Is The Next Useful AI Model One That Excels In Function, Not Sentence Writing?
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: Is The Next Useful AI Model One That Excels In Function, Not Sentence Writing? on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on office and shipping supplies

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

TL;DR

TypeSafe AI has launched Jev, a new AI model designed for structured decisions rather than text generation. It aims to improve automation speed and reliability, challenging the dominance of traditional language models in enterprise settings.

TypeSafe AI has introduced Jev, a groundbreaking AI model that departs from traditional language models by producing structured, typed decisions instead of generating text. This development challenges the assumption that every enterprise problem requires large language models (LLMs) for natural language processing, emphasizing speed, cost-efficiency, and reliability for decision automation.Jev is part of TypeSafe’s System One Models, designed specifically for decision-making within software. Unlike conventional LLMs, which generate prose or dialogue, Jev responds to structured questions with typed answers, including probabilities and confidence levels. It handles three question types: choices, scores, and yes/no probabilities, and returns results in a single, parallel pass, making it suitable for automation workflows. Launched with $40 million in funding led by DCVC, Jev was developed by Diogo Almeida, a co-inventor of RLHF and InstructGPT at OpenAI, who argues that traditional reinforcement learning techniques are less suited for automation tasks. Instead, Jev employs Reinforcement Learning for Calibrated Decisions (RLCD), which aims to produce more reliable, calibrated outputs. In practical applications, Jev replaces the typical language-based output with structured data, such as “team: billing, confidence: 0.94,” enabling software to act directly on the decision without parsing text. TypeSafe claims Jev answers in 70 to 500 milliseconds at a cost of approximately $0.042 per million input tokens, asserting it is significantly faster and cheaper than existing models—claimed to be 193.6 times faster and 444.6 times cheaper based on System One workflows. The model is marketed as eliminating errors caused by output formatting, although it does not address errors arising from incorrect decision logic or judgment. Initial benchmarks show Jev matching or slightly below the accuracy of frontier models like GPT-6 Astra and Claude Fable 5.1, with some overconfidence in its probability estimates. Independent testing on phishing detection tasks revealed lower accuracy (62.6%) compared to Claude Haiku 4.5 (81.3%), indicating that the model’s effectiveness depends heavily on how it is used and the quality of question design.
At a glance
breakingWhen: announced September 15, 2026
The developmentOn September 15, 2026, TypeSafe AI announced Jev, a decision-focused AI model that produces typed decisions with probabilities, not free-form text, for enterprise automation.

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

A typical LLM

“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.

Jev
team: "billing"
confidence 0.94threshold 0.80: auto-route

Software reads it and acts. Nothing to parse.

How they differ

LLMJev
OutputText written for peopleA choice, a score or a yes/no probability
SpeedSeconds per call70–500 ms*
PriceInput and (pricier) output tokens$0.042 per million input tokens, output free*
Knows when it’s unsureOften sounds confident when wrongConfidence score on every answer
Explains its answerYesNo, which matters for audits
Best atReasoning, writing, open questionsRouting, 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 own workflow benchmark
Jev (ties Claude Sonnet 5)
67.8%
Independent test: 2,000 phishing emails
Jev, asked one question
62.6%
Claude Haiku 4.5
81.3%
Jev, split into five narrow questions
95.0%

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

Jev decides
“duplicate listing”, confidence 0.62
Above: act automatically. Tag, route, merge. Most of the volume ends here.
Below: escalate the unsure few to an LLM or a person.

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

High volumeThousands of small judgments, not a handful of big ones.
Narrow questionRelevance, category or duplicate checks. No reasoning needed.
Cheap errorsA wrong answer costs little, or unsure cases go to something smarter.
Heuristic failureA keyword rule is visibly getting it wrong.
All four true: Jev is a strong candidate
Any one false: use an LLM, or keep your rule

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

Transforming Enterprise Automation with Structured Decisions

Jev’s shift from text generation to structured decision-making represents a significant evolution in AI for enterprise use. By producing typed answers with calibrated probabilities, it aims to improve automation reliability, reduce costs, and eliminate common errors associated with language models. This approach could redefine how businesses implement AI-driven decision processes, moving away from reliance on large language models for every task and toward more precise, schema-conformant automation. If successful, Jev could enable faster, more dependable workflows, especially in areas like support ticket handling, risk assessment, and compliance checks. However, the model’s accuracy and reliability depend on careful question design and calibration, and it is not yet clear how well it will scale across diverse real-world scenarios.
Amazon

enterprise decision automation AI tools

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Evolution of AI in Enterprise Decision-Making

Over the past three years, AI development has focused heavily on improving language models like GPT and Claude, emphasizing reasoning, longer context, and code generation. These models have become central to enterprise AI applications, often replacing manual processes with conversational interfaces. However, critics have pointed out issues such as hallucinations, overconfidence, and mode dropping, which limit their reliability for critical decision-making. In response, some researchers and companies are exploring alternative approaches. Notably, Diogo Almeida, a key figure behind InstructGPT, has argued that reinforcement learning from human feedback (RLHF) is less suitable for automation due to inherent flaws. TypeSafe’s Jev is part of this emerging trend, offering a decision-oriented model that emphasizes schema conformance, speed, and cost-efficiency, aiming to address the limitations of traditional LLMs in enterprise environments.

“Jev is designed to produce decisions that software can act on directly, rather than generating text that requires parsing or interpretation. This makes automation faster, cheaper, and more reliable.”

— Diogo Almeida, CEO of TypeSafe AI

Amazon

structured decision-making AI software

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Limitations and Validation of Jev’s Performance

While initial benchmarks and independent tests provide some insights, Jev’s accuracy varies depending on task complexity and question design. Its performance on real-world enterprise tasks remains to be fully validated, and the model’s calibration of probabilities needs further refinement. Additionally, the reliance on specific workflows and benchmarks means broader applicability is still uncertain. It is not yet clear how Jev will perform across diverse industries or in high-stakes decision-making environments, and whether it can scale effectively without introducing new errors.
Amazon

AI model for business automation

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Future Testing and Integration of Decision-Focused Models

TypeSafe plans to expand Jev’s deployment across various enterprise use cases, gathering real-world performance data and user feedback. Further independent benchmarking and validation are expected to refine its accuracy and calibration. The company may also develop new models based on the same principles, potentially integrating decision-focused AI into broader automation platforms. Additionally, industry adoption and comparative studies will determine whether this approach can replace or complement traditional language models in enterprise AI workflows.
Amazon

decision automation software

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Key Questions

How does Jev differ from traditional language models?

Jev produces structured, typed decisions with probabilities, rather than generating free-form text. This allows for direct automation without parsing or interpretation.

What are the main advantages of Jev?

Jev offers faster response times (70-500 ms), lower costs ($0.042 per million tokens), and improved reliability by eliminating output formatting errors.

Can Jev replace all language models in enterprise use?

It is not yet clear if Jev can replace all language models. Its strengths are in decision automation, but complex reasoning or nuanced language tasks may still require traditional models.

What are the limitations of Jev?

Jev’s accuracy depends on question design, and it can still make incorrect decisions if the input questions are poorly formulated or if the decision logic is complex.

When will Jev be widely available?

TypeSafe plans to expand deployment gradually, with further validation and user feedback guiding broader adoption. No specific timeline has been announced.

Source: ThorstenMeyerAI.com

This content is for general information only and is not financial, tax or legal advice. Consult a qualified professional for decisions about your money.
FALL

Fall Picks

As an affiliate, we earn on qualifying purchases.

You May Also Like

The Internal Customer Perspective: The Key To AI Acceptance

Exploring how internal organizational factors and employee acceptance influence AI deployment effectiveness in 2026.

The Strategic Advantage Of Cheap AI In The Open-Weight Arena

Alibaba’s release of a low-cost, capable open-weight AI model is reshaping developer adoption and distribution in the AI landscape, with geopolitical implications.

Technology Operations Signal Monitor: The Future Of Flipper Zero Development

A new technology operations signal monitor is being tested to track developments like Flipper Zero’s future, aiding small software teams in decision-making.

Software-Defined Warfare: How Ukraine’s Delta Turned the Battlefield Into a Shared, Real-Time Map

Ukraine’s Delta system, a cloud-native battlefield management tool, enhances real-time situational awareness and operational speed, marking a shift in modern warfare.