The 2025 AI Breakthroughs
Hey everybody! I hope you had a fantastic Christmas and new year’s eve.
Like every day, today is day 1.
To start 2026, I wanted to share:
The 14 best AI resources I wrote in 2025
My best clips from 2025 interviews (founders + top funds)
The companies I backed in 2025 + the founder pattern behind them
40% off The AI Opportunity
Before, I wanted to thank you all that have been following, reading, learning and supporting me these years.
So I created a 40% discount for The AI Opportuity annual plan.
Start 2026 strong: get 40% off The AI Opportunity annual plan today and lock in a full year of market analysis and breakdowns from the best in the market.
Want to Sponsor The AI Opportunity & get your product in front of thousands of founders worldwide? Reach out to g@guillermoflor.com
Trending right now: The Data Room Template VCs love, 27 Most Promising AI Startup Pitch Decks Backed by Top Investors in 2025
The AI Opportunity: the 14 best AI resources I wrote in 2025 (annual recap)
If you only read a few things from The AI Opportunity this year, make it these.
These are the posts where the frameworks were the clearest, the opportunity was the biggest, or the thesis actually changed how you see the market.
A) The core frameworks (how to think)
1) Sequoia’s AI Playbook: The Agent Economy, Application Layer, and How to Win in 2025
The foundational AI strategy doc.
LINK↗
2) Sequoia AI Ascent: lessons by OpenAI, Nvidia, Anthropic, LangChain and Ramp
The most actionable “operator-level” insights from the people shipping the frontier.
LINK↗
3) Andrej Karpathy’s full guide on how to use LLMs (Jul 26)
The practical user manual for getting real leverage out of LLMs.
4) The 7 Most Powerful Moats for AI Startups (YC)
The moat checklist every AI founder and investor should internalize.
LINK↗
5) The Vertical AI Playbook + 200 Vertical AI Startups
A full framework of what wins in vertical AI, plus a market map you can steal.
LINK↗
B) The big opportunity shifts (where the money moves)
4) Inbox → Autopilot: AI Personal Assistants Are the Next $100B Land-Grab
Why the assistant layer becomes the new distribution surface.
LINK↗
5) Outcome based AI is killing SaaS
Why pricing and GTM are about to change more than product does.
LINK↗
6) The battle for search is on (and there’s a huge opportunity window)
The shift from Google-first to GPT-first discovery, and how startups exploit the crack.
LINK↗
7) The Next Billion-Dollar Opportunity in Voice AI: Licensed Voice/Image Rights
A new “rights economy” forming under voice, avatars, and synthetic media.
LINK↗
C) The execution playbooks (how to build)
8) The Enterprise AI Playbook: what the successful 5% are doing different
What real deployments look like, and why most companies fail at adoption.
LINK↗
9) Why AI Code Review Is the $1.5T Blind Spot in Software Engineering
Where the next productivity explosion happens inside engineering teams.
LINK↗
10) The $1.5T AI Roll-Up General Catalyst Playbook
Buy distribution, automate the work, scale step by step. A blueprint for AI-enabled consolidation.
LINK↗
11) The AI Consulting Playbook: how this founder is building McKinsey of AI
What to sell, who to sell to, and the delivery stack that actually works.
LINK↗
D) The frontier signals (what to watch next)
12) The Rise of Neolabs: where the next AI breakthroughs will come from (11 labs to follow)
A watchlist of the next wave of frontier builders.
LINK↗
13) Inside Mira Murati’s Thinking Machines Lab: the $2B seed and the next AI startup wave
What “new lab formation” signals look like, and how the ecosystem changes around them.
LINK↗
14) Sh*t, are we in a bubble? 10 things you need to understand about the OpenAI bubble
The clearest risk framework I wrote this year. Useful whether you agree or not.
LINK↗
2. 2025 Interviews 🎙️
This year, I was really lucky
In between building, investing, and writing, I had the opportunity to sit down with exceptional people who’ve built companies, funds, and careers at a level most of us spend years trying to understand from the outside 😅
Throughout these interviews, one thing became clear very quickly: sometimes the highest-leverage thing you can do isn’t moving faster. It’s pausing and listening
Listening to how great founders actually think through decisions.
Listening to how top investors talk about risk, patience, and conviction when there’s real capital on the line.
Listening to the trade-offs they made, the things they got wrong, and the principles that stayed consistent over time.
Those conversations of course don’t replace learning by doing, but they compressed many years of it.
CLIPS OF THE YEAR 🔥
Mark Cuban: Entrepreneur & investor
When Mark spoke to us about why most founders shouldn’t raise venture capital. Why avoiding VC can be the most underrated path to building real wealth.
Barbara Corcoran: Entrepreneur & Shark Tank investor
When Barbara told us to “lose the suits“ 😅; on why founders should show their faces instead of hiding behind polished brands.
Robert Herjavec: Entrepreneur & Shark Tank investor
Robert on how to pitch. Why great pitches start with trust and presence, not product slides, and why investors back people before they back ideas.
Daniel Lubetzky: Founder of KIND Snacks
Daniel on why he avoids direct AI startup bets and expects a “bloodbath” 🔴. He explains why he prefers backing a small number of elite funds instead of chasing individual winners.
Chris Best: Co-founder & CEO of Substack($1.1B+)
Chris on how to get people to actually use a new app, how they did it and why trust and habit matter more than features early on.
Victor Riparbelli: Founder & CEO of Synthesia($3B+)
Victor Riparbelli on grinding for three years without PMF at Synthesia. Now they turned down a reported $3B acquisition offer from Adobe 🤯
John C. Santora: CEO of WeWork ($47B+ Peak)
John C. Santora on how there is no substitute for the work. No shortcuts or hacks. Just consistency over time. 💪
Staffan Helgesson: Partner at Creandum
Staffan (with 1 in 6 portfolio companies becoming unicorns 😳) on why the best investors back founders who don’t need help to win, and how Creandum’s real edge comes from long-term conviction and founder network.
Martin Mignot: Partner at Index Ventures
Martin Mignot on backing Revolut($75B+) before the numbers worked, and why conviction in founders and trajectory mattered more than early margins or revenue.
Anna Piñol: Partner at NFX
Anna Piñol (Ex-YC) on what NFX looks for in startups: strong founding teams that can build & sell, shared history, speed, contrarian thinking, and customer problems people feel Monday morning.
Spenser Skates, founder of Amplitude
And many more conversations that shaped how I think about building, investing, and choosing what actually matters, but…
What stayed with me?
Across very different people and paths, the same patterns kept coming up:
The best founders are obsessed with clarity, not cleverness.
The best investors optimise for conviction and ownership, not activity.
Almost everyone underestimated how long the important things would take and overestimated how much speed alone would matter.
These conversations didn’t hand me answers, but they changed how I evaluate decisions.
What deserves speed?
What deserves patience?
Where does conviction actually matter?
Looking ahead…
I want to do more of this.
More conversations with people who’ve actually built things. More time spent listening before forming opinions. And more effort turning those insights into frameworks that are useful beyond a single interview.
That’s something I’ll be doubling down on in 2026.
3. Companies I Backed in 2025 💰
My Founder Thesis: Investing in best founders and helping them go viral
Looking back this week, I found myself thinking about why I back in the founders that I back. What pulls me towards some more than others?
After some thinking, I realised something. On the surface, the investments span very different areas within AI, devtools, healthcare, fintech, vertical software and more, but somehow, all of the companies I backed shared some very specific traits, more precisely, the founders.
This is something Harry Stebbings once articulated very well:
A deep superiority complex: a real belief they can do what others can’t.
And a deep inferiority complex: the feeling they’re never doing enough.
That mix creates a specific kind of founder: confident enough to take on hard problems, restless enough to keep pushing…
Almost every founder I backed this year fits that pattern. They’re not trying to be good, they’re trying to be world class at one very specific thing.
I didn’t set out with a rigid thesis, but looking back, this one revealed itself.
Im grateful that this list ended up longer than I expected 😅
Here’s a snapshot of some of the teams I backed in 2025:
HappyRobot (a16z, YC) — Intelligence that runs your operations
Pablo Palafox: Previously PhD Candidate @ TU Munich working on Computer Vision, research intern at Reality Labs (Meta).
Javier Palafox: The co-founder and COO of Happyrobot. Previously, he served as CFO for a $300M CPG brand.
Luis Paarup: Co-founder of Happyrobot (happyrobot.ai), working on AI agents that speak on the phone.
Pally (Pioneer Fund, YC, Founders Inc) — Personal relationship management
Haz Hubble: High-school dropout and super connector. Previously founded the London Founder House. Designer, salesman, engineer, accountant (weird mix but it works).
Wyatt Lansford: ML Engineer in robotics & autonomy. Won the DARPA AlphadogFight Trials, built autonomous planes & robots. Acquired by Shield AI. Now building human x AI interfaces.
Throxy (YC, Base10) — Vertical AI agents automating the full sales funnel
Bergen Merey: Consulting generalist turned builder. Realised decks don’t scale, systems do. Built my first online business at 14. Now building throxy, where we turn “Software-as-a-Service” into Service-as-a-Software, using AI Agents to transform sales. We automate the repetitive.
Arnau Ayerbe Garcia: Founder focused on making it easier to speak with your ideal customers. Eliminating repetitive manual outreach. Believes the future of growth is human connection.
Pablo Jiménez de Parga Ramos: Founder obsessed with running outbound for founders, so they can focus on actually running their business.
Azimov (a16z Speedrun) — Building the missing half of the digital world
Kevin Jiang: UCLA CS/Econ drop out. Building viral products since 12. ML patent holder at 18. Connection is the only algorithm that matters.
Nelson Lai: CTO of Riot Games China. CEO of Blitz.gg China. Founding Engineer at Picasa (acq. by Google in 2004). Studied EECS at MIT. Born in Montana. Grew up in California.
Cubic (Peak XV, YC) — AI-powered code review
Paul Sanglé-Ferrière: Founder & CEO. 3x founder, previously an ML engineer and a product manager at Tessian.
Nozomio (LocalGlobe, CRV, YC) — Context augmentation for agents
Arlan Rakhmetzhanov: 18. Finished homework at 6 a.m. to build during school hours — until investor calls landed him in the principal’s office. Dropped out in 11th grade and raised $1M within days. Launched his first startup at 15, scaling to 20,000 users. A year later, cold-emailed into Stanford research and worked with a Caltech professor on Cobra, a tool for detecting defects and suspicious code patterns.
QFEX (General Catalyst, YC) — A 24/7 exchange for traditional markets
Annanay Kapila: Annanay is Founder and CEO of QFEX. He met his co-founder Joshua at Cambridge, where he studied Maths and conducted Quantum Computing research. Since then, he has worked as a Quantitative Trader at Flow Traders and Tower Research Capital in London, and is now building QFEX as the exchange he always wish he had to trade on.
Joshua Wharton: Former equities engineer at Citadel, previously a senior software engineer across high-performance and security-focused teams. Cambridge-educated, with deep experience in trading systems and production-grade infrastructure.
Minerva Intelligence (YC) — AI replacing SMB bookkeeping
Peter Zhu: Founder building AI accounting agents that streamline operations for accounting firms and SMBs. Backed by Y Combinator, Amino Capital, Goodwater Capital, and senior operators from DoorDash, Affirm, and Bill.com. Background in Math and Computer Science from Brown.
Om Agarwal: Founder building AI accounting agents for accounting firms and SMBs, backed by Y Combinator, Amino Capital, Goodwater Capital, and senior leaders from DoorDash, Affirm, and Bill.com. Previously built next-gen practice management software used by hundreds of clinicians and advised state health authorities on AI.
Novoflow (YC) — AI employees for medical operations
Mathieu Rihet: Emory University Scholar and former medical interpreter. Now building AI employees to automate up to 99% of manual workflows in medical clinics.
Georges Casassovici: Former Lawrenceville School student turned Carnegie Mellon admit (dropped out). Technical since age 8, with 5+ years in AI. Currently building AI employees for medical clinics.
YouShift (YC) — The operating system for healthcare staffing
Lucía Vives Martorell: Barcelona-born biomedical scientist, educated at UCL and Harvard.
Jota Chamorro: Focused on making doctor scheduling easy, fast, and fair. Background in computer science at Harvard.
Adolfo Roquero Gimenez: Former Google Ads Infrastructure engineer, building large-scale, low-latency systems for serving key-value data. Studied Computer Science at Harvard, graduating with honors.
Klavis AI (YC) — Open-source MCP integrations for AI apps
Xiangkai Zeng: Former Senior Software Engineer on Google Gemini at Google DeepMind and co-author of the Gemini paper. MS in Computer Science from Carnegie Mellon; BEng in Computer Science from Beihang University.
Zihao Lin: Former Senior Software Engineer at Lyft, leading recommendation systems and data infrastructure serving millions of users. MS in Computer Science from Northeastern; BEng from Beihang University. Research in NLP and internship at Baidu.
Clova (Founders Inc) — Cursor for video editing
William Wang: Previously founded DashBill, processing $2M+ for 2,500+ users. Filmmaker by craft, engineer by instinct; has worked with UC Riverside, the City of Riverside, and Justin Kan (Twitch).
Ryan Bailey: Previously co-founded DashBill, scaling a nonprofit finance platform to $2M+ processed volume and thousands of users. Background in operations and product execution.
Gyan Prakash: Co-founder of Clova
Cua (Peak XV, YC) — Docker for computer-use agents
Francesco Bonacci: Founder and systems-first engineer. Ex-Microsoft/Xbox Gaming AI, with 7+ years building cloud and desktop platforms from POC to production. Deep in computer-use and UI-focused AI agents, open-source by default, community-driven, and relentlessly practical.
Flai (YC) — AI that turns every call into a sale
Ari Polakof: Co-founder & CEO of Flai. Former founding engineer at HappyRobot (S23).
Juan Alzugaray: Co-founder & COO of Flai. Former data scientist at Netflix.
Alen Polakof: Founder & CTO of Flai. Former founding engineer at HappyRobot (S23); infrastructure engineer at Uber (4+ years).
Veritus Agent (YC) — AI agents for consumer lending
Joshua March: Founder & CEO of Veritus Agent. Serial founder with multiple exits; previously founded Conversocial (acquired by Verint) and co-founded a16z-backed SCiFi Foods. Author of Message Me.
David Schlesinger: Co-founder at Veritus Agent. 10+ years in startups, including Divvy Homes and Cloudflare.
Joseph Stein: Co-founder of Veritus Agent. 10+ years in startups; previously built a company acquired by Fitbit → Google.
Keystone (YC) — Your team’s on-call AI engineer
Pablo Hansen: Finished a master’s in AI at 19. First hire at Onyx (YC W24) for a year before founding Keystone.
OrangeSlice (YC) — AI agents that find buyers ready to convert
Vihaar Nandigala: Co-founder & CEO of OrangeSlice. Sold a previous company at 19; ex–J.P. Morgan.
Kishan Sripada: Co-founder & CTO of OrangeSlice. Previously bootstrapped FORMI to 100k+ users (used by Broadway, Disney, NBA teams). Former intern at Ramp; worked at Tour (YC S21).
Hera (YC) — AI motion designer
Peter Tribelhorn: Co-founder of Hera. Previously acquired and operated large YouTube channels totaling 30M+ subscribers (including Economics Explained and The Game Theorists).
Chia-Lun Wu: Founder of Hera. First engineer at Flagright (YC W22), building and scaling core infrastructure. Early engineer at Vyond, where he built Vyond Studio used by 20,000+ businesses. MS in Computer Science from National Taiwan University.
Floot (YC) — Vibecoding for non-coders building real apps
Yujian Yao: Former staff engineer at Retool, where he rebuilt the runtime powering every Retool app. Previously an early engineer at Asana. Gold medalist in Singapore’s National Olympiad in Informatics.
Edward Look: Bootstrapped a SaaS company to $5M+ ARR and 10k+ customers. Built low-latency distributed systems on bare metal. Former AWS infrastructure engineer; CS at UC Berkeley.
Slashy (YC) — The AI for work
Pranjali Awasthi: 18 y/o founder building the future of work. Never had a job. Built a high-school startup that went through HF0 (AI copilot for researchers).
Harsha Gaddipati: COO at Slashy, a personal agent that acts across Gmail, Notion, Slack, and more. CS graduate from Georgia Tech; former AWS. Paper published in Space Policy.
Dhruv Roongta: Co-founder of Slashy. CS background from Georgia Tech; internships at Groq, Pulse (S24), and Greptile (W24). Prize winner at CalHacks and HackMIT.
April (YC) — AI voice executive assistant
Neha Suresh: Co-founder & CEO of April. CMU alum. Former engineering lead at Zoho Books, building the accounting platform for Middle Eastern markets. Previously bootstrapped a tutoring startup.
Akash Thakur: Co-founder & CTO of April. Former Apple engineer.
Omnia — LLM search optimization
Daniel Espejo: Founder & CEO of Omnia. Former Klarna product leader, where he launched the Klarna Card and helped scale Klarna in Spain end-to-end (product, GTM, ops). Industrial engineer by training; repeat 0→1 builder focused on how companies grow.
ETN. — European new tech media
Ronan Chambers: Co-founder of etn. Building a fast-growing European tech media platform.
Luke Knight: Co-founder of etn. Previously in GTM and growth roles across startups and scaleups in Europe.
Hyperspell (YCF25) — AI writing and editing assistant
Manu Ebert: Former neuroscientist turned 4x founder (2 exits). 15+ years in ML and ~25 years coding. Built and led teams across data science, machine learning, and fintech; designed trading algorithms for a $63M AUM quant fund.
Conor Brennan-Burke: Second-time founder. Previously built software to $50M+ ARR and advised $1B+ in PE/M&A. Started college at 13; also built in music (soundtracks, band, AI music startup).
Imagine AI (YCF25) — Generative AI for creative content
Sky Yang: Co-founder & CEO of Imagine AI. Youngest-ever elected Student Body President at UC San Diego, leading 32,000 students and raising $150M for infrastructure. Founded Break the Outbreak, scaling to 400 people across 53 cities in 18 states.
Neo Lee: Co-founder & CTO of Imagine AI. UC Berkeley researcher in mathematics and AI, focused on LLM interpretability and non-cooperative stochastic games.
Selfin (YCF25) — AI companion for self-reflection
Paula Gutierrez: Aerospace engineering background at Imperial, MIT, and UC Berkeley. Former quant at Bank of America; experience at NASA and Rolls-Royce; software engineer at startups.
Joel Tomas Pimentel: Rocket science training at Imperial and Stanford. Now building a fintech company.
Karumi (YCF25) — Mobile developer tooling studio
Toni Lopez: Co-founder at Karumi (YC F25). Former founding engineer at StackAI; previously built and scaled AI systems and engineering teams across startups. Background in distributed systems, AI infrastructure, and 0→1 execution.
Pablo Omenaca Muro: Co-founder at Karumi (YC F25). Former founding engineer at StackAI; ex–BCG Gamma data scientist. MIT MBA (La Caixa Fellow). Background in data science, optimization, and large-scale digital transformation.
Sellraze (YCF25) — AI-powered sales enablement
Jeff Mao: Co-founder & CEO of SellRaze. Former sneaker reseller who built custom tooling in high school; reverse-engineered Amazon’s anti-bot systems during COVID and sold the tech. Now building tools for online buying and selling.
Tyler Ma: Co-founder & CTO of SellRaze. Dropped out of Georgia Tech to work on the company full-time. Engineering-focused founder.
Zalos (YCF25) — AI business intelligence platform
William Fairbairn: Former UK GM at Agicap, scaling the team 2× in 18 months. Previously in McKinsey’s Tech practice; built and sold a bootstrapped B2B SaaS.
Hung Hoang: Built Apple Pay products including Buy Now, Pay Later, serving millions daily. Worked on AI initiatives at Apple and startups including Twin.so and GitGuardian.
Fernstone (YCF25) — Cloud infrastructure automation
Luke Button: Co-founder of Fernstone. Previously led growth at Traba (Series A, Founders Fund), Head of Growth at Antimetal, and co-founder at Contrast (acquired).
James Chen: Co-founder of Fernstone. Built and led Traba’s Enterprise function to 25% of revenue. Ex-BCG; University of Chicago. Former #1-ranked U.S. debate champion.
Bryant Le: Co-founder of Fernstone. Former engineer at Google and SentiLink (Craft-backed Series A). Previously acquired and operated an Australian raffle business, doubling profits via new tech tooling.
GETASAP Asia (YC) — On-demand logistics platform in Asia
Raghav Arora: Founded GETASAP at 14. Building a Dashmart-for-B2B model—8-hour inventory delivery to retailers using retail data and micro-fulfillment centers.
Pranavkrishna Suresh: 19 y/o CS at Georgia Tech. Former AWS and GovTech founder. Co-building GETASAP, operating optimized fulfillment centers to deliver lowest prices with same-day (8-hour) delivery across Southeast Asia.
Peec AI (Series A) — AI research and insights tool
Marius Meiners: Founder of Peec AI. Built an AI-search analytics company to ~$5M ARR in ~11 months, serving 1,400+ customers. Former software developer and Top-100 League of Legends player.
Daniel Drabo: Co-founder & CRO of Peec AI. Took the company from $0 to ~$5M ARR in ~11 months by brute-force execution (2,000+ sales calls, two-person sales team). Background in early-stage VC and law; operator focused on GTM, distribution, and winning new categories fast.
Replit (Series E) — Collaborative cloud coding platform
Haya Odeh: Co-founder and VP of Design at Replit. Led product design, UX, and brand for a platform used by millions of developers worldwide.
Faris Masad: Early engineer and co-founder at Replit. Helped build and scale the core product in its early years.
Amjad Masad: Co-founder of Replit.
Momentic — AI user insight & feedback analysis
Wei-Wei Wu: Co-founder & CEO of Momentic. Formerly built infrastructure and real-time systems at Assembled; first employee and founding engineer at Nashi (acquired by Density in 2021).
Jeff An: Technical co-founder. Ex-engineer in cloud and test infrastructure at Splunk, Google, and Robinhood; University of Waterloo alum.
Openroll - AI-native operating system for compensation
Mattias Lindell: Mattias is the CEO and co-founder of Openroll, where he is building the AI-native operating system for compensation. Before Openroll, he competed in elite youth football, co-led multi-billion procurement projects at Swedbank, and launched multiple ventures - from a bedsheet-cleaning service to consumer apps - while studying. He left his MSc in Finance at Sweden’s top university, SSE, to work full-time on Openroll.
Porsev Aslan: Building the infrastructure for how the world gets paid
Agaton — AI Agents for Sales Teams:
Infinite Watch
Stealth (AI Code Generator)
Stealth (Voice AI Customer Support)
Stealth - AI Marketplace
Stealth - AI Video Generator
Stealth — AI Vectors (round led by a16z)
Stealth - led by Coinbase Ventures
These are some of the companies I backed in 2025. From YC Fall ’25 to Series A+ scale (even Replit’s Series E 😅).
Different stages, same ambition: replace broken workflows, not patch them.
Some are early, some already scaling fast, but all of them playing the long game.
I’m grateful to be backing such exceptional founders. Excited to see how they start to really break out and what 2026 may bring!
Big Thanks & Appreciation to the coolest people I met this year
Ryan K. Rigney, partner at A16Z Speedrun who helped me out so much, always with great attitude and openness!
Chris Tottman, founder of Notion Capital, greatest mentor I’ve had so far. Incredible to get support from someone so experienced and successful and still so kind!
Ruben Dominguez, founder of The VC Corner, coinvestor and fellow adventurer, probably one of the smartest persons I know
Ivan Landabaso, founder of Startup riders, coinvestor and BJJ goat. Always positive and dreaming big
Didac Lee, founder of Galdana Ventures, the real life James Bond.
Toni Lopez and Pablo Omenaca Muro from Karumi, both my age, born in 1993, whom I’ve felt we are sharing the same journey of going into the unknown and dreaming very big
John Kristensen from Agaton
Johan Nordenström who I had one of the best conversations I had in the whole year just the day we met. Such a cool human
Jorge Ferreiro from Smashsend & resident in Founders Inc (he got me to see Steven Bartlett first row)
2026 Crazy Aspirations
Nº 1 Aspiration: keep doing exactly what I do, but more and better. Change the world in 10 years
Working and coinvesting with Harry Stebbings 20VC
Coinvesting with Mark Cuban
Meeting & Collaborating with Steven Bartlett, Naval Ravikant, PBD, Andrew Schultz, Alex Hormozi
Closing long term partnership with Lovable Anton Osika
Interviewing Sam Altman, Marc Andreessen, Bryan Johnson, David Goggins, Jose Andrés (working on it)
Building product for founders & VCs of PMF (working on it)
First 4 full time hires (working on it)
Raising 10M for rolling fund (ongoing)
Build up my community of founders = one in person event (SF - working on it)
Find and invest in the 4 best companies of 2026
Being proud of myself (working on it)
Lets go 🚀