Two partners from YC and Lightspeed: For consumer-grade AI, the real entry points lie in these three types of products.

In 2025, consumer-grade AI tools flooded the screens: photo editing, translation, writing, PPT creation... However, the app that can make it to the home screen of ordinary people's phones and be opened daily has yet to emerge.

So, where exactly lies the next wave of opportunities?

On November 28, 2025, Y Combinator's podcast program, Startup Podcast, released an episode of dialogue.

The host was Garry Tan, the president of YC, and the guest was Michael Mignano, a partner at Lightspeed. He co-founded Anchor, transforming podcasts from a niche hobby into a mainstream tool for expression. In 2019, Anchor was acquired by Spotify.

Within less than five minutes of the opening, Mignano clearly pointed out the most challenging aspect of consumer-grade products:

The truly difficult part is not spotting the trend but seizing the right moment:

When will users truly fall in love with this product? No one can know in advance.

However, they provided an answer: The greatest entry points in the AI era are hidden in places where you think there are no opportunities.

This dialogue is not a grand narrative but a practical methodology for founders: Stop asking whether consumer-grade AI is still viable.

The real question to ask is: Have you spotted an opportunity that others haven't?

Section 1 | Is there still an opportunity for consumer-grade products after large models?

Sora can generate videos, Grok can have conversations and write code, and Claude can summarize long texts...

In 2025, as AI models become more powerful, it's getting increasingly difficult to develop consumer-grade products.

Many entrepreneurs are trapped in a common belief: Since models have become the infrastructure, how can applications compete?

However, Michael Mignano, a partner at Lightspeed, sees it differently: The more powerful the models are, the more new products can be developed, especially those that were previously impossible to create.

Take music as an example.

Over the past decade, there has been a constant stream of social platforms, short - video apps, and photo - editing tools, but no one has truly lowered the threshold for music creation.

What Suno achieved was not just generating melodies. It allowed ordinary people to write songs for themselves, listen to their own creations, and put them on repeat for the first time.

Mignano said, "I can't think of any other content form where I've seen such behavior."

What AI brings is not just a new function but the invention of a new behavioral scenario.

This is also the reason why they are optimistic about consumer - grade AI: The opportunities don't lie in the models themselves but in the new windows opened by the models.

Section 2 | Which three types of products are on the verge of explosion?

So, specifically, which products are they?

The opportunities seen by the two partners share three common characteristics:

The user scenarios are real, but no one has specifically designed tools for them;

AI has lowered the threshold, motivating people who have never tried before to get started;

Once they explode, it's not just a tool iteration but a change in behavioral habits.

① Niche but high - frequency small tools

The most typical examples are the system - level entrances that have been forgotten by the times: email, calendar, task manager, and notes.

However, everything changed with the emergence of AI.

Granola is a case in point: It's not a traditional note - taking tool but an intelligent assistant that can remember your work habits. It doesn't require you to manually organize things but automatically understands your meeting content, to - do lists, and work rhythm.

The value of these tools doesn't lie in how beautiful the interface is but in the fact that AI can understand what you're doing and then automatically help you with the next step.

② Light - entertainment apps that seem like toys

Many people look down on things like music generation, photo filters, video synchronization, and virtual avatars, thinking that they have low user retention rates, users are reluctant to pay, and they can't grow into large - scale businesses.

However, these products target not the need for tools but the desire for self - expression.

Character.ai is a typical example. Users aren't just "using a chatbot"; they're creating their own virtual characters and having repeated conversations with them. This kind of behavior doesn't exist in traditional products.

The essence of these products is not content tools but turning creation into entertainment.

Unlike Notion and Figma, which require certain skills, they are more like Instagram and Snapchat in the past:

They turn creation into a fun thing that makes people unable to stop using them.

AI makes this possible for the first time.

③ Memory - based AI products based on private data

The third type of product represents the long - term potential that YC values the most: your own AI memory.

These products don't focus on instant conversations. Instead, they internalize all your photos, chats, browsing history, and health data into a personal knowledge base.

Nory: It integrates with Apple Health data and becomes your health advice assistant;

Rewind: It records your screen, voice, and documents comprehensively and allows you to replay your entire workday;

Dtronic: Based on medical literature and patient records, it predicts your treatment path.

In these scenarios, the model is not just a question - answering machine but a part of your life.

These three types of products neither seem grand nor are they the kind of platforms that large companies are scrambling to develop.

However, they all share a common feature: They can be used by people, people are willing to use them repeatedly, and they'll feel uncomfortable without them.

This is what the two partners emphasized repeatedly: Products that aren't supposed to succeed are often the most worthy of investment.

Section 3 | How can small teams win? The 15% weekly growth rule

Now that we know what products to develop, how can small teams verify that they've made the right choice?

Mignano's answer is: growth. It's not about developing a perfect product first and then seeking growth, but finding the right product through growth.

When Anchor was on the verge of bankruptcy, they set a growth bottom line for themselves: 15% per week.

To achieve this goal, they made three unconventional decisions:

1. Prioritize user growth over a perfect product

The team was caught in a common trap at that time: There were too many features, but the number of users wasn't growing.

So, they set a strict rule: If we don't grow next week, we won't survive.

This life - and - death pressure forced the team to abandon the self - satisfying product development path and focus on the actual strong needs of users: one - click distribution to Spotify and Apple Podcasts.

"Since these platforms didn't open their APIs, we manually helped users upload their content. When users clicked a button, we uploaded it for them in the background."

Yes, they started doing manual and inefficient work. It wasn't because they didn't know how to automate it but because it was the only way to immediately meet the real needs of users.

This is especially important in AI entrepreneurship.

Many teams start with the model and spend 90% of their energy on prompts, interfaces, and UIs, ignoring the real motives of users at a certain moment.

YC's methodology is the opposite: First, meet the seemingly inefficient needs, and then scale up after verification.

2. Distribution is about getting users to spread the word, not just buying traffic

Michael Mignano, a partner at Lightspeed, clearly pointed out that the biggest challenge for today's consumer - grade products is not functionality but distribution.

The traffic - distribution mechanism of platforms has changed, and the traditional model of "launching a product and waiting for users to find it" is no longer effective.

Instead, it's:

Short - video + influencers (Suno, Character.ai)

Content platforms + template sharing (Obo, Midjourney)

Anonymous communities + user word - of - mouth (Granola, Rewind)

"Distribution is an art, but it can't rely on burning money. You either have to be a creator yourself or be able to get creators to speak for you."

The growth strategy they favor is leveraging creators: Instead of running ads, make the product itself something that creators are willing to share.

This is also why they are optimistic about small teams developing consumer - grade products: As long as they can find a point that users are willing to share voluntarily, the product may explode rapidly.

3. Focus on whether users will use it repeatedly, not on creating a closed - loop system

Ultimately, growth boils down to one question: Will users come back and use the product again?

Many AI products fall into a misunderstanding: They try to create all - in - one tools but lack a core function that makes users use them repeatedly.

Instead, ask three questions:

Why did the user come for the first time?

Will they want to use it again tomorrow?

Will they be willing to tell their friends about it?

A product that can meet all three criteria is a viable one.

Section 4 | Why are niche products the most worth developing?

We've analyzed three types of products and the growth criteria. But the more difficult question is: How do you know you've chosen the right direction?

1. Is there no opportunity in popular fields?

Mignano used browsers as an example.

Two or three years ago, he saw several AI browser projects but decided not to invest: He thought that Chrome and Safari had already monopolized the market, leaving no room for new entrants.

However, today, products like ChatGPT Atlas and Perplexity are proving that AI has transformed browsers from search interfaces into operating systems, and products that were once impossible have become new entry points.

This leads to a new criterion for judging directions:

If it succeeds, will it completely change a user's behavior pattern?

2. What's truly worth betting on is culture, not technology

In consumer - grade products, it's never the technology itself that determines success but whether it can integrate into users' daily lives.

The key isn't how smart you are but whether you can capture the shift in people's attention.

This is also why they are optimistic about products that seem unremarkable, fun but not very practical:

Instagram isn't a better camera, but it defined the new behavior of taking a photo and sharing it immediately;

TikTok isn't a better video - making tool, but it made 15 - second videos a new way of expression.

The common feature of these products is that they start from niche areas and then force the mainstream to accept them.

3. The core of judgment is people, not ideas

In this dialogue, the two partners emphasized that when making investments, they not only look at the direction but also at whether the founders can turn an unpromising idea into a success.

Their selection criteria are:

Give priority to founders who can create a market rather than follow it;

Pay attention to products that can inspire new motivations rather than just meet existing needs.

How can you judge whether a niche direction is a real opportunity? You can ask three questions:

If it succeeds, which behavior pattern will it change?

Will this behavior integrate into users' lives, or is it just a tool function?

Can this direction only be pursued now, or could it have been done five years ago?

A direction that can answer all three questions is a new market truly opened up by AI.

This is also the core message that this dialogue wants to convey to entrepreneurs:

Don't do what everyone thinks is a good idea. Do what only you can see the opportunity in.

Conclusion | It's not about having more features but about capturing users' attention first

The challenge of consumer - grade AI lies not in developing the product but in getting users to open it, stay, and come back.

The answer given in this dialogue is not to compete in model development but to find directions that no one else has noticed but that you must pursue.

Because real blockbusters always emerge from niche areas.

As models become more powerful, it's getting easier to develop products, and choosing the right direction is becoming increasingly important. In the AI era, it's not about who has more features but who can get users to open the app every day.

You don't need to predict the future.

You need to have the courage to bet on an opportunity that others haven't seen right now.

📮 Original links:

https://www.youtube.com/watch?v=Z4L4ZqL1xqQ&t=605s

https://www.linkedin.com/in/mignano/

https://migna.no/

https://mignano.medium.com/lightspeed-ahead-4109742bcd75

https://lsvp.com/newsletter/2024 - 09/

https://www.bloomberg.com/company/stories/vc - michael - mignano - on - embracing - a - grow - or - die - mentality - cornell - tech - bloomberg/

This article is from the WeChat official account "AI Deep Researcher". The author is AI Deep Researcher. It is published by 36Kr with authorization.