This TechCrunch Disrupt 2025 panel discussion brought together entrepreneurs and investors to share deep insights on the journey to finding Product-Market Fit (PMF) and strategies for success. The discussion focused on lessons learned from failure, the importance of listening to customer voices, and the evolving nature of PMF in the AI era. Panelists emphasized how difficult it is to build products, set appropriate prices, and meet customer expectations, sharing real experiences on navigating the confusion and uncertainty inherent in this process. The session offered essential advice on what it takes to become a successful startup amid today's market dynamics.

1. Learning the Importance of PMF Through Failure

The panel opened by quoting the famous business strategist Mike Tyson:

"Everyone has a plan until they get punched in the mouth."

In the startup world, "getting punched" is the moment your product first meets customers. Getting back up after the punch and improving the product is the key to success. Ann Bordetsky, a partner at NEA, shared her most memorable PMF failure from her 12 years working at startups: a company called Wheels, launched in San Francisco in 2012.

Wheels pioneered a new category of P2P (peer-to-peer) car sharing around the same time as Uber and Lyft. It had the same backdrop as today's successful companies -- mobile payments, the commerce boom, and the emergence of the P2P sharing economy. But Wheels "narrowly" failed at PMF.

Ann explained three reasons:

1. Usage frequency: P2P car sharing was a weekly or monthly service, like weekend grocery runs, whereas ride-sharing (Uber, Lyft) was used daily, even multiple times a day. Ann emphasized that daily habitual usage is critical for PMF.
2. Asset-heavy model: Wheels required parking spaces, infrastructure, and in-vehicle hardware installation -- an asset-intensive model. Ride-sharing, by contrast, was an asset-light model where drivers operated cars via mobile apps.
3. Market size (TAM): The market for people who occasionally need a car was far smaller than the market for everyone who needs transportation for daily commutes, airport trips, and more.

Ultimately, Wheels attempted several pivots but failed, eventually merging with RelayRides (now Turo) in a "graceful exit." Ann advised that as the AI era brings many similar products to market, the key is finding "super engaging" daily usage behaviors and giving products "stickiness."

2. PMF in the AI Era: Focusing on the Voice of the Customer

Murali Joshi, a partner at ICONIQ, discussed PMF in the AI era, emphasizing that what matters is whether initial experimental interest in AI products translates into "durability of spend and usage." He said the voice of the customer is the core source of truth -- understanding how customers talk about the product, how they use it, and "how sticky is it for their workflows."

For enterprise AI products specifically, he advised asking decision-makers like CIOs, CISOs, and CXOs:

• "What role does this product play in the existing stack?"
• "How can we integrate this product more deeply into core workflows to make it more sticky?"

Murali explained that getting companies to move AI solutions from experimental AI budgets to core CXO budgets is crucial, ensuring AI becomes "a tool and a solution that's a platform that's here to stay" rather than a passing fad.

3. Chef Robotics' 'Sell Before You Build' Strategy

Rajat Bhageria, founder and CEO of Chef Robotics, went beyond listening to customers to introduce a unique PMF journey: "selling before building."

Rajat focused on robotics to solve the severe labor shortage in the food industry. Five to six years ago, the robotics industry was in a downturn, so he knew he had to do things differently to succeed. He knew customers wanted robots, but was uncertain about what form and features would be optimal.

So Chef Robotics devised the following strategy:

1. Early contracts: Before building the product, they signed contracts with at least three customers. This prevented over-customizing for any single client.
2. Performance-based payment: Customers paid a robotics-as-a-service subscription fee, but only when predefined success criteria were met -- for example, "must be food-grade, must be a certain size, and must perform these specific functions."
3. Refundable deposits: Customers paid deposits to show commitment, but could get refunds if the company failed to meet the success criteria.

This approach brought Chef Robotics several advantages:

• Validated customer demand: Since customers signed contracts and invested real money, the company could be confident in actual demand.
• Clear product requirements: By synthesizing requirements from three customers, they created a product requirements document (PRD) that wasn't biased toward any single client.
• Design partners: Early customers became early adopters who closely collaborated with the company, providing feedback and helping steer product improvements.

Rajat emphasized that the "sell before you build" approach is particularly useful for enterprise product development. After speaking with 60-70 robotics founders, he realized many followed Eric Ries' Lean Startup methodology, but hardware products have long prototype cycles (3-6 months), making rapid iteration difficult. Understanding customer requirements upfront and getting it right the first time was therefore essential.

He also explained why Chef Robotics uses AI as a core feature. Traditional industrial automation robots were specialized for performing identical tasks repeatedly, like those in Ford factories 40 years ago, but struggled with the inherent organic nature of food -- every piece of chicken looks different. AI, especially computer vision, is essential for building flexible robots that can perceive and adapt to organic variation. Most of Chef Robotics' team consists of AI and computer vision software engineers focused on flexibly leveraging commercial hardware.

4. How PMF Is Changing in the AI Era: An Investor's Perspective

Ann noted that finding PMF today looks completely different from the past, particularly for AI infrastructure and application companies. She said the traditional Lean MVP (Minimum Viable Product) approach is no longer fully valid.

Key changes in PMF for the AI era that Ann identified:

1. Technology dynamism: AI technology is constantly evolving, discovering emergent properties and capabilities. PMF depends on leveraging these new "Lego blocks" to build great products.
2. The momentum game: Every category is over-invested and extremely noisy. Ideas and products alone aren't enough -- you need to "tell the story, get people to pay attention" and stand out in the market.
3. PMF as a continuum: PMF isn't something you reach at a single point -- it's a continuum that must be strengthened over time. You need to move users from seeing the product as "interesting" to viewing it as "essential."
4. Speed and adaptability: PMF is about velocity, discovering the value AI can deliver to users, and storytelling. You must constantly adapt to market dynamics and capture value as the environment changes every few months.

Murali agreed, noting that as an investor, he looks for whether companies that have already found PMF have an "innate desire to keep searching and finding." He emphasized the importance of understanding whether AI is a tailwind or a headwind for every business, and strategically thinking about how to capture changing customer and market needs in the AI era.

5. Finding PMF Amid Uncertainty: Opportunity and Capability

How do you find PMF when market fluidity and uncertainty are increasing? Ann argued that this change represents a tremendous opportunity for entrepreneurs.

"The fact that the world is changing so much is one of the biggest opportunities founders will have over the next decade."

Periods where business habits, enterprise buyers, and consumer habits are all susceptible to change are rare. People are open to trying new things and experimenting with massive corporate budgets on various tools. This is a "time-limited incredible opportunity" -- a prime chance for startups to grab a piece of the AI future.

As an investor, Ann said she invests in "founders that can navigate a dynamic era" -- leaders who combine technical ability and business insight to continuously navigate all the dynamism of markets, technology, and the macro-political environment.

Rajat mentioned that the pandemic helped crystallize Chef Robotics' business plan. As labor avoidance intensified, the labor shortage worsened, and food manufacturers urgently felt the need for robotic automation. Initially focused on ghost kitchens and fast-casual restaurants, the company discovered new market opportunities when large food manufacturers like Amy's Kitchen proactively requested automation. The fact that customers "convinced us we had to do this" was itself a powerful signal.

6. PMF Metrics: Balancing Quantitative Data and Qualitative Feedback

How do you know if you've achieved PMF? Especially with AI products that have high flexibility and diverse use cases, what metrics should you use?

Murali noted that every business is different so there's no one-size-fits-all metric, but he looks at:

• Customer acquisition velocity: Whether friction in selling the product is decreasing and customer additions are accelerating. This indicates not just market "push" but also "pull."
• Large customer acquisition: The ability to land bigger, more sophisticated enterprise clients.
• Utilization and engagement: For AI products especially, measuring not just budget allocation but how frequently end customers within organizations use and engage with the product. Beyond consumer metrics like DAU-to-MAU ratio, checking end-user engagement frequency within enterprises.

Ann warned that at early stages (seed, Series A), data may be insufficient or misleading, and offered these metrics:

• Product velocity: How quickly the team ships new features and products (weekly, even multiple times per week).
• User growth, engagement, and retention: User growth and minimizing churn are important, but above all, the "fanaticism" of the early user base matters. Is there a fandom saying "I can't live without this product"?
• Must-have vs. nice-to-have: Whether the product is a "nice to have" or a "must-have" -- this applies even in entertainment categories. Qualitative interviews reveal how deeply integrated the product is into customers' lives.

On balancing qualitative and quantitative data for PMF, Ann said at the early stage you need to consider both. Since you must make decisions with incomplete information, she focuses on founder-idea fit or founder-market fit -- understanding "why can this team uniquely succeed in this business?" This includes understanding the team's technical capabilities and ability to realize their vision.

Murali also emphasized the importance of founder-market fit, explaining that deep insight and knowledge of a specific industry or market can lead to success. Founders like Procore's Tooey and ServiceTitan's Ara built successful software businesses by identifying unique market needs through years of domain experience. Ultimately, PMF must be assessed through a combination of qualitative understanding of customer voices and data metrics (ACV, growth rates, retention rates, etc.).

Rajat agreed that "utilization" is the key metric for enterprise products. He candidly admitted that early on, the product frequently broke down with many software bugs, resulting in low utilization and uptime. But customers showed tremendous patience, saying "No, come back tomorrow and fix it. We really need this product."

"Having customers tell us 'No, we want this, just figure it out' was incredibly motivating."

The customers' strong "desire" was the driving force that kept the team from giving up and overcoming technical challenges, Rajat recalled.

7. The Importance of Price Model Market Fit

Pricing expert Chris pointed out that many companies achieve product-market fit but fail at "price model market fit." The question arose whether companies consider appropriate pricing models too late.

Rajat explained from Chef Robotics' experience that PMF and pricing models are closely intertwined. Food industry customers were accustomed to equipment purchases and preferred a capex model, but Chef Robotics pursued a subscription-based "robotics as a service (RaaS)" model to account for software costs. Initial customer pushback was fierce, but the company held firm and proved the model's effectiveness through successful customer implementations and case studies.

"We got very good evidence from many customers about why the robotics-as-a-service model is useful."

Rajat said that once the RaaS pricing model was established and case studies were secured, "sales velocity just like accelerated so much." He added that showing clear ROI through value-based pricing -- setting prices slightly below labor costs -- is essential.

Murali emphasized that pricing is a highly iterative process. Even companies that have grown from millions to hundreds of millions in revenue have changed their pricing models dozens of times to adapt to market and customer changes. In the AI era, "usage-based" or "meter-based" billing models are becoming increasingly important, with various thresholds, credit limits, and tiers to drive upselling. Ultimately, you must constantly optimize pricing models by listening to customer voices and analyzing the competitive landscape.

8. The Evolution of Lean Startup in the AI Era

The audience asked how the traditional Lean Startup approach should change in the AI era and what new adaptations are needed for finding PMF.

Ann said the "timeless wisdom" of the Lean Startup framework remains valid. Building a minimum viable product (MVP), learning through user feedback, and improving the product are still fundamentally important. But in the AI era, there's a key change:

"Building a lean prototype is literally easier than ever before."

In the past, building lean prototypes required considerable effort, but today you can easily build apps, landing pages, first-version products, and agent workflows to test any idea. The emphasis should now shift from building prototypes to the "iterative aspect."

That means spending more time closely collaborating with early design customers, users, and potential enterprise clients to identify truly valuable and essential core business workflows or utilities. Then building "more durable" product versions based on those findings. Prototyping is easy, but "building deeper workflows and product durability" is today's core challenge.

9. VC Fundraising Strategy: When and How to Approach

Daniel, a new founder with 6-12 months of operating capital from angel investors, asked about receiving outreach from over 50 VCs -- when to start conversations and how to filter them.

Ann noted that much of VCs' outbound outreach exists, but warned that these contacts may not translate into real investment interest. Real interest can only be gauged by directly engaging with VCs. She advised not spending too much time on VC meetings until you're ready to fundraise. When you've decided to raise:

• VC selection: Research who the best investors in your space are and whether they've shown signals of interest in what you're building.
• Warm-up: Take 1-3 months before starting fundraising to build relationships.
• Investor-company fit: Conversations with VCs are most likely to lead to actual investment when there's "investor-company fit."

Rajat emphasized that "fundraising isn't the goal."

"Fundraising isn't the goal -- revenue is the goal."

He recommended not talking to investors until you have personal conviction in your product and real evidence from customers. He feels the pressure of having to return money if things fail after taking investor funds.

Rajat also noted that most VC outbound contacts are "noise", and getting a warm introduction to a "general partner" is far more effective than going through analysts or associates. Connecting through other founder communities is a good way to achieve this.

The panelists agreed that the most effective approach is to focus on PMF and approach investors with sufficient data as operating capital runs low. Deep-tech companies that need significant capital for prototype development may be exceptions, but even then, the founder's "conviction" and willingness to persevere through difficulties are paramount.

Ann closed with a fundraising maxim:

"Ask for money and you'll get advice. Ask for advice and you'll get money."

This saying perfectly captures the right posture for conversations with investors.

Conclusion

Today's panel discussion deeply explored how difficult and complex the process of finding PMF is, and how learning from failure is essential. The arrival of the AI era is significantly changing the nature and approach to PMF, and the core message was that listening to customer voices, exploring innovative business models, and constantly adapting to market dynamics are keys to success. Ultimately, we reaffirmed that outstanding founder capability, persistent execution, and customer-centric thinking are the most important elements for navigating an uncertain future and finding PMF.