Aurora Mobile Limited (NASDAQ: JG) announced that its enterprise AI agent platform, GPTBots.ai, has integrated Jev, a decision model developed by TypeSafe AI, to create a two-layer AI architecture within the platform.

The integration, announced Sept. 22, 2026, comes one week after TypeSafe AI released Jev on Sept. 15. The model has also been integrated by Vercel, Cloudflare, LangChain, and Langfuse.

According to the company, the architecture separates decision-making tasks from reasoning tasks. Jev handles routing, filtering, classification, and relevance scoring, while general-purpose large language models such as GPT and Claude handle text generation and open-ended dialogue.

Three existing GPTBots.ai features are now powered by Jev: a model auto-router that directs queries to the most suitable model, a Dynamic Top-K retrieval filter that scores document relevance before passing content to an LLM, and an intent classifier used in its FlowAgent and Workflow modules.

TypeSafe AI's published benchmarks cite Jev's end-to-end response time at 70–500 milliseconds, compared to 3–329 seconds for frontier LLMs. The model is priced at $0.042 per million input tokens with no charge for output, which TypeSafe claims is up to 445 times cheaper than comparable LLM decision tasks. The benchmarks also cite a 0% structured output error rate.

"What Jev gives us is a dedicated decision engine that handles these judgments at a speed and cost point that changes the economics of the entire pipeline," said Chris Lo, Founder and CEO of GPTBots.ai. "Instead of paying for a full LLM call on every routing decision, we now pay a fraction of a cent per judgment — and get calibrated confidence scores to boot."

The platform includes a configurable confidence threshold system, allowing enterprises to set per-workflow rules that determine whether a decision executes automatically, is flagged for review, or escalates to a human agent.