Jev AI and similar "judgement models" enable zero-shot outcome prediction for businesses, but most users measure only technical accuracy rather than business value. Accuracy metrics can be misleading: a system predicting "no" for every customer message achieves 80% accuracy whilst identifying no at-risk customers. To properly evaluate Jev, businesses must calculate potential earnings. For example, a B2B company might offer a $250 retention discount to customers flagged as likely to cancel, averaging $700 in future earnings from correctly targeted customers who stay. The challenge lies in determining the right probability threshold for action. Setting it too low wastes money on customers who weren't leaving; too high misses genuine retention opportunities. Each mistargeted discount costs $250. Businesses must directly measure how their Jev-based systems impact revenue, not just technical performance metrics.

A16z leads $870m round for US AI startup TypeSafe. TypeSafe AI, a startup that builds AI models for automation, raised US$870 million at a US$7.5 billion valuation in a round led by Andreessen Horowitz, with Sequoia and existing investor DCVC participating. The financing came after the September 15 release of Jev and followed the company's earlier US$40 million seed round when it emerged from stealth in 2026. TypeSafe said one-third of Fortune 500 companies already use Jev. The model uses a transformer architecture, but it is not a large language model and outputs probabilities rather than text. The company said Jev is designed for automation and uses fewer tokens than large language models. TypeSafe was founded in 2024 by former OpenAI researcher Diogo Almeida, former Meta research engineer Sasha Sheng, and engineer Erik Gafni. Stay updated on the go with our mobile app. Get latest insights with smoother, more personalized experience through TIA mobile app. How would you feel if you could no longer use Tech in Asia? Share, tag us, and land on our Wall of!

TypeSafe AI Series A raises $870M at $7.5B valuation. October 9, 2026 · openai/gpt-oss-120b Bottom line. TypeSafe AI closed a $870 million Series A financing round at a $7.5 billion post-money valuation, signaling a major bet by Andre-And et Horowitz, Sequoia Capital and other top investors on the company's Jec "Jev" decision-model platform. What the raise means for developers. Takeaway: TypeSafe will double-down on the features that made Jev popular and add more "machine-native" models, positioning the stack as the premier infrastructure for building AI-enhanced software. * The company promises to push all of the aspects developers love about Jev "to the extreme." * New machine-native models will be released, expanding the range of tasks Jev can handle out-of-the-box. * TypeSafe aims to become the "best infrastructure for building smart software," implying tighter integration with tooling, observability, and deployment pipelines. "We intend to take all of the things you love about Jev to the extreme" - TypeSafe AI blog post What the raise means for enterprises. Takeaway: TypeSafe is positioning Jev as an enterprise-grade decision engine, leveraging existing Fortune 500 traction and promising additional features demanded by large organizations. * A third of the Fortune 500 are already using Jev, according to the announcement. * The company claims to have saved customers "millions of dollars in production" and will now add the enterprise-level capabilities that customers have asked for (e.g., SLAs, audit logs, role-based access). * The funding will likely be used to build out sales, support, and compliance teams to service these large accounts. "A third of the Fortune 500 are getting their Jev on" - TypeSafe AI blog post What the raise means for talent recruitment. Takeaway: TypeSafe is using the financing round as a recruiting signal, emphasizing stability, culture, and growth opportunities. * The blog post states, "We are a real company and we're here to stay," directly addressing concerns about startup longevity. * The tone is deliberately informal ("We're pretty fun") and includes a link to their careers page, suggesting an aggressive hiring push. * Martin Casado, a well-known technologist, joined the board, adding credibility that may attract senior engineering talent. Community reaction on hacker news. Takeaway: The HN discussion highlights skepticism about the valuation, concerns over product moat, and curiosity about how the capital will be spent. * Product moat doubts: Several commenters note that Jev's functionality can be replicated quickly with open-source models and that OpenAI's own Decisions API already competes directly. "It's weird because two days after Jev was released there were a dozen decision models... OpenAI's own Decisions API beats it" - @armcat * Valuation skepticism: Users question how a product with "no moat" commands a $7.5 B valuation, suggesting the round may be driven more by hype than technical differentiation. "a product which has no moat, was already available, was duplicated within a couple of days is valued at 7B" - @prometheus1992 * Marketing vs. engineering: Some see the success as a result of strong marketing rather than a unique technology. "It's clear that they don't have a product with some incredible moat... but they clearly have good engineering and product people... and very strong marketing muscle" - @christina97 * Potential use of funds: Commenters ask how the $870 M will be allocated, with speculation ranging from product development to aggressive PR/astroturfing. "Can someone who actually knows these things share how might a company like this spend $870M over the years?" - @rokhayakebe * Industry context: A few remarks compare the valuation to historic tech bubbles, implying the round may be a sign of a broader AI funding frenzy. "This looks like the top of the dot-com bubble..." - @collimarco Why the raise matters now. Takeaway: The financing underscores the current AI investment climate, where venture capital is willing to pour hundreds of millions into companies that can claim early enterprise traction, even if the underlying technology is rapidly commoditized. * The involvement of Andreessen Horowitz and Sequoia signals confidence that TypeSafe can translate early adoption into a sustainable business model. * The $870 M capital gives TypeSafe the runway to build out a full-stack AI platform - covering model development, infrastructure, enterprise compliance, and talent acquisition - that could lock in market share before larger cloud providers roll out comparable decision-model services. * If successful, TypeSafe could set a precedent for other niche AI startups to secure massive rounds based on early enterprise contracts rather than proprietary breakthroughs. Bottom line recap: TypeSafe AI's $870 million Series A at a $7.5 billion valuation reflects a high-stakes bet that its Jev decision-model platform can become the de-facto infrastructure for AI-driven business logic, despite community concerns about product differentiation and the sustainability of such lofty valuations. Sources. * Dispatch OpenAI Positioned to Fast-Follow TypeSafe's Jev Classification Model OpenAI can likely replicate and integrate TypeSafe's Jev classifier because it already uses token-level micro-classifiers and has the data, compute, and product expertise to turn Jev's approach into a built-in capability. * Dispatch TypeSafe AI Jev System One Model: Fast, Structured Decision-Making AI TypeSafe AI's Jev, the first System One model, delivers frontier-level intelligence for structured decision tasks at 40-200x lower latency and 100-500x lower cost than comparable LLMs, while guaranteeing type-safe outputs and calibrated confidence scores. * Dispatch Laya Open-Source System 1 Decision Model Beats TypeSafe Jev Laya delivers sub-35 ms calibrated decisions across 100+ languages, open-source weights, and 6-8x lower latency than TypeSafe AI's proprietary Jev. * Dispatch Cloudflare Clef Decision Models: Open-Source, Vision-Enabled, and RL-Fine-Tunable Cloudflare released Clef and Clef-flash, open-source decision models that outperform Jev on benchmarks, add vision support, and include a new reinforcement-learning fine-tuning service. * Project kydlikebtc/awesome-jev awesome-jev is a curated catalogue of public examples, benchmarks, and measurements of **Jev**, TypeSafe AI's typed-decision model (Choice, Score, Noul). It organizes 1 207 resources by decision pattern, records source citations and caveats, and includes independent negative-result benchmarks. The repo does not contain the model itself, but serves as a searchable evidence base for developers and researchers wanting to understand how Jev is used in agents, RAG reranking, safety gating, and more.