I’m a fan of ‘requests for startups’ lists as they provide a Zeitgeist for what buyers are in search of (and suppose it’s a very good alternative’ in a interval in time. This yr, I’ve already revealed two of those lists – you’ll be able to verify Half 1 and Half 2 which had been revealed in Q1 2025. In comparison with older lists, like this one I revealed in March 2024, it’s clear to see how a lot AI and AI Brokers have taken over.
As soon as once more, quite a lot of funds and accelerators have refreshed their RFS (requests for startups), so I’ve collected all of them collectively beneath. Full disclosure, RFS usually are not a assure of funding. committing your self to a startup as an entrepreneur ought to come from self conviction and after a technique of analysis and market validation. However of us in search of concepts would possibly get some inspiration from the lists beneath.
Y Combinator’s newest request for startups Summer time 2025
That is the yr of the AI Native or ‘Full Stack AI’ startups. I’ve beforehand coated the Spring 2025 RFS, beneath are concepts that YC is actively trying to fund within the Summer time 2025 batch based mostly on the YC weblog:
1) Full-stack AI Corporations
Begin your personal AI-staffed firm to tackle slow-moving incumbents.
These firms can fully rethink legacy industries through the use of AI-native operations, rebuilding whole services or products classes from scratch.
2) Extra Design Founders
Alternative for designers to make use of AI themselves to launch their very own merchandise and construct their very own firms with style.
Designer-founders who deeply perceive consumer expertise can now independently prototype and launch stunning, practical instruments utilizing AI.
3) Voice AI
Cellphone calls haven’t modified in 100 years, alternative to vary that with voice AI bots.
Advances in voice synthesis, recognition, and contextual understanding make it potential to create clever brokers that may maintain actual conversations.
4) AI for Scientific Development
Use AI to resolve complicated issues in drug discovery, chemical course of optimization, metals & mining, or energy grid optimization.
Founders can use AI to unlock insights in scientific domains that historically require years of experimentation and modeling.
5) AI Private Assistant
Create techniques that really deal with your emails, scheduling, and duties with out supervision.
The objective is a seamless assistant that acts like a chief of workers for people, releasing individuals from the cognitive load of day by day administration.
6) Healthcare AI
Scale back the $1+ trillion in administrative prices by automating healthcare knowledge techniques.
AI startups can streamline the whole lot from billing to documentation, addressing deep inefficiencies within the U.S. healthcare system.
7) AI Private Tutor
Ship customized schooling via multimodal AI explanations of complicated subjects.
Adaptive studying instruments powered by AI can present instantaneous, tailor-made suggestions and instruction in a means human tutors can’t scale.
8) Software program Instruments For Robots
Construct the software program basis for the “ChatGPT second” in robotics.
As bodily robots turn into extra succesful, there’s a large want for intuitive, general-purpose interfaces and programming environments.
9) The Way forward for Training
Reshape how 1.5 billion college students be taught with AI-powered customized instruction.
Training is ripe for reinvention utilizing AI to offer every learner a novel path, adjusted in real-time to their progress and wishes.
10) AI Residential Safety
Deliver commercial-grade AI security measures to the $20B house safety market.
Good house gadgets with AI can present 24/7 safety, anomaly detection, and decision-making that reduces false alarms and will increase peace of thoughts.
11) Inner Agent Builder
Create instruments for workers to construct their very own automation brokers for repetitive duties.
Non-technical staff ought to be capable of construct their very own customized workflows and automations utilizing intuitive AI interfaces.
12) AI Analysis Labs
Fund deep, open-ended AI analysis which will take years to commercialize.
It is a name for founders constructing the following DeepMind or OpenAI: organizations advancing the science of intelligence itself.
13) AI Voice Assistants for E-mail
Construct voice interfaces that allow individuals course of emails hands-free.
There’s a possibility to rethink how we work together with e-mail via voice-based summarization, response drafting, and prioritization.
14) AI for Private Finance
Present unbiased, customized monetary recommendation at near-zero value utilizing LLMs.
Founders can now automate customized cash teaching—serving to customers save, make investments, and plan with excessive belief and accessibility.
A16Z Speedrun
Beneath are a few of the concepts that companions at A16Z $600M Video games fund want to actively fund. The workforce has shared extra color about these in a collection of movies.
- AI Agent-Powered Artistic Storytelling Platforms: Constructing platforms like a “next-gen Wattpad or Roblox” the place AI brokers act as artistic assistants to assist people compose their concepts into wealthy transmedia tales. These platforms ought to supply an end-to-end iterative workflow, probably incorporating voice mode, multiplayer options, and specializing in area of interest distribution to assist customers go from thought to first draft shortly whereas permitting finer management.
- Social Simulation for Conduct Prediction: Creating techniques to create a “crystal ball for human habits“. This entails constructing fashions, probably utilizing generative brokers in simulated environments, to predict human attitudes and behaviors. Such expertise might be utilized to reply questions in areas like sociology, politics, and advertising.
- AI-Pushed Recreation Experiences Breaking Guidelines: Creating new sport experiences that leverage AI to problem conventional guidelines of sport improvement and design. This contains alternatives like making world constructing non-uniform or hyperpersonalized to gamers or enabling solo builders led by storytellers to create complicated video games. They’re concerned about firms utilizing AI for real-time content material and interactive experiences that evolve with gamers.
- AI/Recreation Merging Client Apps: Creating client purposes throughout varied verticals (resembling psychological well being, relationship, health, social, schooling, studying, and well being) that incorporate design, development, and monetization parts from video games. These apps ought to leverage AI to create participating and evolving experiences.
- Vibe Creation Platforms for Shoppers/Proumers: Constructing AI-native platforms particularly concentrating on customers and “proumers” that allow them to simply create varied types of content material like video, visible tales, video games, and even apps. These platforms require an end-to-end workflow orchestrated by brokers to deal with complexity, permitting customers to generate content material shortly whereas sustaining artistic management.
- AI/XR for Reinventing Core Client Verticals: Using AI and/or XR expertise to basically reimagine and enhance the core experiences in massive, established client classes like relationship, studying, and well being. The objective is to disrupt present fashions and ship considerably higher worth to customers in these areas.
South Park Commons requests for curiosity
San Francisco accelerator South Park Commons (SPC) with a give attention to AI shared what they referred to as their ‘Requests for Curiosity‘. Beneath are some areas that counsel potential focuses for startups:
- Software program Improvement & Administration for Huge AI-Generated Code: Startups addressing the challenges of a future the place code quantity will increase 10x to 100x resulting from AI. This contains rethinking model management, code assessment, maintainability, improvement planning, and interleaving code with design and different workstreams.
- Managing an Agentic Workforce: Corporations constructing options to coordinate and oversee a fleet of AI brokers utilized by companies for capabilities like customer support, gross sales, HR, and accounting. This entails questions on shared techniques of report, connectivity materials, and international deployment administration views.
- Co-pilots & Brokers for Technical Operations (NOC/SRE): Creating AI assistants and brokers particularly for Technical Operations, Community Operations Facilities, and Website Reliability Engineers. Focus areas embody managing present deployments, dealing with elevated AI-generated code, and figuring out the steadiness between automation and help on this area.
- Novel Basis Mannequin Architectures: Startups difficult core assumptions about present basis fashions to create architectural advances. This contains exploring methods to embed customized data immediately into fashions and allow interpretability as a core function.
- {Hardware} & Software program for Ubiquitous AI Inference: Options for enabling “always-on” AI inference effectively, probably involving novel chip architectures past present ARM designs or up to date software program like a restricted model of CUDA. Key concerns are energy consumption, thermal administration, and compute bottlenecks.
- Drone Software program Stack Improvement: Constructing a unified, horizontal software program platform for core drone functionalities (navigation, communication, knowledge processing) or specialised vertical purposes on high of such a platform, aiming to decrease obstacles to entry for software program engineers.
- AI-Powered Artistic Tooling for Youngsters: Creating multimodal AI instruments that provide youngsters intuitive and self-directed methods to create photographs, movies, tales, cartoons, and music. The objective is a straightforward “multimodal canvas” like a contemporary “Photoshop” for youths.
- Multi-player AI Interfaces: Creating AI-native platforms that permit a number of customers to collaborate on content material creation and question administration, addressing challenges like shared context, privateness, and question arbitration.
- Disrupting Incumbent Credibility Buildings: Startups difficult established, low-quality types of credibility in industries dominated by a couple of gamers (like company auditing or credit score rankings). This entails leveraging AI democratization to compete on perceived high quality.
- Uniquely AI-Unlocked Client Experiences: Creating novel client networks and purposes which might be solely potential as a result of availability of “intelligence on faucet”. This contains exploring how AI mediates creator-consumer relationships, new non-text interplay loops, and customized content material discovery.
- The Way forward for the Artistic Course of: Instruments and paradigms that allow creators to be restricted solely by their concepts, not lower-level abilities. This entails new instruments for design and manufacturing, evolving paths towards mastery, and new design paradigms for iteration.
- Content material Consumption in a Zero Manufacturing Value World: Platforms or strategies addressing how content material consumption modifications as creation prices strategy zero. This contains exploring hyper-personalization, the worth of higher-friction experiences, and interactive media resembling video games.
- Person-Owned Information Infrastructure: Constructing the infrastructure required for customers to personal their knowledge and probably port customized fashions throughout providers, addressing privateness and the way companies adapt to customers valuing their knowledge.
- Advertising and marketing & Promoting in an Agentic World: Creating new types of advertising, promoting, and model experiences for a world the place client brokers mediate interactions. This entails adapting advertising metrics and efficiency advertising for agent-first situations.
- Reimagining Digital Environments for Brokers: Creating digital environments and interplay patterns optimized for agentic techniques reasonably than human limitations, resembling web sites readable by computer systems or techniques interacting immediately with databases.
- API Design for an Agentic World: Creating new API architectures, safety protocols, and probably a common language or protocol for communication between AI brokers and interactions with the bodily world.
- Excessive-Throughput, Low-Latency Human-Pc Interplay: Exploring new “Human APIs” and ensuing purposes/marketplaces enabled by quicker, extra pure interactions between people and computer systems.
- Empowering Small/Solo Enterprise House owners Towards Personal Fairness: Instruments, liquidity options, and novel enterprise fashions that empower small and mid-cap enterprise house owners and solo entrepreneurs to compete with non-public fairness rollups.
- AI for Public Sector Workflows: Creating AI co-pilots and tooling to boost the productiveness and scaffold the workflows of public servants, specializing in human-to-human interfaces and needed security guardrails.
- Subsequent-Era Search Engines with Excessive Compute: Constructing engines like google that leverage considerably extra compute to carry out deeper evaluation, mix advice techniques with LLMs, and clear up issues justifying increased per-search prices.
- Expertise Enhancing Human Non-Intelligence Points: Creating expertise that amplifies the emotional, creative, communal, and embodied facets of being human as AI takes over clever duties. This contains enhancing relationships, constructing neighborhood, and creating emotionally impactful experiences.
- Expertise for Finish-of-Life Care: Creating options for decreasing long-term care prices, combating loneliness in growing old populations, supporting retiring digital natives, and enhancing end-of-life planning.
Requests for Startups (RFS) lists could be a beneficial sign for founders and buyers alike, providing perception into the sorts of issues buyers imagine are price fixing. Nevertheless, it’s necessary to keep in mind that RFS lists usually are not exhaustive roadmaps to success—they have a tendency to give attention to what’s newly potential reasonably than what’s essentially viable, and following them too actually can result in overcrowded areas or spinoff concepts. The very best use of RFS is as a directional compass, not a blueprint—nice startups usually emerge by difficult the consensus, not following it.
That being mentioned, it’s time to construct! We’re actively investing in pre-seed AI native (together with B2C) startups with a give attention to Israeli-founded firms. It’s by no means too early to get suggestions.
I am a former common associate at google ventures, head of Google for Entrepreneurs in Europe and founding head of Campus London, Google’s first bodily hub for startups.
I am additionally the founding father of Techbikers, a non-profit bringing collectively the startup ecosystem on biking challenges in assist of Room to Learn. Since inception in 2012 we have constructed 11 faculties and 50 libraries within the creating world.
Associated
