funding market.

On one side sat Monish Darda, Co‑founder and CTO of Icertis, a global CLM SaaS leader. On the other side sat the managing partners of Pentathlon Ventures – Madhukar Bhatia, Sandeep Chawda, Girendra Kasmalkar, and Navjeev Kumar representing Pentathlon Ventures.

Monish Darda opened with a story from inside Icertis. 15 leaders in a room, a whiteboard, and one question written in thick marker “What will AI do to contract lifecycle management in five years?” The company was already at serious scale – roughly 350 million ARR, 2,400 people, and about 500 enterprise customers.

Then the room split cleanly into two camps.

Running two experiments instead of one big bet

As Monish Darda tells it, half the Icertis leadership team became “the mavericks.”

They argued that everything would change. CLM, as they knew it, would be rebuilt from the ground up by AI and customers would buy something very different from what the company sold at that moment. The Kodak story came up, film as “art” and digital as a toy, until digital swallowed the category. The disbelief that a phone without a keyboard could work, until nobody wanted physical keyboards any more.

The other half were “the conservationists.”

They pointed to blockchain and how slowly it had reshaped real enterprise workflows. They talked about how long it took the internet to change procurement, legal and finance in large organisations. AI was powerful, they agreed, but enterprise adoption had always lagged the hype. Tools were slower than promised and risk and change management were real. Their reference points were the early struggles of copilots bolted onto big legacy productivity suites.

If you are the founder looking around that room, this is your leadership team, responsible for 1000s of people and a large revenue base, holding incompatible views of the same future. That discomfort is what pushed Icertis into its next move.

Monish Darda explains that there was no grand speech declaring loyalty to one camp. There was a capital allocation decision.

• The mavericks got a defined innovation budget – in the next twelve months, deliver 5 concrete AI‑driven use cases. If even one held up in the real world, it would be pulled into the mainstream product.
  
• The conservationists had their budgets cut by about 10 %. Revenue targets stayed where they were and the company still aimed to grow in the mid‑teens to mid‑twenties. Services, support, customer success and sales costs did not disappear. Their job was to use AI to win back that 10% through efficiency so the cut would not show up in performance.

In practice, Icertis started running two experiments in parallel:

• At the edge: can we build the future product we are both scared of and excited about?
• At the core: can we use AI to make the current business meaningfully better on real metrics, with less money?

Both groups misjudged one thing: the speed of change.

Nobody in that room predicted how fast models and infrastructure would improve, how quickly capital would flood into AI, or how sharply the value of players like OpenAI and NVIDIA would spike. A model company going from relative obscurity to one of the most valuable startups in history. A chip company moving past older incumbents and flirting with a three‑trillion‑dollar valuation in roughly two years.

For founders and investors listening to Pentathlon Day, the pattern was clear. You do not need to pick one camp and stay there forever. You can design your own version of these twin experiments.

The hard questions become:

• How much budget do you give to maverick bets?
• What efficiency target do you expect from the core?
• How quickly will you kill or scale experiments when data comes

A young fund learning its own lessons

Once Monish Darda had set that frame, the conversation shifted to the other side of the stage.

Moderator Monish Darda turned to Madhukar Bhatia, Managing Partner at Pentathlon, and asked a simple question: in a world changing this fast, have founders changed too ?

Madhukar Bhatia describes himself as a relative newcomer to venture. Pentathlon started in 2020, at the tail‑end of a frothy cycle where valuations floated above fundamentals. Back then, in India and globally, founders could often assume a new round every eighteen months. Some could afford to respond to diligence questions with “we have two other term sheets, decide fast”.

5 years and one funding reset later, he says the conversations in their office sound different.

Founders are “more ambitious and more grounded” at the same time.

• Ambitious because AI makes it feel possible to build larger companies with leaner teams
• Grounded because the era of “growth at all costs” is clearly over

Rounds are slower, valuations have come back to earth and founders talk more about gross margins, payback periods, and paths to profitability than they did in 2021.

Madhukar Bhatia also sees a shift in who is starting up:

• More repeat founders, including some with a “failed” company behind them, bringing scar tissue and deeper domain knowledge
• A wave of young teams straight out of college, who have grown up with large language models and vector databases and see no reason they cannot go straight at complex industries

If you are pitching into a B2B‑focused fund like Pentathlon today, you sit somewhere on that spectrum.

If this is your first company, you are competing with founders who have already lived through an entire cycle.

If you have done this before, this is a moment when experience and discipline are valued in a way they simply were not when capital was almost free.

Why Pentathlon keeps saying “use case first”

Underneath those founder stories, the Pentathlon team laid out a clear strategic choice. They have decided not to play the foundational AI game.

That means:

• No bets on training giant models
• No pushing limited seed‑stage capital into core infrastructure layers where cheque sizes are huge and competition is global

Instead, as Gireendra Kasmalkar and Sandeep Chawda explained, Pentathlon lives at the application layer. They look for software that attacks very specific B2B problems, often in:

• Logistics and transportation, where physical infrastructure is improving and digital infrastructure is racing to catch up
• Trade, finance, and retail processes that still run on email chains and spreadsheets
• Enterprise functions where data exists, but workflows are fragile and slow

In most of these spaces, AI is a powerful lever, but it is only one part of the stack. Integrations, rules, reporting, onboarding, and change management matter as much. That is why the partners are wary when a founder leads only with “AI‑first” as a slogan. A few years ago every pitch said “cloud‑based.” Today almost every one says “AI‑powered”.

As Gireendra Kasmalkar put it on stage, if everyone is AI‑powered, you have to stand out somewhere else.

The questions they keep coming back to sound like:

• Who exactly is your buyer and what is their daily pain?
• What process are you changing and how is it done today?
• What data and context do you really have access to?
• Where does AI move a number that matters, like time‑to‑approve or cost‑to‑serve?

For the founders in the room, the message was instead of saying “we are the next big AI company,” frame yourselves as the team that knows a painful process inside‑out and is now using AI, and other tools, to solve it better than anyone else.

The quiet ways the ecosystem is maturing

While the room was animated about AI, Navjeev Kumar, brought up something quieter: regulation.

Over the last couple of years, India’s securities regulator has rolled out a stream of new requirements for venture funds. Navjeev walked the room through a few that hit Pentathlon directly:

• At least one key person in every AIF now has to clear a certification exam. That pushed senior partners, who had not sat an exam in more than twenty years, back into study mode
• Funds must undergo annual cybersecurity audits. Devices and systems need vulnerability assessments, patching, and formal reporting
• LP units and portfolio holdings need to move into demat form, which forces coordination between founders, finance teams, and investors
• LP portals and even public websites must meet accessibility standards so people with disabilities can use them effectively

From a founder’s perspective, none of this appears on the product roadmap. But it shapes the capital pool around them.

More regulation means:

• Investors run deeper diligence, especially on security and governance
• Funds spend more time on compliance and reporting
• LPs gain confidence that the ecosystem is not running only on handshakes and spreadsheets

Monish Darda connected this to what he had seen in Europe when AI and data rules tightened. Early on, it felt like friction. Over time, companies that treated those constraints seriously often ended up better positioned when the rest of the world moved in the same direction.

India’s venture market, he argued, is walking a similar path.

Hard questions about where to sell

One of the most practical parts of the panel had nothing to do with model architectures. It came from a founder in the audience who asked about geography.

Do we:

• Build in India and sell in India first?
• Go straight to the US?
• Treat the Middle East as a bridge between the two?

Underneath that question sit common anxieties:

• Will an India‑first GTM make US investors hesitate later?
• How much capital do we need to fund a serious push in a high‑cost market?
• How do we avoid spreading ourselves too thin?

The Pentathlon team did not offer a single playbook. Instead, they suggested a set of filters:

• Bandwidth: Do you truly have the leadership time to commit to a new geography?
• Focus: Have you defined a narrow ICP, or just said “US market”?
• Budget: Have you priced in flights, hotels, local talent, and the opportunity cost of being away from your core team?
• Sequence: Is now the right moment, or should you deepen your India base first and revisit later?

On exits, Gireendra Kasmalkar and Madhukar Bhatia shared how the fund thinks about geography differently. A strong Indian base with solid domestic logos is attractive to certain strategic buyers and later‑stage funds. Early traction in the US or the Gulf opens other doors. 

Outcomes they work toward are not limited to IPOs:

• M&A, ideally with clean liquidity for founders and funds
• Secondaries in later rounds, where new investors buy out earlier ones through platforms or bankers

For founders planning their next market, the advice was to hold all of this together:

• Where can you prove value fastest?
• Which customer logos will matter to your eventual acquirer?
• What story will make sense when an investor five years from now looks back at your journey?

Matching investors and thesis as both sides scale

Later in the discussion, Monish Darda asked how Pentathlon itself is changing as an organisation.

They explained that funds evolve just like companies.

• Fund I drew heavily from individual entrepreneurs and angels, many in and around one city. Cheques were smaller, relationships were very personal
• Fund II widened the circle. More business families from other regions joined. A Gulf‑based family office came in, creating a bridge for portfolio companies looking at that region

Fund III, which they are already preparing for, will look different again. There is a pipeline of institutional LPs who like the thesis but could not fit into Fund II because their minimum cheque sizes were too large relative to the fund. Many institutions have rules about what percentage of a fund they can own. That made participation impossible earlier. As funds get larger, that constraint eases.

So this is not just a story about a fund getting bigger. It is a story about fit. Different pools of capital play different roles.

• Large global funds backing AI‑native law or insurance plays are taking very long‑dated, high‑risk bets
• A seed‑stage B2B SaaS fund like Pentathlon backs teams that are closer to revenue and closer to P&L metrics business families understand

The fit has to work both ways:

• Funds need LPs who understand and believe in their strategy
• Founders need funds for whom their business sits near the centre of the map, not at the edge

When that happens, as Monish pointed out, you get more than money. You get time, pattern recognition, and help on the problems that matter most.

Making the E bigger, together

Toward the end of the panel, Monish Darda returned to something very simple from inside Icertis.

In one all‑hands, a young engineer pointed out that the last of those often got less airtime than the others. His comment was blunt: without execution, the rest are just nice words.

In response, the company made a small but memorable change. On its values slide, it increased the size of the letter E. A visual reminder of what actually moves numbers.

On stage at Pentathlon Day, the fund beside him described its own version of “making the E bigger”:

• Two partners dig into each opportunity
• Customers are called directly
• Internal and external investment committees, including Monish Darda, debate each case
• Navjeev’s team periodically reviews companies they passed on that later raised elsewhere, to learn what they might have mis‑read

If you are building or investing in 2026, it is easy to get lost in the big narratives: AI everywhere, new regulations every quarter, commentators arguing about de‑globalisation, bubbles, and the next crisis.

Monish Darda and the Pentathlon partners did not claim to have a perfect answer to any of that. What they modelled on stage was something more grounded:

• Hold a clear thesis, yet be willing to update it
• Make room for mavericks and conservationists in the same organisation
• Focus on real use cases and value creation
• Take regulation seriously instead of treating it as noise
• Choose your capital as carefully as you choose your market
• Execute relentlessly

That brings the story back to where it started.

The first room was that offsite inside Icertis, debating “What will AI do to CLM in five years?” The second was Pentathlon Day 2025, where the same founder sat with the fund’s managing partners and replayed the story in front of B2B founders and investors.

The question on the whiteboard had expanded from “What will AI do to CLM?” to “What will AI, regulation, and a changing funding market do to all of us?”

If you walked out of that hall with one thought, it was probably this: you do not get to choose whether the world changes. You only get to choose how deliberately you prepare for it.