Tremendous

An angel investor's take on life and business

  • Moonshot Drops Kimi K2.6 – Cheap, Powerful, and Worth Trying for Builders

    Moonshot just dropped a new, more powerful Kimi: K2.6. I tested it this morning, and it’s the best open source model I’ve used yet.

    Moonshot claims that Kimi outperforms GPT-5.4 and Claude Opus 4.6 on some benchmarks. Better yet, Kimi is dirt cheap to run, making it a popular choice for developers. 

    I ran Kimi through 3 difficult tests this morning: researching startup funding rounds, prepping me for a meeting, and helping me study Japanese. Overall, Kimi turned in a strong performance, although it struggled with some web search tasks. 

    Let’s see what the new Kimi can do! 

    Round #1: Researching the Cursor Seed Round

    SpaceX recently agreed to acquire Cursor for $60 billion. Almost every seed investor in the world is slapping their forehead, wishing they’d been in on it… including me!

    I’m curious: what did the company look like at the time? Maybe studying such a successful company can help me find the next Cursor. 

    Cursor had already passed $1 million ARR by the seed round…unusually strong traction. 

    Cursor also had an exceptional founder. CEO Michael Truell had interned at Google in high school, which is highly unusual. 

    The deal wasn’t cheap — around $60 million. But especially in 2023, seed stage companies with over $1M in ARR were quite rare. 

    Cursor was a hot startup with serious traction and a great founder. It had every factor I look for.

    The price was steep. But that shows the importance of focusing on median entry price, rather than an absolute price limit. Expensive deals make sense occasionally, so long as they’re balanced out by cheaper ones. 

    Kimi did a great job, citing reliable sources and giving me the info I wanted. I’m giving this round an A! 

    Round #2: Stalking a Founder 

    Can Kimi can help me find the next Michael Truell? Next week, I’m meeting a fellow who might fit the bill…

    This young man has a very interesting startup in robotics. I asked Kimi to tell me about his background and any exceptional achievements.

    With a little help, Kimi pulled some interesting info (name redacted). 

    He managed to work in ROTC and run a startup at the same time. That’s impressive! I’m looking forward to meeting him. 

    Kimi found useful info, but only with a lot of coaxing. Kimi claimed it couldn’t access Linkedin, so I had to screenshot his LinkedIn and upload it. 

    I gave the same prompt to Grok, and it found a bunch of information on the founder right away.

    I’m giving Kimi a C for this round.

    Round #3: Kimi the Language Learning Coach

    It’s less than a week until I ship off to Tokyo! I’m desperately trying to brush up on my Japanese. Can Kimi help? 

    Kimi gave great advice. It suggested I listen to the Nihongo con Teppei podcast and repeat everything he’s saying.

    I pulled up a short video of Teppei’s and tried it. It was a wonderful way to get myself speaking and thinking in Japanese.

    The only thing I didn’t like about Kimi’s response was that it didn’t give me a link to the podcast. I still had to Google it.

    Still, Kimi was super helpful! I’ll give this round an A- 

    Wrap-Up

    I’m giving Kimi K2.6 a B- overall in my testing. It’s light years ahead of K2.5, which I gave a D in February.

    Kimi is good at analyzing information, but sometimes struggles in web searches. The hosted version of Kimi definitely isn’t the best model around.

    But for a cheap, open-source model to use in your application, Kimi is a great choice. 

    A lot of startups begin with a frontier model like Claude or Grok and get the output they want. Then, they use a cheaper model and try to get the same results at scale. 

    Kimi is an excellent choice for that cheaper model. If you’re building an AI, give the new Kimi K2.6 a shot! 

    More on tech: 

    I Put Elon’s New Grok 4.3 Beta to the Test — Here’s Why It’s My Favorite Model So Far

    NVIDIA Nemotron: Great for RAG, Weak on Web Search

    Kimi K2.5 Is Slow and Stupid

    Save Money on Stuff I Use:

    Fundrise

    This platform lets me diversify my real estate investments so I’m not too exposed to any one market. I’ve invested since 2018 with great returns.

    More on Fundrise in this post.

    If you decide to invest in Fundrise, you can use this link to get $100 in free bonus shares!

    Wispr Flow

    I used this app every single day to dictate to my computer, I’m even dictating this text using Wispr Flow! It’s way better than Apple’s native dictation.

    My productivity is up about 25% since I started dictating rather than typing. I’m also less tired and stressed.

    Get a free month of Wispr Flow Pro here!

  • I Put Elon’s New Grok 4.3 Beta to the Test — Here’s Why It’s My Favorite Model So Far

    Elon has quietly released Grok 4.3 Beta, xAI’s most powerful model yet. I tested it this morning, and it’s the best model I’ve ever used.

    I put 4.3 Beta through a grueling series of tests this morning. I had it analyze 5,000 venture deals, predict the future of the real estate market, even find under-the-radar travel recs for my trip to Tokyo.

    Elon’s new model excelled across the board. Let me show you what the new Grok can do…

    Round #1: Real Estate Investment Outlook

    High interest rates have meant a tough couple of years for real estate. That includes my Fundrise portfolio, which invests in housing across America. 

    What do the next couple of years have in store? Let’s ask Grok… 

    Grok predicts that Fundrise will see 8-15% returns over the next couple of years. This would be a big upshift from the low single-digit returns of late.

    Gradually declining interest rates should support real estate. Strong residential demand also plays into Fundrise’s strategy, which is heavy on multifamily. 

    Grok cited solid sources like CBRE and Reuters. Its analysis was clear and easy to understand. 

    Compared to the prior model, Grok 4.3 produced a more readable response. It’s shorter with better formatting, making it easier to digest. 

    I’m giving this round an A!

    Round #2: Planning My Trip to Tokyo

    A week from today, I’m getting on a plane to Tokyo! Lately, it’s all I can think about.

    It will be my first visit to Japan in six years. Ever since I first visited in 2014, Tokyo has felt like my home away from home.

    Can Grok recommend some awesome, under-the-radar spots for me to try? 

    I’ve actually been to around half these spots, including Todoroki and Hamarikyu! In each case, Japanese people brought me there.

    This tells me that Grok’s recommendations are superb. They’re the type of under-the-radar spots actual Tokyo residents recommend.

    The only flaw is a lack of dining options. Grok recommended some beautiful natural areas, but not much in the way of food.

    I’m giving this response an A-. 

    Round #3: Analyzing 5,000 Venture Deals

    Back to business…

    For the final round, I gave Grok 4.3 a difficult task: analyzing the nearly 5,000 deals I’ve passed on in the last couple of years. Can Grok find mistakes and opportunities to improve?

    I fed Grok a monster spreadsheet that includes details on every deal. Here’s what it found… 

    Grok thinks my approach is solid, carefully hewing to my thesis. I’ve systematically filtered out companies without builder founders, products or customers.

    My biggest opportunity to improve: tracking the close calls.

    Now and then, I find a company that almost meets my bar for investment. I haven’t systematically tracked the outcomes of those companies. That’s something I plan to do more of, assisted by AI. 

    Grok took its time, running analysis for around eight minutes. Still, that’s nothing for the amount of data I threw at it!

    Grok’s analysis was thorough and helpful. I’m giving this round an A+. 

    Wrap-Up

    Grok 4.3 Beta earns an A overall in my testing.

    It’s not a quantum leap from Grok 4.20. But its analysis is tighter, its formatting better, and its outputs more usable. 

    That makes Grok 4.30 Beta the best AI model I’ve used yet. If you’re looking for a frontier model, Grok 4.3 Beta is it. 

    For now, 4.3 Beta is only available to paid Supergrok subscribers. Plans start at $30 a month. It’s not available via API yet, but I expect that to be released soon.

    Everyone is excited about Anthropic right now. But xAI is building some incredible stuff in the shadows. 

    Don’t be surprised Elon if leapfrogs Dario soon…

    More on tech: 

    Grok Imagine Quality Mode: Jaw-Dropping Images, But Text Still Struggles

    Meta’s New Model is Worth $60 Billion — But Struggles in Real World Testing

    NVIDIA Nemotron: Great for RAG, Weak on Web Search

    Save Money on Stuff I Use:

    Wispr Flow

    I used this app every single day to dictate to my computer, I’m even dictating this text using Wispr Flow! It’s way better than Apple’s native dictation.

    My productivity is up about 25% since I started dictating rather than typing. I’m also less tired and stressed.

    Get a free month of Wispr Flow Pro here!

  • NVIDIA Nemotron: Great for RAG, Weak on Web Search

    NVIDIA sells $200 billion worth of chips every year. But Jensen isn’t stopping there. NVIDIA recently dropped its most powerful AI model ever. 

    This morning I tested Nemotron 3 Super, NVIDIA’s most sophisticated model. Nemotron is designed to let us build and run powerful agents cheaply.

    Nemotron does well if you give it data (also known as Retrieval Augmented Generation, or RAG). But it struggles with complex jobs and finding data on its own. 

    Let me show you where Nemotron can help and where you’re better off looking elsewhere…

    Round #1: Finding the Next Unicorn on Product Hunt

    Several times a week, I troll Product Hunt looking for the next great startup. Can Nemotron automate this? 

    Nemotron wouldn’t help, saying that it can’t browse the internet. For a model designed to create agents, this seems like a huge problem! 

    You can add a web browsing tool yourself. But it would be much easier if Nemotron included it automatically. 

    What’s more, Nemotron’s knowledge cutoff is July 2024, nearly two years ago. This makes it harder to build useful tools. 

    I love Jensen, but I have to give this round an F. 

    Round #2: Researching Drone Technology

    Earlier today, I chatted with a founder making very innovative drone motors. His motors don’t use any rare earth minerals.

    English major that I am, I don’t understand how existing drone motors use rare earths. Can Nemotron explain?

    Nemotron gave a great breakdown: the rare earths are used in a magnet that the motor needs to function.

    Nemotron’s explanation was a little technical. I asked a few follow-up questions, and it did a good job of breaking things down more simply.

    But there’s one big problem: Nemotron didn’t cite any sources. The information sounds plausible, but I can’t verify it.

    This makes it hard to rely on Nemotron. I’m giving this round a B-minus. 

    Round #3: Automating Intros to Startups

    Yesterday, I met a really cool biomanufacturing startup. Which investors in my network should I introduce them to?

    I fed Nemotron my spreadsheet of 75 top investors. Nemotron did a great job of looking through the list and finding investors interested in deep tech.

    When I give Nemotron all the data it needs, it parses it very well. You can take advantage of that in applications you build, RAG-ing Nemotron and getting outputs cheaply.

    I’m giving this round an A. 

    Wrap-Up

    I’m giving Nemotron a C- overall. 

    Nemotron really deserves two grades: an F for tasks requiring web search and an A for parsing data you provide. 

    Nemotron is way behind frontier models. But because it’s cheap to run, Nemotron could be a good choice for your application. 

    Just be sure to give it very specific instructions and lots of custom data. Then, you can get the output you need at lower cost. 

    I’m excited to see what Jensen does with Nemotron. Tenacious as he is, he just might score a come-from-behind victory.

    More on tech:

    Google Drops New Gemini macOS App — A for Easy Tasks, F for Complex Jobs

    Grok Imagine Quality Mode: Jaw-Dropping Images, But Text Still Struggles

    Meta’s New Model is Worth $60 Billion — But Struggles in Real World Testing

    Save Money on Stuff I Use:

    Fundrise

    This platform lets me diversify my real estate investments so I’m not too exposed to any one market. I’ve invested since 2018 with great returns.

    More on Fundrise in this post.

    If you decide to invest in Fundrise, you can use this link to get $100 in free bonus shares!

    Wispr Flow

    I used this app every single day to dictate to my computer, I’m even dictating this text using Wispr Flow! It’s way better than Apple’s native dictation.

    My productivity is up about 25% since I started dictating rather than typing. I’m also less tired and stressed.

    Get a free month of Wispr Flow Pro here!

  • Google Drops New Gemini macOS App — A for Easy Tasks, F for Complex Jobs

    Google just dropped a new Gemini app for macOS. It excels on simple tasks like summarizing documents. But for complex jobs, Gemini is surprisingly weak.

    This morning, I ran the new Gemini app through a three round test. Let me show you where this app is strong, where it disappoints, and why Google is falling behind…

    Round #1: Summarizing a Pitch Deck

    I look at thousands of pitch decks every year. Can Gemini summarize them and lighten the load? 

    This afternoon, I’m meeting with a fascinating startup working on biomanufacturing. Gemini did a great job of summarizing the deck (key details redacted). 

    Gemini pulled out everything I want to know, including what the company does, the founders’ backgrounds, and terms of the raise. I love how it formatted the response using bullets and bold text to make it more readable.

    I’m giving this round an A! 

    Round #2: Practicing Japanese For My Upcoming Trip

    Next week, I’m off to Japan! I couldn’t be more excited! It’s my first time since COVID.

    I’ve studied Japanese since 2014, but I’m getting a little rusty. Can Gemini help me practice? 

    I told Gemini to give me some common signs I’ll see in Japan, along with English translations.

    It gave me a useful list, including signs for Entrance and Exit at train stations and Open and Closed at stores. Unfortunately, the text is quite small and there’s no way to make it bigger.

    On mobile, I can actually speak Japanese with Gemini Live. That function isn’t in the MacOS app yet. 

    I’m giving this round a B-. Once Google adds Gemini Live to the macOS app, I’ll bump it up to an A.

    Round #3: Searching Gmail for Company Updates

    For our final round, I gave Gemini a challenge: running a very complex search in my Gmail.

    I want to see if any of my investments haven’t sent an update in the last nine months. But I don’t want to manually search each company.

    Gemini failed me completely on this task.

    I used the Pro model and let the search run for 10 minutes. Gemini didn’t produce any output. 

    Then, I tried it again on Fast mode. This time, Gemini gave me results. But they were completely incorrect! It flagged companies that just sent updates in the last few days.

    A year ago, this would’ve been too much to ask. But with the enormous investment in compute, I expect Gemini to be capable of more complex jobs by now. 

    I’m giving this round an F. 

    Wrap-Up

    I’m giving the new Gemini macOS app a C overall.

    It’s great for basic tasks like summarizing a PDF, but for anything more complicated, Gemini falls short.

    I can’t have a voice conversation with Gemini Live. The app can’t run complex searches on my email. 

    Gemini’s new app feels several months behind the cutting edge. It needs to work with multiple modalities and deeply integrate into my Google services. 

    Gemini has some great capabilities. But once again, Google is beginning to fall behind Anthropic and xAI. 

    Unless its next release is much more competitive, Google is in trouble.

    More on tech: 

    Grok Imagine Quality Mode: Jaw-Dropping Images, But Text Still Struggles

    Meta’s New Model is Worth $60 Billion — But Struggles in Real World Testing

    Gemma 4: Strong on Basic Tasks, Weak on Real Research

    Save Money on Stuff I Use:

    Fundrise

    This platform lets me diversify my real estate investments so I’m not too exposed to any one market. I’ve invested since 2018 with great returns.

    More on Fundrise in this post.

    If you decide to invest in Fundrise, you can use this link to get $100 in free bonus shares!

    Wispr Flow

    I used this app every single day to dictate to my computer, I’m even dictating this text using Wispr Flow! It’s way better than Apple’s native dictation.

    My productivity is up about 25% since I started dictating rather than typing. I’m also less tired and stressed.

    Get a free month of Wispr Flow Pro here!

  • Peter Thiel: America Is Winning — And Pulling Away from the World

    Headlines scream “American decline.” Peter Thiel isn’t buying it. 

    In a panel discussion in Japan this week, Thiel argues that on technology, demographics, and economic dynamism, the U.S. isn’t just holding on — it’s pulling away.

    “On a relative basis, it seems to me that the U.S. is winning against the whole world.” Peter Thiel

    America Makes, The World Takes

    “… the technological future is still being built in the US.” – Peter Thiel

    AI is producing much of the world’s economic growth. But AI tools are only built in a handful of places.

    The United States is by far and away the leader. Every frontier model was built here. Almost every major technology company is based here. 

    China is a strong competitor, producing capable models like Kimi and Qwen, and innovative robotics projects like Unitree. But outside the US and China, most countries are spectators in the AI race.

    What important technology is being built in Germany, France, or Russia? France’s only significant AI company, Mistral, is dwarfed by Anthropic and OpenAI.

    America’s Demographic Edge: More Babies, More Immigrants, More Growth

    The U.S. fertility rate has slipped below replacement levels. But compared to many of our rivals, we’re producing babies at an incredible clip.

    Our current fertility rate is 1.6, versus less than 1 for China. China is beginning to shrink rapidly even as our population still grows.

    And what’s true of China is true of most major countries: Russia, Germany, England, all are producing far fewer children than the United States.  

    We’re also able to grow our population through immigration.

    China has no history of immigration. Their language is not spoken worldwide like English is, which makes it hard for immigrants to integrate.

    Wrap-Up

    America is growing and creating new technology at a pace no other country can match.

    We have problems, to be sure. Our debt is heavy. Our politics are fraught.

    Nonetheless, America is pulling away from the rest of the pack. If we continue to work hard and keep our markets and people free, we can prosper like never before. 

    “… I am not convinced that the US is in decline relative to the rest of the world.” – Peter Thiel

    Have a great weekend, everyone!

    More on tech: 

    Zero to One

    Three Management Secrets from Jensen Huang

    Three Lessons on Business from Warren Buffett

    Save Money on Stuff I Use:

    Fundrise

    This platform lets me diversify my real estate investments so I’m not too exposed to any one market. I’ve invested since 2018 with great returns.

    More on Fundrise in this post.

    If you decide to invest in Fundrise, you can use this link to get $100 in free bonus shares!

    Wispr Flow

    I used this app every single day to dictate to my computer, I’m even dictating this text using Wispr Flow! It’s way better than Apple’s native dictation.

    My productivity is up about 25% since I started dictating rather than typing. I’m also less tired and stressed.

    Get a free month of Wispr Flow Pro here!

  • Why Almost Every Investor Passed on DeepMind — And How to Spot the Next One

    Demis Hassabis pitched countless investors when he started DeepMind. Almost everyone said no. Why did these VC’s miss out on the opportunity of a lifetime? 

    “Finding initial funding for this was very hard,”  Demis later recalled. 

    Demis and DeepMind are leading the race to AGI. They won the 2024 Nobel Prize in Chemistry for their protein folding technology, AlphaFold. 

    Passing on DeepMind is one of the biggest misses ever. But when Demis was raising his first capital in 2010, most investors gave him the cold shoulder.

    That mistake was much far reasonable than it seems…

    Why They Passed

    As someone who looks at a couple thousand startups a year, I can give you some great reasons to pass on DeepMind:

    1. No product
    2. No revenue
    3. No path to revenue
    4. Founder has no experience

    Demis told a great story about inventing machines that can think. But how would he get the money to keep this business alive for the years it would take to reach that goal?

    Would I Have Invested in DeepMind?

    Demis is very smart — that would be obvious to anyone. But given the way I judge investments, I would have passed on DeepMind too.

    I’d love to tell you that I would have seen the future back in 2010 and written a huge check! But for a pre-revenue company with no real path to market?

    Not likely. 

    How Investors Can Get These Decisions Right — The Wildcard

    Passing on DeepMind was a huge mistake. 

    But if we rip checks into pre-revenue companies with no business strategy all the time, we’ll go broke. We’ll no longer be able to make investments. 

    How can we change our rules to capture the occasional moonshot like DeepMind, while avoiding some of the failures? 

    What if we gave ourselves a wildcard?

    Maybe 80-90% of our investments have traction or at least a path to market. We could reserve 10 to 20% of our capital for moonshots like DeepMind.

    How Investors Could’ve Seen DeepMind’s Potential at the Time

    Every company comes down to one factor: the founder.

    In 2010, Demis didn’t have much business experience. But he was exceptional in one way: he was a chess prodigy. 

    As a child, Demis was the second-highest-rated chess player on earth.

    He spent much of his youth traveling all over Europe for chess tournaments. He beat grandmasters decades older.

    When Elon Musk hires someone, he looks for evidence they’re exceptional in some area — any area. Demis was clearly exceptional in chess.

    The One Question I’ll Be Asking Founders From Now On

    To make sure we don’t miss this crucial fact, we can ask founders, “What achievements are you most proud of?”

    Demis would have probably mentioned his past as a chess champion. We’d know that, despite his lack of track record in startups, he’s achieved something big. 

    Judging talent like Elon does helps us make far better decisions. And sure enough, Elon was one of the first investors in DeepMind!

    I’m going to start asking founders this question. I can’t miss the next Demis. 

    Wrap-Up

    It’s easy to sit here in 2026 and call the VCs that passed on DeepMind idiots.

    But none of these people are stupid. Their decision was wrong, but it made sense at the time.

    We need to learn from their mistake. We need to adjust our rules so we don’t miss the next DeepMind. 

    The key is finding founders who have done something exceptional. For the next Demis, it’s worth breaking the rules.

    Would you have invested in DeepMind?

    More on tech:

    How OpenAI and Anthropic Will Crush Harvey and Legora

    How Ron Conway Landed His Investment in Google

    Speaking Portuguese to an AI — Google Translate Practice

    Save Money on Stuff I Use:

    Fundrise

    This platform lets me diversify my real estate investments so I’m not too exposed to any one market. I’ve invested since 2018 with great returns.

    More on Fundrise in this post.

    If you decide to invest in Fundrise, you can use this link to get $100 in free bonus shares!

    Wispr Flow

    I used this app every single day to dictate to my computer, I’m even dictating this text using Wispr Flow! It’s way better than Apple’s native dictation.

    My productivity is up about 25% since I started dictating rather than typing. I’m also less tired and stressed.

    Get a free month of Wispr Flow Pro here!

  • Software Is Eating Law: Why AI-Native Firms Will Win Big While Most Lawyers Lose

    New law firms will be 95% AI. They will bill at low, flat rates. And they will crush existing firms. 

    Clients want lower, predictable bills. New AI-native law firms are in the best position to provide that. 

    This new world will create a handful of winners. But most lawyers, firms and law schools are in deep trouble. 

    How New, AI Native Law Firms Win

    Here’s what the new, AI native firms will look like…

    Instead of a team of ten associates and one partner working on a deal, picture one associate and one partner. The partner is handling the face-to-face relationship with the client. The associate is mostly running a bunch of Claude agents.

    These firms can put in low bids at flat rates because their costs are so much lower.

    AI native firms are already cropping up. Crosby recently raised $60 million from Index, Lux and Sequoia. 

    A handful of top partners at these new, AI-native firms will make way more money than any lawyer makes today. Think hundreds of millions or even billions of dollars. 

    Those top partners will be recognizable leaders in their field. They will build client trust.

    But most of the actual work will be done by computers. 

    Existing Firms Can’t Compete

    Big firms today do a little flat-rate work. Their clients demand it.

    But they can never fully switch away from the billable hour. 

    Old school firms have an enormous number of lawyers on the payroll. To feed those mouths, they need to bill lots of hours.

    AI native firms will win one client after another. Clients will find their low-cost, flat-rate bids irresistible.

    Slowly losing business, many old school firms will wither away. 

    Law Grads and Law Schools Left Behind

    AI doing most legal work means we need a lot fewer lawyers.

    Law grads will have a harder time finding work. As their unemployment rate rises, fewer people will go to law school.

    This will affect lower-tier law schools and their graduates first. Then, it will expand into mid-tier schools.

    Harvard and Stanford aren’t going anywhere. But everyone else will shrink or shut down. 

    Finite Work vs. Infinite Work

    If you make software engineers 10x more productive with AI, you actually might hire more engineers. 

    These supercoders can build you new features customers love. You sell more software and your business expands. 

    Software engineering is “infinite work.” There’s no limit to how much we might want. 

    But AI doesn’t boost lawyers in the same way. Legal services are “finite work” — we only want so much of it. 

    Eventually, you’ve negotiated all your customer contracts, handled any lawsuits, and you’re good to go. You just don’t need any more legal work. 

    When AI takes work away from human lawyers, most human lawyers lose. 

    Wrap-Up

    The biggest winner in all this is clients. 

    Clients will get lower, predictable bills. Their work will be done faster and more reliably. 

    A legal market transformed by AI is great for most of us. But for most lawyers, it’s a crisis.

    As legal work becomes software, law is headed to the same winner take all dynamics we see in tech.

    The winners will become fabulously rich. The rest will be left behind.

    More on tech: 

    How OpenAI and Anthropic Will Crush Harvey and Legora

    Request for Startups: American-Made Solar, Rare Earths and Next Gen Drones

    SaaS Is Cheaper Than Ever — But It’s Not Going Anywhere

    Save Money on Stuff I Use:

    Fundrise

    This platform lets me diversify my real estate investments so I’m not too exposed to any one market. I’ve invested since 2018 with great returns.

    More on Fundrise in this post.

    If you decide to invest in Fundrise, you can use this link to get $100 in free bonus shares!

    Wispr Flow

    I used this app every single day to dictate to my computer, I’m even dictating this text using Wispr Flow! It’s way better than Apple’s native dictation.

    My productivity is up about 25% since I started dictating rather than typing. I’m also less tired and stressed.

    Get a free month of Wispr Flow Pro here!

  • How OpenAI and Anthropic Will Crush Harvey and Legora

    “I just use ChatGPT, it’s really good.” I was chatting with a lawyer we’ll call Mary. She’s the assistant general counsel of a late stage fintech. Mary knows all about Harvey and Legora, but she’s not interested.

    ChatGPT already does everything she needs. 

    Lately, I’m hearing from more and more lawyers like Mary. They’re bypassing the hyped legaltechs and going straight to ChatGPT and Claude. 

    For VCs that poured a fortune into legaltech, this is a huge red flag. 

    A Torrent of Capital 

    Hundreds of millions of dollars have poured into legaltech in the last couple years. Harvey recently raised $200 million at an $11 billion valuation. That valuation is around 50x their current ARR. 

    Legora raised $550 million at a $5.55 billion valuation just days before. Legora is also valued at around 50x revenue.

    This is an incredible amount of money pouring in at extremely high valuations. 

    Meanwhile, public SaaS companies trade at 4x revenue. Harvey and Legora are growing much faster than those companies. But there’s still a long way between a 50x multiple and a 4x.

    To grow into these valuations, Harvey and Legora need to more than 10x their revenue. If growth slows, they may struggle to raise additional capital at these high valuations.

    Foundational Models Can Buy More Data and Compute 

    To handle legal tasks with AI, you need two inputs: data and compute. OpenAI and Anthropic can outcompete legaltechs on both.

    Companies like Harvey and Legora buy legal data and feed it to their models. But OpenAI and Anthropic have more money and can buy more data.

    They also have far more access to compute. OpenAI has locked up compute years in advance. 

    How do Harvey and Legora beat OpenAI and Anthropic when the labs have more access to the key inputs? 

    Legaltechs Can’t Win the Talent Wars

    There are some incredibly smart people at Harvey and Legora. But can they compete for talent with Anthropic? 

    A handful of top researchers are driving a lot of today’s progress in AI. Anthropic has so much excitement and capital behind it. It can attract talent in a way no legaltech can. 

    The best researchers want to work with other great researchers. This means OpenAI and Anthropic hoover up the top candidates. 

    ChatGPT and Claude Have the Distribution Edge

    ChatGPT and Claude have another advantage: distribution. 

    If you’re used to using ChatGPT in your everyday life, you’re going to use it at work too. To get lawyers to use a different tool, it has to be dramatically better than ChatGPT. 

    And it just isn’t. 

    Wrap-Up

    It’s no surprise that base models excel at legal work. 

    It’s perfect for an LLM. There’s tons of training data. All the inputs and outputs are text. 

    Harvey and Legora just can’t compete with OpenAI and Anthropic for capital, compute, talent, or data. Disadvantaged at every turn and selling an undifferentiated product, legaltechs will fall behind.

    I’m not an investor in any legaltech startups. But if I were, I’d be looking to unload a big piece of my position.

    The party may not last much longer.

    More on tech: 

    How One Founder Used Warm Intros to Raise $17 Million

    Request for Startups: American-Made Solar, Rare Earths and Next Gen Drones

    SaaS Is Cheaper Than Ever — But It’s Not Going Anywhere

    Save Money on Stuff I Use:

    Fundrise

    This platform lets me diversify my real estate investments so I’m not too exposed to any one market. I’ve invested since 2018 with great returns.

    More on Fundrise in this post.

    If you decide to invest in Fundrise, you can use this link to get $100 in free bonus shares!

    Wispr Flow

    I used this app every single day to dictate to my computer, I’m even dictating this text using Wispr Flow! It’s way better than Apple’s native dictation.

    My productivity is up about 25% since I started dictating rather than typing. I’m also less tired and stressed.

    Get a free month of Wispr Flow Pro here!

  • How Fake Investors Scam Founders: 5 Red Flags

    A venture studio is promising you a huge check. But there’s one catch: $20,000 a month in “service fees”.

    A founder on Reddit is facing this situation right now. “Service fees” are one of several scams that so-called “investors” run on startups.

    Startupland is the Wild West of capitalism. You can make a fortune, but you can also get ripped off.

    Here are five common scams to look out for:

    1. Service Fees. No investor should be charging you service fees.

      They’ll claim that they help with marketing, sales, etc. In truth, their help will be worth little or nothing.

      Investors are supposed to help their startups for free! Charging money shows they’re not real investors.
    2. “Sweat Equity”. Some “investors” will demand equity in exchange for services. These are the cousin of the service fees we covered in #1.

      These firms are usually dev shops disguised as investors. They’ll get you a little help from some offshore devs who probably aren’t competent. In exchange, they’ll take a fat slice of your company.

      Never make a deal like this. Investors invest actual cash, not “services”.
    3. Syndication Fees. A venture firm promises to syndicate your deal to their network of wealthy investors. You’re sure to bring in millions!

      All they need is a small syndication fee.

      Run like hell. Syndication fees are one of the most blatant scams in startups.

      I’ve invested in numerous syndicate deals. The startup never pays a fee. We investors pay all the costs of setting up the legal structure.
    4. Ridiculous Equity Grants. The venture studio the Reddit founder is joining wants cash, but many other venture studios take a giant equity slice instead.

      It’s common for a venture studio to take 40% of your company. This breaks your cap table and kills future financings.

      Some studios make ridiculous claims that they’re a “co-founder.”

      No venture firm is a co-founder of your company. And if YC only gets 7%, why should some no-name venture studio get 40%?
    5. Due Diligence Fees. Every investor does due diligence before they invest in a startup. Only scammers try to charge for it.

      Never pay a due diligence fee. The VC should be bearing those costs.

      Not only should you not pay a due diligence fee, you shouldn’t accept capital from any firm that tries to charge you one. Dealing with them is going to be a nightmare.

      Wrap-Up

    Whenever I hear of a so-called “investor” trying to scam a founder, it makes me extremely angry.

    We’re supposed to be helping founders, not stealing money from them.

    Inexperienced founders fall for these scams all the time. Scammers target founders with less access to capital: first-timers, folks outside the U.S., women, and minorities. 

    Real investors don’t take money from you. We give you money.

    Stay miles away from any “investor” running any of these scams.

    You don’t need these jerks! There are real investors out there who want to help you.

    It may take time, but you will find them.

    More on tech: 

    Five Things Founders Should Never Pay For

    The Venture Studio Trap

    Which Accelerators Are Worth Your Time? Here’s How to Find Out.

    Save Money on Stuff I Use:

    Fundrise

    This platform lets me diversify my real estate investments so I’m not too exposed to any one market. I’ve invested since 2018 with great returns.

    More on Fundrise in this post.

    If you decide to invest in Fundrise, you can use this link to get $100 in free bonus shares!

    Wispr Flow

    I used this app every single day to dictate to my computer, I’m even dictating this text using Wispr Flow! It’s way better than Apple’s native dictation.

    My productivity is up about 25% since I started dictating rather than typing. I’m also less tired and stressed.

    Get a free month of Wispr Flow Pro here!

  • How One Founder Used Warm Intros to Raise $17 Million 

    “Mike” has raised over $17 million. The way he gets investor intros is totally different from most founders. Here’s how it works…

    A Radically Different Approach

    On Tuesday, I was chatting with Mike (not his real name). I’ve invested in his startup twice, and he’s gearing up to start a new fundraise. 

    Most founders ask, “Do you know any other investors you could introduce me to?” This leaves the ball in my court. 

    Mike inverts this. He does everything for me. 

    Mike had me send over a list of investors I know. Then, he picked out the VC’s he thought were a good match.

    Mike is drafting custom messages to each person. I just have to cut and paste!

    Why Mike’s Method Wins

    Mike has taken all of the guesswork out.

    I don’t have to pick investors for him. I don’t have to write intros. 

    Make it easier for an investor to make an intro, and it’s more likely to happen!

    Mike uses every person on his cap table to meet more investors. He knows that people with skin in the game are a critical resource.

    Think about it from a VC’s perspective. Which message are you more likely to reply to?

    1. Intro request from someone you already know who has invested in the startup
    2. Random cold message from a founder you’ve never heard of.

    What If I Haven’t Raised $17 Million — Or Anything?

    “That’s great for Mike, but he’s already raised 17 million! What about the rest of us?”

    Even if you’ve only raised a small sum, you can use those investors to meet more investors. As your network expands, so will the amount of money you raise.

    Even if you haven’t raised a cent, you can still get warm intros!

    Ask your founder friends for intros to their investors. Ask your customers if they know any VCs. 

    How to Draft a Message for Intros

    For every intro, make sure you have a pre-written message, ready to cut and paste. Keep it short. Here’s an example:

    “I’m building UberCab, the easiest way to get a ride. 

    Just request a car on your phone and a driver shows up in minutes. You can see the car on a map, so you never have to wonder where your ride is.

    We have 3 cars in operation and have done over 100 rides so far.

    We are now raising a $1 million pre-seed to scale throughout SF.

    Got 15 minutes?”

    Wrap-Up

    Getting warm intros is all about making it easy for the other person.

    If you make them find the right people and draft messages, it’s not going to happen. If you do everything yourself, you’ll get way more intros.

    Those warm intros are the key to closing your round.

    Just remember: when you make intros easy, you get more of them! 

    Have a great weekend, everybody!

    More on tech:

    Your Deck Probably Sucks. Here’s How to Fix It.

    How to Tell If Investors Are Really Interested

    Request for Startups: American-Made Solar, Rare Earths and Next Gen Drones

    Save Money on Stuff I Use:

    Fundrise

    This platform lets me diversify my real estate investments so I’m not too exposed to any one market. I’ve invested since 2018 with great returns.

    More on Fundrise in this post.

    If you decide to invest in Fundrise, you can use this link to get $100 in free bonus shares!

    Wispr Flow

    I used this app every single day to dictate to my computer, I’m even dictating this text using Wispr Flow! It’s way better than Apple’s native dictation.

    My productivity is up about 25% since I started dictating rather than typing. I’m also less tired and stressed.

    Get a free month of Wispr Flow Pro here!