About
In-person work is now a choice. Make it a good one.
We're building a new way to…
We're building a new way to…
Activity
16K followers
-
Rafi Sands shared thisW26 just broke every Y Combinator record on the books (freakishly strong). 14 companies at $1M+ ARR before raising. One at $27M. Average weekly growth of 14% across nearly 200 startups... the fastest in YC history! We have 100+ YC teams working out of our crashpad office in SF while they look for space of their own. So I notice when something's off, or when something's actually working. Here are the ones that stood out to me: → Pocket (YC W26): AI Personal Assistant (Gabriel, Akshay) → Zatanna: APIs for AI agents (Tarun, Rithvik, and Alex) → OpenSpec (YC W26): 27k+ stars, spec framework for coding agents (Tabish) → Patientdesk.ai (YC W26): AI OS for dental clinics (Oncel, Emre, Fikri) → Alt-X: AI agents for real estate underwriting (Ryan, Michael) → Remix: Auto-generate social content (Sam) → Haladir (YC W26): Building Operational Superintelligence (Jibran, Quan, Preston, Joseph) → Human Archive (YC W26): World's largest multimodal robotics dataset (Raj, Rushil, Samay, Shloke) → HLabs: US-made robot actuators (Paul Hetherington) → ZeroSettle (YC W26): Skip App Store fees, direct billing (Gabe, Ryan) → Avoice (YC W26): Harvey for Architecture (Chawin, Chawit) → Asimov (YC W26): On-demand human intelligence for robots (Anshul, Lyem) The whole distribution shifted, not just the top few. That's the part that's hard to explain and easy to feel if you're in it.
-
Rafi Sands shared thisadaption just raised $50M at seed to kill monolithic AI. The bet is laid out in a paper Sara Hooker published in December. It said bigger models aren't winning anymore. It said a model with 4.5% of the parameters can beat one 20x its size. It said the relationship between compute and performance is "highly uncertain and rapidly changing." She called it "the slow death of scaling." Then raised $50M to prove the alternative: adaptability over scale. The team she's built to do that: → Sara: Co-Founder & CEO. Studied economics, built a nonprofit to get into AI, research scientist at Google Brain, led Cohere. Named to TIME100 AI. → Sudip: Co-Founder & CTO. Seven years at Google Brain, built Cohere's entire inference infrastructure, and left to rebuild it from scratch. → John: AI Researcher, Founding Team. Ex-Cohere Labs. Left a good job at a frontier lab because he believed in a thesis more than a salary. → Isabelle: Founding Marketer. Ex-Cohere. Joined a 6-person team at seed to tell the world that scaling is dead. → Ben: GTM Lead Scaled B2B SaaS companies, most recently orchestrated the commercial strategy behind Snigel's PE exit. Now building Adaption's go-to-market from scratch. → Jack: Founding Designer Pentagram (Paula Scher's team), Magic Leap, Vannevar Labs. Now building the brand at Adaption. One to watch.
-
Rafi Sands shared thisI got to tour 3 office buildings in Boston/NYC last week. The amenities made me question everything I thought I knew about where people work best 👇 • F1 simulator • Sauna • Mother's care suite • Massage therapy • Golf simulator A lot of buildings are transforming their spaces into "social clubs". If you've ever been to a Centurion Lounge or a Sapphire Reserve Lounge, that's the vibe. Pretty amazing the type of stuff they're putting in there! This backs something I've believed for a long time. Physical space changes: 1. How we show up as people 2. How we create 3. How we connect with our colleagues The best founders I work with prioritize space early. It pays dividends for team energy, culture, and attracting amazing talent. Office space should be a tool. The market is finally catching up 🚲
-
Rafi Sands shared this4 years ago, I hiked 2,500 metres and accidentally started a company. Here's what a mountain taught me that every office I'd ever been in couldn't... After climbing to the peak of Platteklip Gorge, I stopped, caught my breath, had a beer, and sat down before the descent. Open sky, no walls, nothing between me and the horizon. I had more clarity about what I wanted to build than I'd had in months... When I got back to my Airbnb, I looked it up. Turns out there's a name for it: the "cathedral effect". The idea that vast, open spaces with high ceilings change how we think. Make the big questions feel answerable. That exact unlock is the kind of thinking that a low ceiling and those fluorescent white lights just... can't. The more you think about it, great prayer spaces across history share the same features. Cavernous. Open. Built to make the act easier. Sitting on that mountain, I wasn't exactly in a cathedral. But I felt it. That was the moment Tandem made sense to me. We spend so much energy optimizing what happens inside our offices. Almost none asking whether the space itself is working for us or against us. Why do we expect the most creative ideas to happen in the places least designed for them? Who wants to join me on my next hike with Henry?? ps. shoutout to Andrew Grabowski for being there every step of the way
-
-
Rafi Sands shared thisThe YC "rejection club" has built more unicorns than most VC portfolios. As someone who faced the YC "no" two times, I thought it would be cool to spotlight those who have seen success after the "fails". 1. Steli Efti (Close): 𝗔𝗽𝗽𝗹𝗶𝗲𝗱 𝟳 𝘁𝗶𝗺𝗲𝘀 before getting into YC. Now running a $50M+ ARR CRM serving thousands of small businesses. The persistence through 6 rejections taught him more than any single "yes" could have. 2. Amjad Masad (Replit): 𝗔𝗽𝗽𝗹𝗶𝗲𝗱 𝟯 𝘁𝗶𝗺𝗲𝘀, rejected without interviews. Got recruited by Sam Altman via Twitter DM after building traction. Now 30M+ users, $220M+ raised, changing how people code. 3. Drew Houston ( Dropbox ): 𝗥𝗲𝗷𝗲𝗰𝘁𝗲𝗱 𝗶𝗻 𝗽𝗲𝗿𝘀𝗼𝗻, literally shooed out of YC offices. Created a demo video that hit #1 on Hacker News, got accepted, and built a $9B company with 700M+ users. 4. Sabba Keynejad (VEED): 𝗥𝗲𝗷𝗲𝗰𝘁𝗲𝗱 𝟮 𝘁𝗶𝗺𝗲𝘀. Rejected by YC. Sold his crypto to survive. Bootstrapped a video editing tool to $40M+ ARR without ever getting in. YC said "no"... he built it anyway. 5. Dave Rogenmoser (Proof): 𝗔𝗽𝗽𝗹𝗶𝗲𝗱 𝟮 𝘁𝗶𝗺𝗲𝘀. First rejection with $5K MRR and 200 customers. Spent 6 months grinding to 3,000 customers and $75K MRR, got accepted to W18 batch. Eventually raised $2.2M+ and built the personalized social proof category. 6. Sangyong Jung and Arthur Kim (Relate): 𝗔𝗽𝗽𝗹𝗶𝗲𝗱 𝟲 𝘁𝗶𝗺𝗲𝘀 over 3 years, pivoted 5 times across different ideas. Finally accepted to S22 batch with a sales CRM that now serves thousands of teams. The iteration through rejections shaped their final product. But the most interesting pattern by far is this: The founders who got rejected multiple times before YC have higher retention rates and raise larger follow-on rounds than first-try acceptances. Rejection isn't a sign to stop, it just mean your time is coming. Anyone else applied multiple times before being accepted into YC? Share your story below 👇
-
Rafi Sands shared thisPaul Graham told our YC S24 batch something that kills the Sunday scaries: "Look to your left. Look to your right. One of you will succeed. The other two will fail. Not because you quit, but because you'll spend years chasing something mediocre with just enough traction to keep you going" Now, every Sunday night, it's a must for me to grab a pen and jot down two things (maybe three). Then I run them through the Paul Graham test: "Is this truly worth my attention or just noise?" For anyone who didn't ask, my Monday usually goes something like this: I get to my desk, and I don't have to think about my to-do list. I just have to hold my ground and say "no" to any crap that distracts me from my Sunday night high leverage tasks. For me, the biggest lesson I've learned while building a business is staying focused on the mission and not getting distracted by all the 'shiny things'. I quickly came to understand that focusing on the wrong tasks for too long can destroy companies 🚲 P.S. Who likes my writing buddy?
-
Rafi Sands shared thisGranola has raised $63M and is coming to San Francisco! They entered one of tech's most crowded niches, spent a year onboarding users one by one, trashed half their features... then the VCs came running. Turns out, doing less was the whole strategy. Here some of the insane numbers: • $20M Series A led by Spark Capital (raised in ~1 week post-launch) • 5-6x weekly user growth since May 2024 launch • solid 70% of users are coming back after their first week • Team scaled from ~5 to 40+ people in under a year • VCs at Benchmark and Sequoia use it internally The founders aren't first-timers. Christopher built Socratic (acquired by Google) to 10M users, won App of the Year, and sold to the beast, Google, in 2018. Sam spent his career as the first designer through the door at startups that needed taste before they had scale. With that impressive growth and enterprise momentum (hello Vercel, Ramp, and Roblox), they're heading straight to the holy grail (aka San Francisco). The team behind their insane growth also needs to be highlighted: They've got Jo, aka chief of staff writing prompts at granola (LLM Whisperer), which tells you everything about how deep they are in the prompting weeds. Mathew, as Chief Architect and Tikhon, as Founding Engineer, built the core infrastructure that handles millions of meetings without breaking a sweat. Michael runs ops as COO, while Rob heads marketing, and Luke keeps the design engineering tight. The team is a small, tight-knit crew with massive ambitions and actual craft. P.S. Who else has spotted these Granola billboards in SF? 🐶 🚲
-
Rafi Sands shared thisOffice space rental is a f**king weird industry. I was having dinner with a few founders the other night and told them about my wildest office stories. Here are the best ones: → One of the companies that rents from Tandem is on a mission to build an 𝗔𝗜 𝘁𝗼𝗶𝗹𝗲𝘁 (not kidding). These guys have been killing it for years. Initially, they mentioned they had "special requirements" for their office space... we had no idea what we were stepping into. It turns out these AI toilets are for senior centers to help identify early health issues. They’re definitely one of the coolest companies we’ve worked with. → Another company needed a space that could fit a 𝗳𝗶𝘀𝗵 𝘁𝗮𝗻𝗸 𝘁𝗵𝗲 𝘀𝗶𝘇𝗲 𝗼𝗳 𝗮 𝗯𝗮𝗰𝗸𝘆𝗮𝗿𝗱! In the office. At first, when they said they were "a little niche", we thought weird hours, extra storage, or something like that. Turns out they built underwater cameras for aquaculture and needed somewhere to test them... we figured it out in the end 😅 → "𝗪𝗲 𝗿𝗲𝗾𝘂𝗶𝗿𝗲 𝘀𝗽𝗮𝗰𝗲 𝗳𝗼𝗿 𝟭𝟱 𝗵𝘂𝗺𝗮𝗻𝘀 𝗮𝗻𝗱 𝟭𝟱 𝗵𝘂𝗺𝗮𝗻𝗼𝗶𝗱𝘀." I had to read that twice. It turns out that half of their "team" consisted of humanoid robots they were currently developing... & they needed a place where the robots could move around without, and I quote, "startling the neighbors." Got a wacky startup? We would love to hear it 🚲 P.S I am here writing this post in a cupboard of GPUs... maybe I fit in with the 'niche' work set ups...
Rafi Sands replied to a comment
1w
Mirit Cohen (she/her) agree, so where does F1 simulator fit into that?
-
Rafi Sands liked thisRafi Sands liked thisI went to 3 very different conferences back-to-back - NVIDIA GTC, CERAWeek, and RSAC - and had a consistent takeaway: 𝗪𝗲’𝗿𝗲 𝗻𝗼𝘁 𝗯𝗼𝘁𝘁𝗹𝗲𝗻𝗲𝗰𝗸𝗲𝗱 𝗯𝘆 𝗶𝗱𝗲𝗮𝘀 𝗼𝗿 𝗱𝗲𝗺𝗮𝗻𝗱. 𝗪𝗲’𝗿𝗲 𝗯𝗼𝘁𝘁𝗹𝗲𝗻𝗲𝗰𝗸𝗲𝗱 𝗯𝘆 𝘀𝘆𝘀𝘁𝗲𝗺𝘀. 1) Demand is not the problem - systems are the bottleneck. 🤖 At GTC, the AI demand story is constrained by infrastructure on the ground. Data center racks sit partially empty because there isn’t enough power to run both compute and cooling. Entire conversations are around cooling as a near-term limiter, not a nice-to-have. ⚡️At CERAWeek, the same dynamic showed up differently: not a lack of technologies but of bankable solutions. The conversations weren’t “does this work?” but “who takes performance risk?”, “who underwrites warranties?”, and “how do we finance early commercial projects without over-relying on equity?” 🔐 At RSAC, in cyber, the ceiling isn’t detection tech but what the systems can tolerate. OT environments are so fragile that most solutions stop at passive monitoring. The tech to do more exists, but the systems don’t allow it. 2) The hardest problems are coordination problems, not technical ones. 🔐 At RSAC, one of the clearest pain points was - who owns the problem? Cyber responsibility in distributed energy systems is fragmented across owners, EPCs, OEMs, and operators. No clear owner = no clear budget = slow adoption. ⚡️ At CERAWeek, scaling climate tech requires alignment across corporates, lenders, insurers, and customers. Financing structures (e.g. offtake agreements, balance sheets, risk-sharing structures) matter as much as the underlying tech. 🤖 Even at GTC, scaling AI infra depends on coordination across utilities, equipment suppliers, and developers - entire supply chains and the grid need to be aligned. 3) What’s not being talked about is also telling - each conference had incredibly smart people having 3 adjacent (but siloed) conversations about the same underlying systems. 🤖 At GTC, surprisingly little discussion of real-world constraints (power, geopolitics, supply chains) that will ultimately shape AI deployment. ⚡️ At CERAWeek, huge uncertainty (aka limited commentary) around geopolitical volatility (e.g., Iran) despite obvious energy system implications. 🔐 At RSAC, a heavy IT focus, but the hardest problems sit at the intersection of cyber + physical infrastructure. We’re entering a phase where the question isn’t whether to build - 𝗶𝘁’𝘀 𝗵𝗼𝘄 𝘁𝗼 𝗺𝗮𝗸𝗲 𝗰𝗼𝗺𝗽𝗹𝗲𝘅, 𝗺𝘂𝗹𝘁𝗶-𝘀𝘁𝗮𝗸𝗲𝗵𝗼𝗹𝗱𝗲𝗿 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 𝘄𝗼𝗿𝗸 𝗮𝘁 𝘀𝗰𝗮𝗹𝗲. Curious what signals others saw! Great seeing familiar faces in person - Daniel Goldman Elizabeth Pacelle Ernest J. Moniz Girish Nadkarni Hugh Le Mussadiq Akram Kevin Heard Sila Kiliccote Ayo Olabimtan, CFA Gabriel Scheer Sean Kane James Diaz-Sokoloff Kevin Eggers Tibor Toth David Medoff Peter Sopher Daniel Stack Supratim Das Ignacio Bincaz Brent Lanyon Mohammad A. Alkhadra Chris Graves Natasha Kostenuk, P.Eng
-
Rafi Sands liked thisRafi Sands liked thisThrilled to share that Glimpse has raised a $35M Series A led by Andreessen Horowitz (a16z)! Beyond grateful to be part of such a talented, driven, and thoughtful team - one that moves with speed, precision, and a level of care I haven't experienced anywhere else. Nine months ago I joined Glimpse knowing it would be a special place to work, but I didn't think I'd have this much fun in the process. This raise is a testament to what we've built together, and it's just the beginning. We're growing the team, and I couldn't be more excited about the talent and energy we're about to bring in to kick off this next stage. To my current teammates: thank you for making this place what it is. To anyone thinking about joining - now is an incredible time to be part of Glimpse. Onward.
-
Rafi Sands liked thisRafi Sands liked thisI've been watching my autonomous company build itself for the past few weeks. Mind blown. A few weeks ago, I started working on Paperclip, an open source project that allows you to build autonomous companies that run themselves. Here's what I discovered: The Good: Multiple AI agents (content writers, SEO specialists, developers) actually coordinate well out of the box. You don't even need to know what an agent is. You set the vision as a "board member" and watch agents execute. Like you say "go do blah blah blah" And they all do that. The orchestration was smoother than expected. And was surprisingly easy to get started. The Reality: "Autonomous" still requires daily management. I found my agents would get stuck, need direction, and coordination broke down regularly. The SEO specialist, for example, was very bad at SEO when it first got started. It improved when I asked why nothing was being indexed in Google. Miraculously, it got better without any training. So, it's dramatically less work than manual company building, and you don't have to worry about company culture or morale. (I'm sure morale is bad because they work 24-7). But it's definitely not a fully hands-off experience. But, building isn't the hard part - distribution is. We've always known this but with my autonomous company, that's even more apparent. It has built A LOT. But there is no one coming to the website. And SEO only gets you so far. There's no sales agent to help either. I'm sure if I tinkered with it, I could get it to build an outbound sales process. But, I'm just the VC and Board member – not in the day-to-day here. While not ready for prime time today, Paperclip shows where we're headed. The technology is advancing faster than I imagined.
-
Rafi Sands liked thisRafi Sands liked thisSpaceX launched 119 payloads on Transporter-16 this morning. Multiple spacecraft on that mission are flying parts sourced through Vendra (YC S24). Incredible to be a small part of today's space race. 🚀 🛰️
-
Rafi Sands liked thisRafi Sands liked thisbaited 150+ people to come to our event… turns out u can bait lots of people to come for events in SF if you promise chicken rice (not pizza) tummies were filled and learnt lots of new things from our SEA founders building in healthcare, relocation, AI, simulations, HR… Chinmay Chauhan, Felix Lee, Aravind Kandiah, Justin Gorriceta-Banusing, Jacob Chia, Rachael Annabelle Yong, Kenneth Lou and Ryan Foo thanks Daryl Png, Vincent Song from Singapore Global Network (SGN), Daniel Suh, Ted Wint from Gold House, Ara K. from Sky9 Capital and Winston Gu, James Huie from Paul Hastings for helping to bring this to life again!
-
Rafi Sands liked thisRafi Sands liked thisI enjoyed speaking with CNBC about the various factors that could be driving up consumer energy prices. https://bit.ly/4v28ZTU
-
Rafi Sands liked thisRafi Sands liked thisLast week was one of the best weeks we've had since starting Bilanc. We signed two new customers. Expanded three existing contracts. Met a customer in person. And kicked off onboarding with one of the biggest enterprises we've been talking to for months. Weeks like this don't happen often. When you're a startup, most weeks are grinding. Cold outreach that goes nowhere. Features that break. Calls that don't convert. But then you get a week where everything clicks and you remember why you started. I just want to say thank you to Lambert Le Manh and Dickson Dokowe. These two are the reason any of this is possible. They build, they ship, they stay up late debugging, they show up every single day. Building a company isn't about one big moment. It's about hundreds of small ones that compound. And this week, a lot of them compounded at once. Back to work.
Experience & Education
-
Tandem
View Rafi’s full experience
See their title, tenure and more.
or
Languages
-
English
Native or bilingual proficiency
-
Spanish
Limited working proficiency
Organizations
-
Undergraduate Business Society (UBS)
Consulting Workshop Participant
Other similar profiles
Explore more posts
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content