🔍 Read the full analysis: The Potential Of 'System One' AI To Change How We Think About Artificial Intelligence on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
TypeSafe AI has launched Jev, a ‘System One’ AI model designed for decision automation within software. It produces typed decisions with probabilities instead of text, promising faster, cheaper, and more reliable enterprise AI solutions. This development signals a potential shift in how AI is integrated into business processes.
TypeSafe AI has unveiled Jev, a new ‘System One’ class of AI designed specifically for decision-making within enterprise software, marking a significant shift from traditional language models. Unlike conventional chatbots or reasoning models, Jev produces structured, typed decisions with associated probabilities, enabling faster and more reliable automation. This development could impact how businesses automate routine judgments and decision workflows, reducing reliance on large language models for internal processes.
Jev is built on the concept of ‘System One’ thinking, inspired by psychologist Daniel Kahneman’s distinction between fast, intuitive judgments and slow reasoning. Developed by Diogo Almeida, a co-inventor of RLHF and InstructGPT, Jev is designed to replace free-form text output with structured, schema-conforming answers such as ‘team: billing, confidence: 0.94.’ This allows software to interpret and act on decisions directly, without parsing language or relying on human interpretation.
The model is optimized for automation, answering questions about program state, choices, scores, and yes/no probabilities in a single, parallel pass. It claims to operate at speeds of 70 to 500 milliseconds per query and costs approximately $0.042 per million tokens, making it significantly faster and cheaper than traditional large language models, with claims of being 193.6 times faster and 444.6 times cheaper based on specific workflows.
TypeSafe emphasizes that Jev’s outputs are schema-compliant and free from hallucinations—meaning it cannot generate off-schema or malformed answers—though it can still select an incorrect answer within the allowed options. The company states that Jev’s reliability depends on breaking down complex judgments into smaller, precise questions, rather than asking vague, broad queries.
Jev vs. LLMs: who should make the call?
Jev, from TypeSafe AI, is a “System One” model. It doesn’t write text. It returns a typed decision with a confidence score that your software can act on directly.
Same support ticket, two kinds of answer
“This ticket appears most likely related to billing, although it could also concern account settings or a recent plan change. I would suggest reviewing the invoice history before…”
A person reads it, or code has to parse the prose.
team: "billing"Software reads it and acts. Nothing to parse.
How they differ
| LLM | Jev | |
|---|---|---|
| Output | Text written for people | A choice, a score or a yes/no probability |
| Speed | Seconds per call | 70–500 ms* |
| Price | Input and (pricier) output tokens | $0.042 per million input tokens, output free* |
| Knows when it’s unsure | Often sounds confident when wrong | Confidence score on every answer |
| Explains its answer | Yes | No, which matters for audits |
| Best at | Reasoning, writing, open questions | Routing, tagging, scoring, duplicate checks |
* Vendor-reported. TypeSafe also claims up to 194× faster and 445× cheaper on its own selected workflows.
Accuracy is something you build
Jev is far cheaper and faster, but not more accurate than frontier models. How you phrase the question matters a lot.
TypeSafe’s benchmark scores agreement with two frontier models rather than verified ground truth. The five-question result used weights fitted on 1,000 labelled examples.
The real idea: a confidence dial you control
“duplicate listing”, confidence 0.62
Raise the threshold for fewer mistakes and more manual review. Lower it for more automation and more risk.
Only use Jev when all four hold
Good fits
- Routing tens of thousands of support tickets a day
- Flagging duplicate listings in a product catalogue
- Replacing a keyword filter that mis-tags half its matches
Poor fits
- Drafting customer emails or release notes
- Reviewing a few high-stakes contracts a month
- Anything that needs a written explanation
Implications for Enterprise Automation and AI Decision-Making
The introduction of Jev and the ‘System One’ model represents a potential paradigm shift in enterprise AI, moving away from text generation towards structured decision automation. This approach could significantly reduce costs, increase speed, and improve reliability of internal decision processes, ultimately transforming how businesses implement AI-driven workflows. It also challenges the dominance of large language models in enterprise settings, suggesting a future where AI acts more like a software function than a conversational partner. However, questions about accuracy, scope, and integration remain, and the true impact will depend on real-world adoption and performance.enterprise decision automation software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Evolution of Enterprise AI and the Rise of Structured Decision Models
Over the past three years, the AI industry has seen continuous launches of large language models promising better reasoning, longer context, and improved code generation. Despite these advances, many enterprise applications require reliable decision-making rather than free-form text. The concept of ‘System One’ models, inspired by Kahneman’s fast, intuitive thinking, aims to fill this gap by providing structured, schema-bound outputs that can be directly used by software. TypeSafe’s Jev is the first public example, built with a pedigree of AI expertise, notably Almeida’s background in RLHF and InstructGPT. The model’s launch follows a broader industry trend of seeking more predictable, reliable AI solutions for internal automation, especially as concerns about hallucinations and overconfidence in traditional models grow.“Jev produces typed decisions with calibrated probabilities, making it behave more like a function than a conversational agent.”
— Diogo Almeida, co-inventor of RLHF
structured decision-making AI tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Performance, Accuracy, and Real-World Adoption Still Unclear
While Jev shows promising speed and cost advantages, its accuracy in complex or nuanced decisions remains uncertain. Benchmark tests indicate it can be overconfident or underconfident depending on question type, and its agreement with frontier models is around 68%. Independent tests on phishing detection scored around 62.6%, suggesting room for improvement. Moreover, the model’s effectiveness in diverse enterprise scenarios and its integration with existing systems are still being evaluated. Questions about long-term reliability, error handling, and scalability are unresolved at this stage.
As an affiliate, we earn on qualifying purchases.
Next Steps: Testing, Adoption, and Performance Validation
TypeSafe plans to gather user feedback from early adopters and conduct more comprehensive benchmarks to assess Jev’s reliability across various enterprise use cases. Industry observers will watch for real-world deployments that test its decision accuracy and integration ease. Further development may include expanding the model’s capabilities to handle more complex judgments and refining calibration of probabilities. As adoption grows, comparisons with traditional reasoning models and large language models will clarify Jev’s position within the enterprise AI ecosystem.
As an affiliate, we earn on qualifying purchases.
Key Questions
How does Jev differ from traditional large language models?
Jev produces structured, typed decisions with probabilities, rather than free-form text, enabling direct use by software without parsing or interpretation.
What are the main advantages of ‘System One’ models like Jev?
They offer faster response times, lower costs, schema compliance, and potentially more reliable decision-making for enterprise automation.
What are the limitations or risks of Jev?
Its accuracy in complex scenarios is still being tested, and it can still choose incorrect answers within its schema. Long-term reliability and scalability are yet to be proven.
Will Jev replace human judgment entirely?
Jev is designed for automation of routine decisions within software, not to replace human judgment in complex, nuanced situations.
When will Jev become widely available?
TypeSafe has announced its release, but broader adoption depends on further testing, validation, and integration with enterprise systems.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
