I’m a fan of ‘requests for startups’ lists as they give a Zeitgeist for what investors are looking for (and think it’s a good opportunity’ in a period in time. This year, I’ve already published two of these lists – you can check Part 1 and Part 2 which were published in Q1 2025. When compared with older lists, like this one I published in March 2024, it’s clear to see how much AI and AI Agents have taken over.
Once again, a number of funds and accelerators have refreshed their RFS (requests for startups), so I’ve collected them all together below. Full disclosure, RFS are not a guarantee of funding. committing yourself to a startup as an entrepreneur should come from self conviction and after a process of research and market validation. But folks looking for ideas might get some inspiration from the lists below.
Y Combinator’s latest request for startups Summer 2025
This is the year of the AI Native or ‘Full Stack AI’ startups. I’ve previously covered the Spring 2025 RFS, below are ideas that YC is actively looking to fund in the Summer 2025 batch based on the YC blog:
1) Full-stack AI Companies
Start your own AI-staffed company to take on slow-moving incumbents.
These companies can completely rethink legacy industries by using AI-native operations, rebuilding entire product or service categories from scratch.
2) More Design Founders
Opportunity for designers to use AI themselves to launch their own products and build their own companies with taste.
Designer-founders who deeply understand user experience can now independently prototype and launch beautiful, functional tools using AI.
3) Voice AI
Phone calls haven’t changed in 100 years, opportunity to change that with voice AI bots.
Advances in voice synthesis, recognition, and contextual understanding make it possible to create intelligent agents that can hold real conversations.
4) AI for Scientific Advancement
Use AI to solve complex problems in drug discovery, chemical process optimization, metals & mining, or power grid optimization.
Founders can use AI to unlock insights in scientific domains that traditionally require years of experimentation and modeling.
5) AI Personal Assistant
Create systems that truly handle your emails, scheduling, and tasks without supervision.
The goal is a seamless assistant that acts like a chief of staff for individuals, freeing people from the cognitive load of daily management.
6) Healthcare AI
Reduce the $1+ trillion in administrative costs by automating healthcare data systems.
AI startups can streamline everything from billing to documentation, addressing deep inefficiencies in the U.S. healthcare system.
7) AI Personal Tutor
Deliver personalized education through multimodal AI explanations of complex topics.
Adaptive learning tools powered by AI can provide instant, tailored feedback and instruction in a way human tutors cannot scale.
8) Software Tools For Robots
Build the software foundation for the “ChatGPT moment” in robotics.
As physical robots become more capable, there’s a massive need for intuitive, general-purpose interfaces and programming environments.
9) The Future of Education
Reshape how 1.5 billion students learn with AI-powered personalized instruction.
Education is ripe for reinvention using AI to give each learner a unique path, adjusted in real-time to their progress and needs.
10) AI Residential Security
Bring commercial-grade AI security features to the $20B home security market.
Smart home devices with AI can provide 24/7 protection, anomaly detection, and decision-making that reduces false alarms and increases peace of mind.
11) Internal Agent Builder
Create tools for employees to build their own automation agents for repetitive tasks.
Non-technical employees should be able to build their own custom workflows and automations using intuitive AI interfaces.
12) AI Research Labs
Fund deep, open-ended AI research that may take years to commercialize.
This is a call for founders building the next DeepMind or OpenAI: organizations advancing the science of intelligence itself.
13) AI Voice Assistants for Email
Build voice interfaces that let people process emails hands-free.
There’s an opportunity to rethink how we interact with email through voice-based summarization, response drafting, and prioritization.
14) AI for Personal Finance
Provide unbiased, personalized financial advice at near-zero cost using LLMs.
Founders can now automate personalized money coaching—helping users save, invest, and plan with high trust and accessibility.
A16Z Speedrun
Below are some of the ideas that partners at A16Z $600M Games fund would like to actively fund. The team has shared more colour about these in a series of videos.
- AI Agent-Powered Creative Storytelling Platforms: Building platforms like a “next-gen Wattpad or Roblox” where AI agents act as creative assistants to help individuals compose their ideas into rich transmedia stories. These platforms should offer an end-to-end iterative workflow, potentially incorporating voice mode, multiplayer features, and focusing on niche distribution to help users go from idea to first draft quickly while allowing finer control.
- Social Simulation for Behavior Prediction: Developing systems to create a “crystal ball for human behavior“. This involves building models, potentially using generative agents in simulated environments, to predict human attitudes and behaviors. Such technology could be applied to answer questions in areas like sociology, politics, and marketing.
- AI-Driven Game Experiences Breaking Rules: Creating new game experiences that leverage AI to challenge traditional rules of game development and design. This includes opportunities like making world building non-uniform or hyperpersonalized to players or enabling solo developers led by storytellers to create complex games. They are interested in companies using AI for real-time content and interactive experiences that evolve with players.
- AI/Game Merging Consumer Apps: Developing consumer applications across various verticals (such as mental health, dating, fitness, social, education, learning, and health) that incorporate design, progression, and monetization elements from games. These apps should leverage AI to create engaging and evolving experiences.
- Vibe Creation Platforms for Consumers/Proumers: Building AI-native platforms specifically targeting consumers and “proumers” that enable them to easily create various forms of content like video, visual stories, games, or even apps. These platforms require an end-to-end workflow orchestrated by agents to handle complexity, allowing users to generate content quickly while maintaining creative control.
- AI/XR for Reinventing Core Consumer Verticals: Utilizing AI and/or XR technology to fundamentally reimagine and improve the core experiences in large, established consumer categories like dating, learning, and health. The goal is to disrupt existing models and deliver significantly greater value to users in these areas.
South Park Commons requests for curiosity
San Francisco accelerator South Park Commons (SPC) with a focus on AI shared what they called their ‘Requests for Curiosity‘. Below are some areas that suggest potential focuses for startups:
- Software Development & Management for Massive AI-Generated Code: Startups addressing the challenges of a future where code volume increases 10x to 100x due to AI. This includes rethinking version control, code review, maintainability, development planning, and interleaving code with design and other workstreams.
- Managing an Agentic Workforce: Companies building solutions to coordinate and oversee a fleet of AI agents used by businesses for functions like customer service, sales, HR, and accounting. This involves questions about shared systems of record, connectivity fabrics, and global deployment management views.
- Co-pilots & Agents for Technical Operations (NOC/SRE): Developing AI assistants and agents specifically for Technical Operations, Network Operations Centers, and Site Reliability Engineers. Focus areas include managing current deployments, handling increased AI-generated code, and determining the balance between automation and assistance in this domain.
- Novel Foundation Model Architectures: Startups challenging core assumptions about current foundation models to create architectural advances. This includes exploring ways to embed custom knowledge directly into models and enable interpretability as a core feature.
- Hardware & Software for Ubiquitous AI Inference: Solutions for enabling “always-on” AI inference efficiently, potentially involving novel chip architectures beyond current ARM designs or updated software like a restricted version of CUDA. Key considerations are power consumption, thermal management, and compute bottlenecks.
- Drone Software Stack Development: Building a unified, horizontal software platform for core drone functionalities (navigation, communication, data processing) or specialized vertical applications on top of such a platform, aiming to lower barriers to entry for software engineers.
- AI-Powered Creative Tooling for Children: Creating multimodal AI tools that offer children intuitive and self-directed ways to create images, videos, stories, cartoons, and music. The goal is a simple “multimodal canvas” like a modern “Photoshop” for kids.
- Multi-player AI Interfaces: Developing AI-native platforms that allow multiple users to collaborate on content creation and query management, addressing challenges like shared context, privacy, and query arbitration.
- Disrupting Incumbent Credibility Structures: Startups challenging established, low-quality forms of credibility in industries dominated by a few players (like corporate auditing or credit ratings). This involves leveraging AI democratization to compete on perceived quality.
- Uniquely AI-Unlocked Consumer Experiences: Creating novel consumer networks and applications that are only possible due to the availability of “intelligence on tap”. This includes exploring how AI mediates creator-consumer relationships, new non-text interaction loops, and personalized content discovery.
- The Future of the Creative Process: Tools and paradigms that enable creators to be limited only by their ideas, not lower-level skills. This involves new tools for design and production, evolving paths toward mastery, and new design paradigms for iteration.
- Content Consumption in a Zero Production Cost World: Platforms or methods addressing how content consumption changes as creation costs approach zero. This includes exploring hyper-personalization, the value of higher-friction experiences, and interactive media resembling games.
- User-Owned Data Infrastructure: Building the infrastructure required for users to own their data and potentially port personalized models across services, addressing privacy and how businesses adapt to users valuing their data.
- Marketing & Advertising in an Agentic World: Developing new forms of marketing, advertising, and brand experiences for a world where consumer agents mediate interactions. This involves adapting marketing metrics and performance marketing for agent-first scenarios.
- Reimagining Digital Environments for Agents: Creating digital environments and interaction patterns optimized for agentic systems rather than human limitations, such as websites readable by computers or systems interacting directly with databases.
- API Design for an Agentic World: Developing new API architectures, security protocols, and potentially a universal language or protocol for communication between AI agents and interactions with the physical world.
- High-Throughput, Low-Latency Human-Computer Interaction: Exploring new “Human APIs” and resulting applications/marketplaces enabled by faster, more natural interactions between humans and computers.
- Empowering Small/Solo Business Owners Against Private Equity: Tools, liquidity solutions, and novel business models that empower small and mid-cap business owners and solo entrepreneurs to compete with private equity rollups.
- AI for Public Sector Workflows: Developing AI co-pilots and tooling to enhance the productivity and scaffold the workflows of public servants, focusing on human-to-human interfaces and necessary safety guardrails.
- Next-Generation Search Engines with High Compute: Building search engines that leverage significantly more compute to perform deeper analysis, combine recommendation systems with LLMs, and solve problems justifying higher per-search costs.
- Technology Enhancing Human Non-Intelligence Aspects: Developing technology that amplifies the emotional, artistic, communal, and embodied aspects of being human as AI takes over intelligent tasks. This includes enhancing relationships, building community, and creating emotionally impactful experiences.
- Technology for End-of-Life Care: Creating solutions for reducing long-term care costs, combating loneliness in aging populations, supporting retiring digital natives, and improving end-of-life planning.
Requests for Startups (RFS) lists can be a valuable signal for founders and investors alike, offering insight into the types of problems investors believe are worth solving. However, it’s important to remember that RFS lists are not exhaustive roadmaps to success—they tend to focus on what’s newly possible rather than what’s necessarily viable, and following them too literally can lead to overcrowded spaces or derivative ideas. The best use of RFS is as a directional compass, not a blueprint—great startups often emerge by challenging the consensus, not following it.
That being said, it’s time to build! We’re actively investing in pre-seed AI native (including B2C) startups with a focus on Israeli-founded companies. It’s never too early to get feedback.
I'm a former general partner at google ventures, head of Google for Entrepreneurs in Europe and founding head of Campus London, Google's first physical hub for startups.
I'm also the founder of Techbikers, a non-profit bringing together the startup ecosystem on cycling challenges in support of Room to Read. Since inception in 2012 we've built 11 schools and 50 libraries in the developing world.
- Weekly Firgun Newsletter – February 27, 2026 - February 27, 2026
- The Rise of the Personal AI Agent - February 26, 2026
- Pre-Seed Is Stronger than ever, but less startups are getting funded - February 25, 2026
