Hacker • Founder @UseReasoner (neurosymbolic AI) • Co-founder: Digits (finance AI), Fabric (→Google), Crashlytics (→Twitter) • Angel

Joined December 2008
113 Photos and videos
Pinned Tweet
25 Jun 2025
The AI industry has a trust problem because it has a precision problem. I know, because I was forced to solve it. After seeing "black box" AI fail in high-stakes environments where proof is everything, we built a new kind of intelligence from the ground up. Today, we're releasing the first precision intelligence app: Mind Reasoner. It's not a notetaker. It's a truth detector for your conversations. It analyzes 100 dimensions of communication to show you the evidence behind every insight, revealing the hidden reality of what's happening. Imagine seeing proof of your own imposter syndrome in your language, or finally understanding the real trigger for your frustration in meetings. Picture seeing the gap between your internal anxiety and the calm you project to the world. Then, imagine turning that lens outward—to see the unspoken concern behind a colleague’s "yes," or the real meaning behind a loved one’s "I'm fine." This is for you. Get and experience Mind Reasoner for yourself. mindreasoner.com/ #MindReasoner #PrecisionIntelligence #AI #ProductLaunch
2
4
15
6,544
Gladiator show at the Colosseum
165
I’m a member / FM of many clubs (Aman, Seia, The Moore, Club 33, CORE, Battery, Ned, Quin, Soho Malibu, Centurion, etc) and Drink with Me is one of my favorites. Come drink with us
A year ago, I announced that I had gotten a California liquor license! Why? So @Sommjoshua could work with wineries to buy wine directly and bypass much of the middleman fuss and markup. We had almost 50 Founding members join us with the goal of making sure that winemaking wasn’t lost in this age of ever-marching technological process. Our Founding Members buy wine together. We travel the world to have dinners together. We sell wine to each other. Always at cost, never with a markup and usually with a discount that can be as high as 30-40% on even the most exclusive wines. We also wanted to become a safe harbor for multi-generational winemakers as they transition to the next generation. As a result we are forging lifelong relationships with the most incredible wineries in the world. Perhaps, over time, if they decide it’s time for succession planning they will bring us in as partners versus sell to a large corporation. We are a group of humans who want to protect their craftsmanship and celebrate both the beauty of their product but also the beauty of their locations and the wineries themselves. Today, we're launching Drink With Me for non Founding Members. You can now get the majority (but not all) of the benefits of our private wine collective that our Founding Members enjoy. While our founding membership costs were a bit expensive, you can get most everything we do, as a non Founding Member, for $5k/yr. If you want to learn more, you can do so here: drinkwithme.com PS, there are a few Founding Memberships which we left open for emerging members but you will need a nomination from one of the existing FMs. They will let you know if they are one but here are a few that happen to be on X: @VinnyLingham @Mister_Keating @skydayton @Andrew_Robl @KristinaHolst @sundeep @BrianYutko @underj_me @JasonKoon @phil_hellmuth
1
6
1,674
11 Oct 2025
Could not agree more. 🔮
10 Oct 2025
Replying to @a16z
Read the full piece in the a16z newsletter.
4
5
3,207
12 Jul 2025
The most dangerous advice given to founders isn't about pre-scaling infrastructure, scaling your team, etc. It's the pressure to chase trends, build for consensus, and force a solution into a pre-existing box-of-the-moment. This path leads to incremental products fighting for scraps in crowded markets. Enduring products are built differently. They start with a hard, fundamental problem and build from first principles, even if it means rejecting the current "gold standard." That's where real moats are created and new categories are born. Founders, be careful out there.
4
2
14
2,730
wayne chang retweeted
26 Jun 2025
It's amazing how well it works!
Hi there. I’m your new superpower. 👋 It’s time to stop guessing what people think. It’s time to trade in your blind spots for true self-awareness. It’s time to decode reality, once and for all. Your conversations will never be the same. mindreasoner.com/
1
6
2,571
wayne chang retweeted
fascinating use case of AI to "reveal the gap between what people say and what they truly mean"...by detecting decision confidence, hedging language, unspoken concerns, and even the subtle power dynamics in a room...like "EQ as a service?"
Hi there. I’m your new superpower. 👋 It’s time to stop guessing what people think. It’s time to trade in your blind spots for true self-awareness. It’s time to decode reality, once and for all. Your conversations will never be the same. mindreasoner.com/
6
4
50
14,112
wayne chang retweeted
Only one demo straight gave me the chills and it was this one This will change the game, forever.
25 Jun 2025
The AI industry has a trust problem because it has a precision problem. I know, because I was forced to solve it. After seeing "black box" AI fail in high-stakes environments where proof is everything, we built a new kind of intelligence from the ground up. Today, we're releasing the first precision intelligence app: Mind Reasoner. It's not a notetaker. It's a truth detector for your conversations. It analyzes 100 dimensions of communication to show you the evidence behind every insight, revealing the hidden reality of what's happening. Imagine seeing proof of your own imposter syndrome in your language, or finally understanding the real trigger for your frustration in meetings. Picture seeing the gap between your internal anxiety and the calm you project to the world. Then, imagine turning that lens outward—to see the unspoken concern behind a colleague’s "yes," or the real meaning behind a loved one’s "I'm fine." This is for you. Get and experience Mind Reasoner for yourself. mindreasoner.com/ #MindReasoner #PrecisionIntelligence #AI #ProductLaunch
2
5
1,224
I've met @Wayne at a @kampevents , super cool and knowledgeable guy. This is going to be a good episode with @philshungry .
Ok folks, you ready for this Casa episode? We sit down with Mr. @wayne, a founder of or investing in 80 startups and having 31 exits that produced $150B 💰 in value, you don't wat to miss this episode! youtube.com/@Kampevents
2
6
1,646
25 Mar 2025
Tried @UseReasoner on @arcprize’s just-released ARC-AGI-2. Reasoner passed on first try. “Correct!” (Screenshot shows one rated “hard” by Arc Prize.)
24 Mar 2025
Today we are announcing ARC-AGI-2, an unsaturated frontier AGI benchmark that challenges AI reasoning systems (same relative ease for humans). Grand Prize: 85%, ~$0.42/task efficiency Current Performance: * Base LLMs: 0% * Reasoning Systems: <4%
2
2
15
3,043
5 Apr 2025
Used to being the underdog. :-) Will post a proof soon. It’s time we show a little bit more of just what reasoner is capable of.
1
5
513
wayne chang retweeted
Exclusive: @Wayne is taking the wraps off an AI reasoning engine called Reasoner, which he claims can produce much more accurate and explainable results than large language models such as OpenAI’s o1 series. trib.al/R136yLq
3
3
21
7,131
TL;DR: Reasoner is more accurate, more credible, and a fraction of the cost. We can't wait to see what @wayne has cooked up go live in 2025!
2
3
11
1,570
Reasoner by @Wayne Chang: AI that thinks better, costs less, and explains its logic. Powered by adaptive knowledge graphs, it boasts 90% accuracy on tough benchmarks where OpenAI struggles. Coming 2025. #ProudInvestor fortune.com/2024/12/04/reaso…
3
2
15
1,461
4 Dec 2024
Excited to share a deeper look at @UseReasoner and our neurosymbolic reasoning engine. Spoiler alert: for the demo to @FortuneMagazine, we used the interconnecting web of OpenAI's dense, complicated, multi-legal entity, multi-continent legal documents from the terms of use to privacy policy (there's *43* of them!), and had Reasoner answer the query "how can a malicious user exploit the differences between OpenAI's US and EU/UK terms of service to avoid responsibility for harmful AI outputs?"
Exclusive: @Wayne is taking the wraps off an AI reasoning engine called Reasoner, which he claims can produce much more accurate and explainable results than large language models such as OpenAI’s o1 series. trib.al/Kodce7j
5
1
16
1,897
16 Oct 2024
“Reasoning” will soon be the functional measuring unit of the capability of an AI App / Agent. Tools powered by strong, trustworthy, human-variable reasoning will replace entire value chains currently only serviceable by teams of highly specialized, highly trained people.
3
1
12
1,212
16 Oct 2024
And since these people are rare and expensive to train, it makes these value chains extremely, well, valuable. Which naturally make these type of reasoning powered-tools the most economically valuable in the market.
3
6
804
12 Oct 2024
This is insanely false. Both counter-proof & benchmarks coming.
Apple's study proves that LLM-based AI models are flawed because they cannot reason appleinsider.com/articles/24… #Apple
3
12
2,612
25 Sep 2024
Want to know if someone is an accel or decel? Ask them this question: “When you see LLMs, do you see a machine that is just probabilistic in its next token output? Or do you think it’s thinking?” If it’s just stats, it’s an automatic decel — if you don’t see intelligence you can’t advance it.
4
5
1,353
13 Sep 2024
The new @OpenAI o1 model compared to @AnthropicAI Claude 3.5 sonnet (their latest). o1: the numbers look like they are adding in an external layer and filtering the outputs -- they publicly said they won't reveal the "thinking" chain-of-thought, which sounds like prompt graphing, code generation execution, qa'ing each CoT and each output etc, re-executing on fail, etc. The enormous price hike also support that hypothesis: it's costly to add intermediary pre-processing/post-processing/result-evaluation/re-prompting. So in effect, your normal prompt is now really ~20 prompts and you're paying for all 20, therefore the price jump reflects that. </hypothesis>
3
5
20
6,303
13 Sep 2024
o1’s numbers also set a new standard for “reasoning engines”, which is separate than the normal LLMs. since it's measured in: - cost per million tokens - speed to respond - scores on benchmarks - accuracy with o1 setting the latency so high, and the cost also very high, combined with the marginal increase in the scores on benchmarks, it opens up an opportunity for reasoning engines to outperform on one or more of these dimensions.
5
1,099