Tremendous

An angel investor's take on life and business

  • My $698,000 Miss on Figure and Anthropic

    I almost made $698,000 — almost. This is how I missed two of the biggest opportunities of my career…

    Three years ago, I had the opportunity to be an early investor in two incredible startups: Figure and Anthropic. 

    I wound up passing on both. 

    How to Lose $698,000

    Today, Figure is valued at $39 billion. Anthropic is valued at a whopping $380 billion. 

    Both are about 100x higher than when I passed.

    At the time, I was writing $5,000 checks (I’ve since doubled that).

    Both these opportunities came via syndicates, so I would have lost 20% of any gains. I also would’ve been diluted as they raised more money.

    Still, I’d be sitting on paper profits of approximately $698,000 if I’d invested in both. Had I done follow-on, I’d have made even more. 

    Why I Passed 

    How did I miss these amazing investments?

    Both had valuations far beyond my normal range. My usual entry price is between $10 million and $15 million. Figure was raising at several hundred million. Anthropic was raising at about $4 billion.

    Also, both shared very few financials. That’s not uncommon when you’re a small check in a huge round. 

    Between high entry prices and lack of financials, I decided to pass.

    The Importance of Exceptional Founders

    What I got wrong was not placing enough weight on the founders’ backgrounds.

    Figure founder Brett Adcock previously founded Archer Aviation. Archer is an eVTOL company worth $5 billion on the public markets.

    Some people criticize Brett because Archer hasn’t shipped a product yet. But the company is worth billions, completely liquid. Financially, it’s a success. 

    Dario from Anthropic also had an incredible background. He was Vice President of Research at OpenAI — one of the top executives. 

    Brett and Dario’s backgrounds alone should have been reason enough to write the checks. 

    Startups Can Get a Lot Bigger Than You Think

    For an investment to make sense for me, it must be able to return at least 300x. I was concerned there wasn’t enough upside with Figure and Anthropic.

    But if humanoid robotics works, creating a hundred billion dollar company shouldn’t be a problem. And if you build a highly successful new base model, a trillion dollar company isn’t out of reach.

    I underestimated how big these startups could get. Today, they’re well on their way to those lofty valuations. 

    What I’ll Do Differently Next Time

    The next time I see a truly exceptional founder, I’m going to be more flexible on valuation and traction.

    Late last year, I paid a higher-than-average price to invest in Cortex AI, which feeds factory data to AI models. The founder, Lucas Ngoo, previously co-founded Carousell, a billion dollar startup backed by Sequoia.

    I’ll continue to bend the rules occasionally for the best entrepreneurs. This could help me nail the next Figure or Anthropic. 

    Wrap-Up

    Nobody is in every good deal. Not Sequoia, not Benchmark, nobody. 

    When we miss great investments, we don’t need to feel bad. But we should take the opportunity to learn.

    In the end, everything worked out. 

    Just a couple of months after I passed on Figure and Anthropic, I invested in a little startup called Micro1. In the last year, they went from $4 million to $200 million ARR. 

    That’s the beauty of venture capital. You don’t have to be right all the time. 

    Just once will do.

    More on tech: 

    Three Tips for a Smooth Fundraise That Brings in Millions

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

    Trading at 0.2 Times Sales — The Most Hated Companies in SaaS

    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!

    Misfits Market

    I’ve used Misfits for years, and it never disappoints! Every fruit and vegetable is organic, super fresh, and packed with flavor!

    I wrote a detailed review of Misfits here.

    Use this link to sign up and you’ll save $15 on your first order. 

  • Automating My Investing with Lovable

    Investing involves a ton of repetitive tasks. What if I could vibecode a tool to automate them? This morning, I built an app on Lovable to research founders. 

    Here’s how it works… 

    Building Founder Intel

    I started with a simple prompt…

    In minutes, Lovable created a nice-looking UX. Founder Intel is born!

    This makes you wonder about the future of Figma. What’s the point of a mock-up when you can pop up a real product so quickly?

    Lovable asked me if I wanted to connect the application to an LLM so it can find real data.

    Weird question, right? Why wouldn’t I want real information?

    Vibecoding still struggles to understand user intent. 

    I told Lovable to hook me up. Lovable connected to Gemini, and I was ready to test my tool for the first time. 

    Taking My New App for a Test Drive

    I used Founder Intel to look up Christina Cacioppo, the founder of the awesome SOC 2 platform, Vanta. 

    It came up with good, accurate information about Christina, providing a clear explanation of what Vanta does. It also included the Series C she raised from Sequoia. 

    But the info is a little dated. It didn’t mention the Series D, raised last July.

    All in all, although Founder Intel is a cool little product, a good LLM would have done a better job. If I put Christina’s name into Grok 4.20 Beta, I get up-to-the-minute info with elaborate sourcing.

    Wrap-Up

    Even if the output wasn’t perfect, making Founder Intel was fun! If you have folks on your team that are struggling with AI, a vibe-coded tool could be a great way for them to interact with it. 

    This little app is as basic as it gets. I’m looking forward to creating more complex automations that can make my work easier.

    What are you guys vibecoding these days? 

    More on tech: 

    A SOTA Model On Your Phone? — Testing Qwen 3.5

    Gemini 3.1 Pro Disappoints with Slow Outputs, Weak Sourcing

    SaaS Isn’t Dead — Time to Buy the Dip?

    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!

    Misfits Market

    I’ve used Misfits for years, and it never disappoints! Every fruit and vegetable is organic, super fresh, and packed with flavor!

    I wrote a detailed review of Misfits here.

    Use this link to sign up and you’ll save $15 on your first order. 

  • A SOTA Model On Your Phone? — Testing Qwen 3.5

    What if you could run a SOTA model on your phone for free? Alibaba claims Qwen 3.5 is it. 

    Qwen 3.5 is a new, small model that can run on device. This means you don’t have to pay for inference. 

    Qwen isn’t the most powerful model out there. But for an LLM of its size, it’s darned impressive!

    Let me show you what this thing can do… 

    Round #1: When Will the Iran War End?

    I’ve been glued to my phone the last few days following the war in Iran. When will it all end?

    Let’s see what Qwen thinks…

    For this round, I used the Thinking feature. Qwen didn’t give a firm prediction, but it did outline scenarios for shorter and longer conflicts. It explained that the duration of the war will depend on how many countries and armed groups get involved. 

    Qwen analyzed about two dozen sources, but the quality of the sources was spotty. It relied heavily on Wikipedia rather than primary material.

    I’m giving this round a B. 

    Round #2: Finding New Restaurants

    I always thought you couldn’t get good dim sum in Jersey. But on Saturday, I had some of the best dim sum of my life at Chengdu 1 in Jersey City.

    This made me wonder: what other restaurants in the area am I overlooking? Let’s ask Qwen… 

    I enabled Deep Research and picked the Advanced setting. Qwen asked me if I was interested in any particular town or cuisine. I told it I was open to anything, and Qwen got to work…

    Deep Research took a long time to run — about 10 minutes. But it produced an incredible report!

    Qwen gave me a 14 page memo on great, overlooked restaurants in the area. Mawo’s Kitchen, a Senegalese place in Jersey City, sounds intriguing!

    However, some information was inaccurate. Ubani, a Georgian restaurant, is in NYC not Jersey City. Darn it Qwen, you got my hopes up! 

    I’m giving this round a B. 

    Round #3: Creating Anime Francis

    Qwen doesn’t just produce text responses. It can create images as well.

    I told Qwen to take my X profile picture and turn me into an anime character. Let’s see how it does… 

    I couldn’t help but smile when I saw Qwen’s anime Francis. But Qwen is way behind Nano Banana — the background didn’t change at all. 

    I’ll give this round a B-. 

    Wrap-Up

    Overall, I’m giving Qwen 3.5 a B.

    Qwen isn’t the most powerful model around. For my personal use, I’m sticking with Grok. 

    But Qwen can be a great choice for developers. If you integrate Qwen into your app and run inference on device, it takes your inference costs to zero. 

    You don’t always need the perfect response. You just need one that’s good enough. For a lot of applications, Qwen can get you there. 

    I’m really excited to see smaller, cheaper to run models. That means more cool AI tools for us to use!

    More on tech: 

    Gemini 3.1 Pro Disappoints with Slow Outputs, Weak Sourcing

    Elon Drops Grok 4.20 Beta, The Best Model Yet

    Can DeepSeek Beat the Best American Models?

    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!

    Misfits Market

    I’ve used Misfits for years, and it never disappoints! Every fruit and vegetable is organic, super fresh, and packed with flavor!

    I wrote a detailed review of Misfits here.

    Use this link to sign up and you’ll save $15 on your first order. 

  • Gemini 3.1 Pro Disappoints with Slow Outputs, Weak Sourcing

    Google recently dropped Gemini 3.1 Pro. Google says it’s the most powerful model ever, but it flopped in my testing. 

    This morning I ran 3.1 Pro through a three-round test. I asked it real questions I need the answer to.

    Let me show you where this model falls short… 

    Round #1: Finding the Next Mass Layoff

    Massive layoffs at Block are the talk of the tech world right now. Who’s next?

    I asked Gemini to find the five public tech companies that make the least revenue per employee.

    I got repeated errors on the main Gemini app on both Brave and Safari. Google really needs to straighten this out.

    I booted up Google AI Studio, and the prompt finally ran. It took several minutes to produce a response — much slower than prior versions. 

    Gemini found that tech outsourcing companies like Wipro and Tata make the least money per employee. This simple white-collar work will be the first target for automation.

    The middle class in India and the Philippines is about to be hit hard. 

    Gemini gave me a lot of trouble in this round. 

    Once the prompt ran, the output was useful. But when I checked the sources, none of them supported Google’s response.

    Who knows if this answer is accurate?

    I’m giving this round a C. 

    Round #2: Help Me Brush Up on my Japanese

    I’ve studied Japanese for almost 14 years. Right now, I’m trying to brush up before my trip to Tokyo next month.

    I’m on Duolingo every day and I’m watching Tokyo Vice. What else can I do to prepare?

    Gemini has some excellent ideas, like talking with native speakers on iTalki. It even reminded me of an awesome YouTube channel I used to watch, That Japanese Man Yuta.

    But Gemini didn’t include any links to the content it mentioned. This means I have to Google it myself.

    I’m giving this round a B. 

    Round #3: Creating Openclaw Agents

    I recently stood up a hosted version of Openclaw. What sort of agent would be useful to me as an angel investor?

    Gemini has some great ideas to make my work easier. Getting company info before my meetings would be a big help.

    But Gemini’s sources don’t back up its response. None of them show that Openclaw actually has the skills Gemini mentions.

    I’m giving this round a B. 

    Wrap-Up

    Overall, I’m giving Gemini 3.1 Pro a C+. 

    Gemini’s new model really disappointed me. It feels like a step back from 3.0 — slower and dumber. 

    Results lack sources, making it hard to rely on them. What’s more, the Gemini app has gotten buggy.

    Just a couple of months ago, Gemini beat Grok in my testing. Now, Gemini feels light years behind the new Grok 4.20 Beta. 

    Gemini needs to make the UX smoother and improve sourcing. It also needs to run much faster.

    Until then, I’ll be sticking with Grok.

    More on tech: 

    Six Reasons Why the Citrini Report Is Wrong

    Elon Drops Grok 4.20 Beta, The Best Model Yet

    Grok Companions — Elon’s AI Girlfriend?

    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!

    Misfits Market

    I’ve used Misfits for years, and it never disappoints! Every fruit and vegetable is organic, super fresh, and packed with flavor!

    I wrote a detailed review of Misfits here.

    Use this link to sign up and you’ll save $15 on your first order. 

  • Trading at 0.2 Times Sales — The Most Hated Companies in SaaS

    RackSpace trades at 0.2 times sales right now. Yes, you read that right. 

    The markets hate SaaS today. This morning, I found three of the most beaten-down SaaS companies out there. 

    Everyone is convinced that AI will kill SaaS. This means that you can pick up these stocks for next to nothing. 

    The markets aren’t entirely wrong. A lot of these companies are bloated and poorly run.

    But when sentiment is at its lowest, that’s usually the time to buy. Let’s take a tour of three of the most hated companies in SaaS. 

    OpenText: The Forgotten Cybersecurity Giant

    OpenText is a Canadian cybersecurity company that the markets have left for dead. It trades at just 1.2x revenues. 

    Unlike a lot of beaten-down SaaS companies, OpenText is profitable. 

    It pulled in $436M in profit in the last year, putting the stock at a PE of 15. Revenue declined, but remains up substantially from 2023. 

    OpenText is an extremely bloated company. It employs over 21,000 people to produce $5.2 billion a year in revenue.

    That’s just $244,000 in revenue per employee. Compare that to Robinhood, which produces almost as much revenue with just 2,900 people.

    Mass layoffs could do wonders for OpenText. But until then, OpenText trades for about what it deserves. 

    RingCentral: More Bloated Payrolls

    RingCentral lets you call and text customers across all your devices. Sounds useful, right?

    But investors have abandoned RingCentral. It trades at about 1 times revenue.

    The company is growing modestly at about 4% year over year. It made a small profit of $13M last year on $2.5 billion of revenue. 

    RingCentral is not as bloated as OpenText. Still, they need to do layoffs. They’re employing 4,300 people right now to produce around half the revenue of Robinhood.

    Minimal growth, minimal profit, bloated payroll…no wonder RingCentral is in the dumps!

    RackSpace: The Most Hated Company in SaaS?

    As miserable as OpenText and RingCentral have performed, they’re nothing compared to the ugliest duckling of SaaS: RackSpace. 

    RackSpace trades at 0.2 times revenue. I’ve never seen a multiple this low in my life.

    RackSpace helps companies deploy cloud services. Revenue has declined for three straight years and losses hit nearly $250 million last year.

    RackSpace employs 5,100 people to produce these miserable results. What the hell do they do? 

    If RackSpace had the same revenue to employee ratio as Robinhood, they’d employee 2/3rds fewer people. 

    Shareholders are left with declining revenue, massive losses, and an obese payroll. Maybe this stock deserves to trade even lower?

    Wrap-Up

    There’s one way these terrible companies could not just recover, but become extremely valuable: mass layoffs. 

    Most CEO’s don’t have the courage to axe half the company like Jack or Elon. But what if they didn’t have to?

    These companies are perfect targets for private equity. And believe me, the private equity guys are not afraid to fire people. 

    I’d have more sympathy if these were blue-collar factory workers. But these are well-off, white collar employees getting a free ride.

    How much do you want to bet they work an hour a day? That was the standard when I was in Big Tech! 

    Corporate raiders coming through with chainsaws is ugly. But it’s a necessary part of capitalism.

    Expect to see buyouts hit the tape soon. 

    Have a great weekend, everybody! 

    More on tech:

    Six Reasons Why the Citrini Report Is Wrong

    SaaS Isn’t Dead — Time to Buy the Dip?

    My 2026 Predictions

    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!

    Misfits Market

    I’ve used Misfits for years, and it never disappoints! Every fruit and vegetable is organic, super fresh, and packed with flavor!

    I wrote a detailed review of Misfits here.

    Use this link to sign up and you’ll save $15 on your first order. 

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

    There are a thousand accelerators out there. Most aren’t worth your time. Here’s how to find the best…

    They MUST Give You Money

    A real accelerator gives you money. Never waste your time with one that doesn’t.

    Don’t accept “sweat equity,” where you give away equity for “advice” or “product help”. Those services will be worth what you paid for them – nothing. 

    And certainly don’t pay to be in an accelerator! I can’t believe I have to say this, but some founders are actually paying so-called “venture firms” to be in their program. 

    These programs are a scam. 

    The Check Must be $100,000 Plus

    Say you found a decent accelerator that’s willing to write you a check. Don’t bother with some little check of $25K or $50K. 

    That check needs to be a minimum of $100K. You’re probably giving up 3 months or more of your time. You need to see some real cash. 

    There are a few pre-accelerator programs, like Founder University, that write $25,000 checks. That’s totally fine – you’re only committing a few hours.

    But for a true accelerator, make sure the check is six figures. 

    The Top Accelerators

    Now that we’ve ruled out the worst programs, let’s run through some of the best. Here are my favorites:

    • YC
    • Launch
    • Techstars
    • ERA
    • PearX
    • HF0
    • a16z Speed Run
    • Sequoia Arc

    All these programs are run by real VCs. They give you a substantial check, ranging from $125,000 to a whopping $1 million for Arc and HF0.

    They’re all very hard to get into. And you get a ton of valuable advice and intros along with that juicy check. 

    For almost every pre-seed founder, these programs are worth doing. But if you’re a highly seasoned founder who’s had a large exit, you may not need an accelerator.

    Wrap-Up

    After you finish these programs, you’re likely to raise a substantial seed round. You should be able to raise $2M-$5M without much trouble. 

    Having that fancy logo stamped on your butt really does make a difference!

    Are you planning to go to an accelerator? 

    More on tech: 

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

    Why the ‘Pretty Good Startup’ Can’t Raise Money Today

    Meet My Latest Investment: Cryopets

    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!

    Misfits Market

    I’ve used Misfits for years, and it never disappoints! Every fruit and vegetable is organic, super fresh, and packed with flavor!

    I wrote a detailed review of Misfits here.

    Use this link to sign up and you’ll save $15 on your first order. 

  • At Trump’s State of the Union, a Total Lack of Civility

    When I was growing up, nobody interrupted the President. At last night’s State of the Union, Democrats displayed a total lack of decorum.

    Reps. Ilhan Omar and Rashida Tlaib shouted, jeered, and pointed. Rep. Al Green disrupted the speech by holding up a sign.

    The first time I saw someone disrupt a major presidential address, it was actually a Republican, Rep. Joe Wilson of South Carolina. He yelled out “You lie!” during President Obama’s address to a joint session of Congress in 2009.

    I was shocked, as were many Americans. I’d never seen anything like it. 

    Now, it’s common. 

    Lack of respect is a bipartisan problem. It just happened to be a Republican speaking this year.

    A lot of people hate Trump. I respect their opinion.

    But let’s conduct ourselves with dignity. Let people speak without pointing and jeering.

    When it’s our turn to speak, we can offer a great rebuttal!

    I actually thought that the Democrat rebuttal, by Virginia governor Abigail Spanberger, was pretty good. She hit hard on affordability.

    That’s the way to fight back: with reasoned argument.

    In time, it’ll be a Democrat standing at that podium. On that day, Republicans should sit quietly and listen with respect.

    When it’s over, they should feel free to argue their side.

    We’re about to celebrate our 250th birthday as a country this summer. Let’s remember that whatever our politics, we’re all Americans. 

    More from the blog: 

    Six Reasons Why the Citrini Report Is Wrong

    TrumpRx: Love Him or Hate Him, These Drugs Are Really Cheap

    The 2025 Tremendous Awards

    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!

    Misfits Market

    I’ve used Misfits for years, and it never disappoints! Every fruit and vegetable is organic, super fresh, and packed with flavor!

    I wrote a detailed review of Misfits here.

    Use this link to sign up and you’ll save $15 on your first order. 

  • Six Reasons Why the Citrini Report Is Wrong

    Citrini’s AI doom report makes fun reading. But it’s wrong. Here are 6 reasons why…

    1. Financial Collapse Would Cut AI Spend — In Citrini’s scenario, there is no brake on AI spending. But if AI causes economic collapse, all capital spending will grind to a halt.

      Companies need to borrow hundreds of billions to build AI infrastructure. Good luck doing that in a frozen credit market.
    2. Everything Gets Way Cheaper — If Citrini is right and AI causes a depression, prices will fall through the floor. Workers might be making less money, but they won’t need as much.

      Citrini’s report actually mentions big falls in house prices. But they fail to consider that lower house prices mean you don’t need to earn as much.
    3. NVIDIA Saves the Day — Citrini imagines a world with runaway AI spend. In that world, companies like NVIDIA and TSMC will shoot the lights out.

      Guess who owns NVIDIA and TSMC? Those same white-collar workers that could be put out of their jobs.

      They may not have work anymore, but many won’t need to work. Expect more workers to FIRE.
    4. A New Wave of Founders — If AI makes us way more powerful, it will be much easier to start a business. Displaced workers will create a massive wave of new, AI-enabled companies.

      So many of the costs of running a business will drop to zero. Legal, accounting, coding, all done by AI.

      Why not start your own business?
    5. Big Companies are Slow as Molasses — Citrini thinks that big companies will adopt AI agents next year. I’m guessing they never worked at a big company.

      Enterprises can barely install SharePoint. Many of them are still on premise, 20 years into cloud computing.

      Companies will adopt AI eventually. But this will play out over decades.

      That gives workers a lot of time to adapt.
    6. The UBI Way Out — Even if Citrini is right and AI destroys the economy, the average person will still make a living. It’ll just come from the government.

      If AI destroys millions of jobs, the pressure for a UBI would be overwhelming. Already, politicians compete to see who can offer the most benefits!

    Wrap-Up

    Citrini is wrong. AI is not going to crash the economy.  

    But they are right that some people will be displaced. Many of them will get worse jobs than before.

    Still, that scenario will take decades to play out. This gives us time to adapt. 

    Best of all, AI will create new opportunities we can’t even imagine. 

    Doomerism gets clicks. But long term, optimists win. 

    More on tech:

    SaaS Isn’t Dead — Time to Buy the Dip? 

    My 2026 Predictions

    Using Grok 3 to Manage My Stock Portfolio

    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!

    Misfits Market

    I’ve used Misfits for years, and it never disappoints! Every fruit and vegetable is organic, super fresh, and packed with flavor!

    I wrote a detailed review of Misfits here.

    Use this link to sign up and you’ll save $15 on your first order. 

  • Grok 4.20 Beta Trounces Gemini in Head-to-Head Test

    Gemini ruled the roost until Elon dropped Grok 4.20 Beta. This morning, I tested them head-to-head. Grok clobbered Gemini. 

    Here’s why Grok now reigns supreme…

    Pushing Grok and Gemini to Their Limits

    I gave them both the same prompt, asking them to analyze the likelihood of war with Iran. I made the prompt complex, instructing the models to analyze numerous factors. 

    Here’s the prompt I used: 

    “How likely do you think a war between the United States and Iran is? Consider these factors:

    1. The likelihood of US strikes
    2. The likelihood of a ground invasion
    3. The likely role of Israel
    4. The effects on the Iranian regime
    5. The effects on the world economy
    6. The effects on the world geopolitical and security situation
    7. Polymarket and Kalshi assessments of what will happen in Iran re. strikes, war and possible regime change.”

    Let’s see what these guys can do! 

    Gemini’s Response: Solid, But Lacking in Specifics

    Gemini 3.0 Pro predicts that the United States will likely strike Iran from the air. But it views the odds of a ground invasion as minimal. 

    Gemini did a good job of explaining how conflict with Iran has already pushed up oil prices. Finally, it notes that prediction markets expect strikes and the possible exit of Khamenei. 

    Gemini doesn’t quantify its assessment of the risks of strikes or full-scale war. It simply says high, low, or very low. I would have liked to see a more specific answer. 

    But there’s a bigger problem with Gemini’s response: it didn’t cite a single source. I’m sure it crawled many websites, but I have no idea which.

    How can I rely on Gemini’s response? Is it accurate, or is it just making stuff up? 

    I used the “double check this response” feature, which made Gemini reveal its sources. It provided about a dozen websites that seemed to be high quality.

    But I shouldn’t have to use some special feature in order to get citations.

    Overall, Gemini’s response was respectable. But it should’ve been more specific and cited its sources. 

    On to Grok…

    Grok’s Response: Deeply Researched, Highly Nuanced

    Grok 4.20 Beta gave a thorough, nuanced response. Grok agrees with Gemini that the United States will probably launch airstrikes on Iran soon, but will avoid a major war. 

    Grok even highlighted certain sites in Iran that we are most likely to bomb. It’s like having my own defense analyst on call! 

    Grok predicts that Israel will join the strikes. It gives even odds that Khamenei will be gone by the end of the year. 

    Grok answered every aspect of my prompt beautifully. It even cited 288 sources versus Gemini’s dozen.

    And it took only 27 seconds! 

    A human would take weeks to read and synthesize 288 sources. By then, the whole situation would have changed! 

    These days, AI models aren’t just automating what humans do. They’re doing new things we couldn’t possibly do. 

    The only thing I didn’t like: it’s hard to tell which of Grok’s statements corresponds to which source. I’d like to see numbered footnotes instead of just a list of sources. 

    Still, Grok did an incredible job overall! 

    Wrap-Up

    Grok won this head-to-head test handily. I’m crowning Grok 4.20 Beta the best AI model in the world. 

    Before the Grok 4.20 Beta upgrade, my testing put Gemini 3.0 Pro as the world’s best. But the tables have turned. 

    I’ve never seen any other model search as widely, deeply, and quickly as Grok 4.20 Beta. For $30/month, subscribing to Supergrok and getting access is a steal. 

    I’m excited to see how Gemini responds! 

    If you’re in the NYC area, be careful out there! We’re at 7 inches and counting here at Chez Francis. Can’t wait for spring!

    What’s your favorite AI model?

    More on tech: 

    Elon Drops Grok 4.20 Beta, The Best Model Yet

    Grok Companions — Elon’s AI Girlfriend?

    I Built an AI Pitch Deck Analyzer in Minutes with Grok Projects

    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!

    Misfits Market

    I’ve used Misfits for years, and it never disappoints! Every fruit and vegetable is organic, super fresh, and packed with flavor!

    I wrote a detailed review of Misfits here.

    Use this link to sign up and you’ll save $15 on your first order. 

  • Elon Drops Grok 4.20 Beta, The Best Model Yet 

    Elon just dropped Grok 4.20 Beta. This model blows away anything else I’ve ever used. 

    From speed of search to depth of reasoning, the new Grok shines. Let me show you what this thing can do…

    Round #1: Energy and the AI Build-Out

    Yesterday, I met with an interesting startup that’s helping industrial facilities cut power consumption. Since AI data centers are power hungry, this could be the right company at the right time.

    But how constrained is AI by energy supply? Let’s ask Grok… 

    Grok analyzed 289 sources in just over a minute!

    Grok explained that the build of new data centers is already constrained by a lack of power, especially in the key Northern Virginia region. Sounds like we need that energy efficiency startup even more than I thought!

    Grok’s response reads like it was written by an analyst with years of experience. A report like this would have taken a human days.

    Grok did it in 79 seconds. I’m giving this round an A+!

    Round #2: Handicapping a War in Iran 

    America has more military personnel in the Middle East than at any time since the invasion of Iraq, when I was in high school. Are we about to go to war with Iran?

    I ask Grok to give its own assessment and include predictions from Kalshi and Polymarket. 

    Grok did an incredibly sophisticated analysis. It explained that the buildup is much smaller than in 2003, even if it’s the largest in recent years.

    Kalshi and Polymarket expect limited US strikes on Iran, but not a full-scale invasion. Grok concurs. 

    Can the analysts in the White House produce a report this good? Another A+ for Grok!

    Round #3: Help Me Find the Next Great Startup

    It’s hard to beat AI infra as an investment these days. I’ve had great success with Micro1, which recently passed $200 million ARR.

    This got me thinking: what are the next cards to turn over as we build the superintelligence?

    Let’s ask Grok… 

    Grok came up with some amazing startup ideas. My favorite was producing energy on-site for AI data centers.

    Grok searched 346 sources, the most I’ve seen so far. It took just 49 seconds. 

    This thing could be a better investor than me!  The only appropriate grade is another A+. 

    Wrap-Up

    Grok 4.20 Beta crushed it in every round, earning a perfect A+ overall. 

    I was astounded by Grok’s speed and depth of research. It reasons incredibly well and explains its conclusions clearly. 

    4.20 Beta is only available to paid subscribers of Supergrok right now. If you’re not paying for Grok, I really suggest ponying up the $30 and giving it a try. 

    The power of this model will shock you.

    Have you tried Grok 4.20 Beta?

    Have a great weekend everyone!

    More on tech: 

    I Built an AI Pitch Deck Analyzer in Minutes with Grok Projects

    MiniMax’s New M2.5 Model Flops in Testing

    Meet My Latest Investment: Cryopets

    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!

    Misfits Market

    I’ve used Misfits for years, and it never disappoints! Every fruit and vegetable is organic, super fresh, and packed with flavor!

    I wrote a detailed review of Misfits here.

    Use this link to sign up and you’ll save $15 on your first order.