millionways

millionways is the frontier lab for behavioral intelligence behind Thorsten, the first Large Psychology Model designed to give AI the ability to read the room.

[millionways @millionwaysai]

AI can talk. It still can't understand people.

Meet Thorsten-4: a Large Psychology Model, built on 15+ years of human research. It predicts how people think, decide and act.

Finally, AI will be able to read the room.

[Max Weidemann @maxgweidemann]

Language models learned to predict words. Thorsten-4 learned to predict people.

Built on years of real human behavior, it maps how someone communicates to 50 behavioral traits. Ask it the same question twice and you get the same read.

Check it out in the link below.

[millionways @millionwaysai]

Explore millionways: https://t.co/2ReuC8Lcqj

Read the story: https://t.co/ZD1g1N3V9S

[millionways @millionwaysai]

AI can talk. It still can't understand people.

Meet Thorsten-4: a Large Psychology Model, built on 15+ years of human research. It predicts how people think, decide and act.

Finally, AI will be able to read the room.

[millionways @millionwaysai]

Today in @semafor: a closer look at millionways and Thorsten-4.

https://t.co/ZD1g1N3V9S

[Santiago @svpino]

This model analyzes conversations and identifies patterns in how people behave, communicate, and make decisions.

It scores 50 behavioral and psychological traits.

Companies would kill to have this!

Think about how this will change customer service and sales.

A company can connect the model to the tools it already uses to understand what users want and how they’ll react.

The model name is Thorsten-4 and was built by @millionwaysai.

[Chubby♨️ @kimmonismus]

Expert psychologists interviewed 7,000+ people repeatedly over four years. The team then checked Thorsten-4's predictions against what those people actually did.

I'd like to see more AI evaluated this way. An agent handling a difficult conversation needs some grasp of why the person is hesitating and how they might respond.

@millionwaysai built Thorsten-4 around that research. The interviews aren't data you can scrape from the web. The model analyzes communication and can work alongside the AI tools companies already use.

Details / demo: https://t.co/9OotKvdqTE

[Ryan Hart @thisdudelikesAI]

AI is getting insanely good at understanding what we say.

But understanding WHY we say it? That’s a different story.

@millionwaysai just launched Thorsten-4, an AI model designed to understand the person behind the words.

It predicts human motives, decision-making, and reactions with 90% validated accuracy.

And it works with existing AI tools.

This could be a big step toward AI that actually understands people.

[Out of Context Human Race @NoContextHumans]

AI can understand what you say, but understanding why you say it is a completely different challenge.

That's what @millionwaysai is tackling with Thorsten-4. Built on years of real psychology research, it aims to help AI understand people's motives and reactions.

Could be a pretty big step forward for AI agents.👏

[Robin Delta @heyrobinai]

AI is lowkey dangerous

you can be dead wrong and it'll hype you up like you just cracked something

agents already run sales calls, support, interviews and negotiations...

but they read the words and miss the person

picture this:

> a prospect says "sounds great" with doubt behind it
> a client replies "fine" and they're clearly frustrated

the agent hears yes... and moves on

and it caves fast. the FlipFlop study (2024) found leading models flip a correct answer 46% of the time when users push back

@millionwaysai built Thorsten-4 for this... they call it a Large Psychology Model

it's built to read the person behind the words:

- motives
- decision style
- likely reactions

so your agent can decide when to push, when to pause and when to adapt

the data is the interesting part. millionways says:

- 50+ trained psychologists interviewed 7,000+ people over 4 years
- 35M+ words tagged by hand
- predictions checked against what people actually did

which is why they say it can't be scraped off the internet

it works with the AI tools you already use (API, MCP, their platform or self-hosting)... it doesn't replace your stack

Martin Cordsmeier, founder of millionways, explains it in the video below

[aman @Amank1412]

AI CAN READ YOUR WORDS. BUT CAN IT READ YOU?

meet Thorsten-4 by @millionwaysai.

→ Understands motives behind conversations
→ Predicts decisions and reactions with 90% validated accuracy
→ Works with existing AI models and agents

Built on 15+ years of psychology research and 35M+ hand tagged words from real conversations.

AI needs to understand people not just process prompts.

[Karen O’Dell @karmarieodell]

Excited to see this work out in the world. Congrats to Martin, Max and the @millionwaysai team on launching Thorsten-4!

[alexi @aleximarkett]

AI has gotten ridiculously good at understanding what we say, but understanding why we say it? That's a whole different challenge.

Just came across @millionwaysai's Thorsten-4, and I think they're tackling something pretty interesting here.

It's built on 15+ years of psychology research and real human conversations that you can't just scrape from the internet.

They claim 90% validated accuracy in predicting people's motives, decision-making styles, and reactions.

And the interesting part? It works with the AI tools companies already use.

I genuinely think understanding human behavior could be one of the next big unlocks for AI agents.

Worth checking out: https://t.co/tKF9hIwdvS

[Robert Scoble @Scobleizer]

AI agents are running sales calls and job interviews now. They hear "sounds great" and miss the doubt behind it.

@millionwaysai built Thorsten-4 to read the person, not just the words. 50+ psychologists interviewed 7,000+ people over four years and hand tagged 35M+ words. No amount of scraping gets you that.

90% validated accuracy on motives and decision style, and it plugs into the AI tools companies already use.

This is the missing piece for agents.
https://t.co/RpYo0z3C5e

[Parul Gautam @Parul_Gautam7]

AI can understand your words. But can it understand the person behind them?

That's the problem millionways is solving with Thorsten-4, a Large Psychology Model built to help AI understand human motives, decision styles, and likely reactions.

Here's what makes it different:
→ Built on 15+ years of psychology research and human data that can't be scraped from the internet.
→ Predicts motives and reactions with 90% validated accuracy.
→ Works with the AI models and agents companies already use. No need to replace existing tools.

The goal? AI that knows when to push, when to pause, and how to respond to people more effectively.

Explore it and request a demo 👇
https://t.co/xow0HJ0ZZq

@millionwaysai

[Ai With Piyas @piyascode9]

AI needs to understand the person behind the words, not just the words themselves.

@millionwaysai is making this possible with Thorsten-4, a Large Psychology Model that understands human motives, decision styles, and likely reactions with 90% validated accuracy.

✓ Human data that can't be scraped online.
✓ Works with existing AI tools.

A smarter way to make AI understand people.

Learn more: https://t.co/Qa61Rvko1W

[Martin Cordsmeier @outofcords]

AI can write code, pass the bar exam and sound human.

It still can’t tell when you’re unsure, holding back or about to walk away.

Today we’re introducing Thorsten-4, a Large Psychology Model that predicts what people will do next.

[millionways @millionwaysai]

Businesses are putting AI agents into sales calls, candidate interviews and negotiations.

These conversations can change direction when someone hesitates or pushes back. How well does the AI understand what prompted that reaction and how to respond appropriately?

We’ve been working on that question. More from millionways tomorrow.