Most startups equate the method of fundraising to relationship – founders need to usually kiss a whole lot of frogs till the discover the correct match. Traders are usually targeted on stage, geo and in some circumstances sector, however there are at all times exceptions so it’s exhausting for startup founders to evaluate who could be to embark on their journey. As well as, founders fascinated by beginning an organization could be overwhelmed by selection, as there are such a lot of issues to deal with with know-how, nevertheless it may very well be comforting to know that buyers are all in favour of these areas within the first place.
That’s why lists like Ycombinator’s Requests for Startups (RFS) assist entrepreneurs fascinated by beginning an organization (or choosing an issue) and is usually a good barometer for founders on what’s attention-grabbing to buyers proper now. It’s been some time since they’ve up to date their RFS checklist and it captures all of the buzzword’s of right now’s tech world: AI, defence tech, local weather, spatial computing, and so forth. I’ve lately come throughout a number of of such lists and I believed it may very well be helpful to combination them and share them right here.
Ycombinator’s RFS 2024
- Making use of machine studying to robotics – Whereas shopper use-cases function closely in science fiction, among the ignored and most instantly addressable purposes for robots are B2B
- Utilizing machine studying to simulate the bodily world – firms changing current simulations with ML-based ones, together with firms utilizing ML-based simulations to open new markets presently unaddressable
- New defence know-how – New firms that promote to the DoD like Palantir and Anduril
- Deliver manufacturing again to America – New ML-based robotics methods will make it doable to automate much more, which can cut back the cost-of-labor arbitrage that pushed manufacturing to different nations within the first place.
- New area firms – If we’re coming into a future with entry to area being as routine and cheap as business air journey, transport or trucking… what new companies does that unlock?
- Local weather tech – We have now a good probability of avoiding catastrophic local weather change if startups supply business options to decarbonize society or take away carbon from the ambiance.
- Business open supply firms – being open supply is a strong solution to achieve developer adoption and promote to enterprises rather a lot sooner.
- Spatial computing – with the launches of the Apple Imaginative and prescient Professional and the Meta Quest 3 a brand new private computing platform could evolve over the following decade. Searching for startups fixing sensible use circumstances that transcend gaming.
- New enterprise useful resource planning software program (ERPs) – new startups that construct software program that helps companies run.
- Developer instruments impressed by current inside instruments – instruments or frameworks that had been constructed by programmers at their earlier firm to assist remedy their very own significantly painful or repetitive issues.
- Explainable AI – For society to reap the total advantages of AI, extra work must be completed on explainable AI. Searching for startups constructing new interpretable fashions or instruments to elucidate the output of current fashions
- LLMs for handbook again workplace processes in legacy enterprises – LLMs enable entire classes of handbook processes to be automated in ways in which weren’t doable till lately. The place there’s linguistic ambiguity or some quantity of subjective analysis wanted, LLMs come into their very own.
- AI to construct enterprise software program – Sooner or later, each enterprise may have their very own customized ERP, CRM or HRIS that’s frequently updating itself as the corporate itself is altering.
- Stablecoin finance – in search of nice groups constructing B2B and shopper merchandise on prime of stablecoins, instruments and platforms that allow stablecoin finance and extra stablecoin protocols themselves
- A solution to finish most cancers – improvements within the MRI {hardware}, the AI algorithms to interpret scans and cut back false positives, and the enterprise fashions and shopper advertising to make it a viable enterprise
- Basis fashions for organic methods – extremely technical founders constructing foundational fashions from scratch in any a part of biology or drugs
- The managed service group mannequin for healthcare – Searching for startups constructing on the MSO mannequin that allows docs to run their very own clinics by (1) offering them software program that may deal with again workplace duties corresponding to billing and scheduling and (2) channeling sufferers to them.
- Eliminating middlemen in healthcare – from utilizing AI to automate repetitive human jobs to exploring new and higher enterprise fashions for offering care.
- Higher enterprise glue – By producing customized code for unwidespread, company-specific use circumstances, massive language fashions have the potential to remove the necessity for glue code altogether together with ETL pipelines, integrations, and customized workflows.
- Small fine-tuned fashions as an alternative choice to big generic ones – when finely tuned with acceptable knowledge, smaller, specialised LLMs can yield comparable outcomes at a fraction of the fee. Searching for firms creating instruments to facilitate development of or fantastic tuning fashions.
Andreessen Horowitz Client Staff Alternatives for startups
I’m a giant fan of the content material A16Z has been publishing on generative AI over time. Certainly one of my favorite items has been The Abundance Agenda, a deck on how AI will remodel shopper know-how, packing inside areas of alternative, or in different phrases, requests for startups.
Content material creation:
- Technology instruments that kill the “clean web page” downside – ideating,
- Making open supply fashions accessible – create the interface to utilise open supply tech within the browser (or an app).
- Creating remixable outputs – permitting creators to show their prompts and make their work immediately remixable
- Enabling shoppers to construct content material creation apps – merchandise that assist customers “chain” collectively fashions and prompts behind the scenes – after which save these workflows and / or publish them for others to make use of.
Content material enhancing:
- Proudly owning multi-media workflows – mix a picture with textual content, music with video, or animation with a voiceover. Alow customers to generate, refine, and sew totally different content material sorts in a single workspace
- Enabling in-platform refinement – AI merchandise may also help customers determine what could be improved, after which mechanically make these adjustments.
- Iterating with clever editors – merchandise that allow customers to take an current output and refine it (ex. regenerate one body or function) with out fully ranging from scratch.
- Mechanically repurposing content material – leveraging AI to automate handbook enhancing to repurpose a bit of content material for various platforms.
Productiveness positive factors
- Brokers that act as methods of motion – brokers that may full widespread shopper duties like reserving a restaurant or discovering and sending a present to a pal.
- Voice-first apps – voice dictation and summarisation AI apps permitting shoppers to share extra complicated and even unfinished ideas.
- Apps that present in-flow help – an AI assistant like Microsoft’s ‘Clippy’ that magically seems the place and when wanted to keep away from context switching.
- “Construct your personal” workflows – no-code automations and workflows – LLMs performing as intermediaries to permit customers to sew collectively rather more complicated flows than pre-AI.
Companionship
- Differentiated worth prop vs. generalist chat merchandise – AI companion merchandise hat concentrate on content material that mainstream fashions aren’t good at (or don’t enable), like fictional function performs or erotica.
- New strategies of interplay – past textual content field interplay, a companion you may summon anyplace, with a voice, avatar, and animation that appears like an actual pal hanging out with you
- Apps that allow reminiscence and development – A companion ought to get to know you higher over time, bear in mind your earlier conversations, and alter the character of the connection.
Social
- Hybrid AI x human communities – messaging & social apps the place bots are handled as equal residents. Bots ought to be capable to be part of chats with you and mates, and weigh in or spark discussions.
- Reside, interactive leisure at large scale – AI permits anybody to be an entertainer with the assistance of generative avatars. AI characters are already internet hosting interactive streams and exhibits wherein they reply to viewers questions, feedback, or votes in real-time.
- Subsequent-gen avatars for next-level communication – AI provides shoppers a hyperrealistic digital likeness they will use to immediately generate and share these belongings.
- AI to make IRL matches – What if you happen to may chat with a bot that learns about you on a deeper degree and makes use of this data to make a curated set of matches?
Training
- Multimodal potential – apps that enable customers to “enter” questions, matters, or concepts in all kinds (audio, picture, textual content, and even video) and get a response within the media kind that helps them be taught greatest.
- New interfaces that break the “edtech” mould – the rise of extra informal, experiential studying at scale.
Private Finance
- Cross-account visibility and administration – Right this moment’s AI merchandise can analyze and transfer cash between accounts – as brokers enhance, they’ll make trades throughout accounts.
- Auto-optimization throughout merchandise – Many shoppers are “overpaying” for his or her debt, insurance coverage, and payments, AI brokers can take over this course of.
- Programmatic investing utilizing pure language – Utilizing pure language or resolution timber, shoppers with no data of code can construct algorithms that execute trades for them.
- Advanced transactions transfer from companies ? AI – tax planning or wealth administration augmented and in some circumstances changed with AI.
Ark Make investments’s Large Concepts 2024
Cathy Wooden, the founding father of Ark Make investments printed the ARK Large Concepts 2024, an annual report titled ‘Disrupting the Norm, Defining the Future’. It identifies 5 innovation platforms converging and defining the technological period: AI, public blockchains, power storage, robotics, and multiomic sequencing.
- Technological Convergence – The worldwide fairness market worth related to disruptive innovation may enhance to 60% by 2030.
- Synthetic Intelligence – Scaling international intelligence and redefining work: AI coaching prices ought to proceed to fall 75% per 12 months.
- Sensible Contracts – Powering the internet-native monetary system, good contract networks may generate charges of $450bn in 2030.
- Digital Wallets – Digital wallets may develop choose vertical software program platforms’ revenues to $27-$50bn in 2030.
- Robotics – Generalizing automation, because of the convergence of AI software program and {hardware}. Generalizable robotics symbolize a $24 trillion-plus international income alternative.
- Digital Customers – Transitioning towards digital leisure, the place spending may educate $23 trillion in 2030.
- Electrical Automobiles – Decrease battery prices powering adoption imply EV gross sales may attain 74 million in 2030.
- Robotaxis – Robotaxi platforms may create $28 trillion in enterprise worth in 2030.
- Multiomic Instruments & Expertise – Translating organic insights into financial worth: R&D spending may decline by greater than 25%, because of multiomic instruments and know-how.
- Reusable Rockets – Satellite tv for pc connectivity revenues may exceed $130bn per 12 months in 2030.
- Autonomous Logistics – World autonomous supply income may attain $900bn by 2030.
- Bitcoin Allocation – Rising the function of bitcoin in funding portfolios. Over the past seven years, bitcoin’s annualized return has averaged round 44%.
- Bitcoin in 2023 – After challenges in 2022, bitcoin’s worth surged 155% final 12 months, reaching $827 billion in market cap.
- Precision Therapies – Curing illness extra effectively and fewer expensively. The enterprise worth of firms targeted on precision therapies may attain $4.5 trillion by 2030.
- 3D Printing – Revenues may develop 40% at an annual price to $180bn by 2030.
Sarah Guo – Conviction Capital and Embed Accelerator
Sarah Guo believes that with present developments in generative AI, and LLMs particularly, good founders could be up and operating with a product that customers need in a matter of weeks. For that function, she launched Embed, an accelerator program.
- Verticalized Video Understanding – machine interpreted video. there’s a wealth of alternative in dwelling safety, public security, office security, well being/health, contextual promoting, automated video enhancing, buyer training/assist, retail (checkout and loss) and parking/visitors administration.
- Deepfake Detection – demand for a cell software that gives deepfake detection, Caller ID, spam blocking and even perhaps dwell translation or agentic experiences for name dealing with.
- Your Workshop Assistant – construct assistive tooling that both helps engineers slender down their design area extra effectively or performs some set of menial duties for them as a place to begin; second, combining generative fashions with post-processing work to “clear up”, like simulation for validation and different post-processing, to cut back the burden on zero-shot mannequin output.
- KYC/KYB Automation – emulate a few of what a human compliance agent would possibly do of their investigation, and, at minimal, can parallelize workstreams and floor potential pink flags.
- One Large Leap for Healthcare Admin -Breakthroughs in transcription, speaker diarization, translation, summarization, LLM net navigation/kind filling, and basic retrieval and reasoning will revolutionize the affected person encounter and claims / prior auth workflows
- Your Private Vendor – Basis fashions allow the at-scale era of different product listings and likewise due to this fact A/B testing of variants to enhance conversion and probably optimize for various audiences
- AI Therapist & Coach – One of many emergent use circumstances with the preliminary wave of LLMs has been a mixture of remedy and private teaching, however the present fashions are poor on reminiscence, security and safety.
- Automated Root Trigger Evaluation – there’s a chance for automated root trigger evaluation to considerably enhance incident decision time, and the expertise for engineers. A light-weight “agent” with entry to logs and metrics can, to begin, retrieve related data (e.g. service statuses, previous error logs, comparable prior incidents) and counsel fixes based mostly on what earlier resolutions had been. After an incident, the identical agent may very well be used to supply a “greatest follow” decision for future incidents. and generate a publish mortem. Long run, brokers would possibly even be capable to mechanically repair widespread reoccurring points
- Finish-to-Finish Authorized Outcomes – automate sure end-to-end, transactional companies supplied by regulation corporations corresponding to immigration agency in a field, trademark submitting or contract evaluations.
- Excessive-Consideration Analysis – LLMs are actually good at shortly studying and synthesizing lots of of pages of content material on any matter. Is there a superb shopper product expertise to be constructed round constructing a dynamic, interactive, complete question expertise for high-intent, high-consideration purchases?
- Workforce Tetris – Recruiting, employment compliance/admin and logistics requires a backend database and scalable workflows, however the subsequent era of workforce software program shouldn’t put that burden on managers or employees.
- Subsequent-Technology Autocomplete – a system that learns your model and completes every thing it’s good to write based mostly in your historical past, with privateness and velocity.
- The All-Seeing Eye – {Hardware} and storage must be rethought from the bottom up within the age of semantic video understanding, and highly effective on-device fashions. A full-stack safety companies agency may see extra, value much less, and supply a step-function higher expertise.
- At all times Choose Up the Cellphone – Voice era high quality and LLM functionality are approaching the power to deal with many transactional calls. What’s lacking is the final mile — distribution, buyer journey design, guardrails and workflow automation.
- Developer in a Field – the power to go from a human description of a difficulty to a draft resolution, in code, to the issue.
- AI Static Evaluation Instruments – inside tooling that mechanically triages and prioritises points recognized by different methods.
- AI-Native MMPGs and Social – What would a recreation world populated by AI’s be like? What if the following era of leisure is customized generations?
- AI Video Technology, Enhancing and Understanding – Firms that democratise video manufacturing, enhancing, personalisation and understanding
- Net Content material APIs – present net content material APIs lack the flexibleness & function set required to energy massive scale net purposes. There doesn’t exist an internet search API that has entry to web page content material, parsed outlinks from the web page, and even edit historical past.
- The Tireless (Junior) Monetary Analyst – international accounting, tax, monetary reporting and compliance requirements are all codified in pure language. Remodel monetary and accounting software program from databases to context-aware, proactive processors.
- Autonomous HR (and IT) Helpdesk – A website populated with course of documentation, ever-changing compliance wants, complicated coverage software, kinds, and pure language communication is ripe for assault by LLMs.
- Technical Help – a “debugging copilot” for the engineers presently working in technical assist.
These are only a small number of ‘requests for startups’ or fairly, alternatives for startups leveraging know-how. The facility of LLMs is advancing quickly, which means that we’re simply at the start of this know-how wave. Founders selecting to work on non-obvious, massive area of interest issues, can nonetheless get an early mover benefit. It’s time to construct! I additionally advocate Erik Torenberg’s podcast ‘Requests for Startups‘ for in-depth deep dives on varied verticals.
Shameless plug: At Remagine Ventures, we again pre-seed startups on the intersection of tech, leisure, gaming and commerce. If you happen to’re an Israeli or UK startup constructing in these areas and also you’d prefer to get fast suggestions and VC-friendly recommendation, don’t hesitate to get in contact.
Eze is managing companion of Remagine Ventures, a seed fund investing in bold founders on the intersection of tech, leisure, gaming and commerce with a highlight on Israel.
I am a former basic companion 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 colleges and 50 libraries within the creating world.
Newest posts by Eze Vidra (see all)
