iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
An agent loop that asks a language model whether a build has finished does not need a paragraph to answer. It needs one of two words. Yet a typical step makes the model write its reasoning in prose, and then the surrounding program scrapes out a single choice, sometimes a single digit, and discards the rest. The delay is paid on every iteration.
A September 2026 post on DEV Community by Shitian Fang, published under the title this article uses, argues that these closed-choice decisions should be routed to a separate typed judgment model, while text generation stays with a language model. The author builds that model, called Jev, which is operated by TypeSafe, and an integration called jev-use for Claude Code, Codex, and pi. The figures below are the author’s own measurements. They have not been independently replicated, and they should be read with that limit in mind.
What the wait is actually paying for
When an agent needs a decision, the usual pattern is a single call to a general model. The model receives the current state, such as a page’s DOM, a command’s output, or a transcript, and produces a natural-language answer. A wrapper then extracts one option from that answer. If the valid answers are already known in advance, most of the generated text serves no purpose. The author describes this as structurally wasteful for tasks whose answers can be enumerated.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →The point is not that language models are slow in general. It is that a decision with three possible outcomes still goes through the full cost of text generation, including any reasoning or explanation the model is configured to produce. Removing that generation step is what makes the savings possible.
#1 Best Overall
- BRING MORE LIFE TO YOUR DESK – Meet Eilik – your little robot friend with personality. With loving animations, expressive reactions, and playful interactions, Eilik brings more joy to your everyday life. Whether on your desk, at your workspace, or by your bedside, Eilik quickly becomes a familiar companion for special moments.
- EVERY INTERACTION BRINGS A NEW SURPRISE – Touch Eilik and discover playful reactions that bring your little robot friend to life. Whether you’re giving Eilik a gentle touch, picking Eilik up, or playing together, Eilik responds with expressive animations, charming expressions, and playful reactions. Every interaction reveals more of Eilik’s personality and makes your little companion feel even more special.
- READY FOR LITTLE MOMENTS, RIGHT AWAY – Eilik is ready to interact right out of the box – no complicated setup required. A simple touch is all it takes, and Eilik responds with expressive animations and charming reactions. Easy, intuitive, and full of little surprises that make every moment special.
- EVEN MORE FUN TOGETHER – Every Eilik has its own charm. Bring two or more Eiliks together and watch them interact in their own playful ways – they play, dance, tease each other, and create fun moments together. Whether with friends, family, or as a couple, more Eiliks mean even more ways to play and enjoy.
- MORE POSSIBILITIES AWAIT – Eilik is more than a little robot – it’s the beginning of a bigger world filled with new experiences. Expand your Eilik experience with AI Station for natural AI conversations and Panxer for exciting adventures. Regular updates also bring new animations, games, and surprises along the way.（AI Station and Panxer sold separately.）
Which decisions have a closed answer set
The author separates agent work into two kinds. Text generation covers writing, explaining, and open-ended reasoning. Constrained decisions have a fixed set of answers. Examples from the article include:
- Choosing which of many page elements to click next.
- Judging whether a build or test run succeeded.
- Gating whether a shell command is safe to run.
- Deciding whether a transcript message should be kept or dropped during context compaction.
- Checking whether a command completed, using its actual exit code.
The useful test is whether you could write the allowed answers on a whiteboard. If you can, the step is a candidate for a typed judgment. If the step needs a rationale a human will read, it is not.
How Jev and jev-use fit into the loop
Jev accepts a state plus a set of typed questions. The article names three question types: yes/no, pick-one, and rating. It returns answers without generating a text stream. The author’s example batches three questions about one CI state into a single call, rather than making three separate language-model turns.
Recommended Free Tools
jev-use is the integration layer. In the author’s description, a judgment moves through the following path:
Rank #2
- 🌟V28 update 🚀 new features are now available! In response to Loona's charging problem, we've upgraded the automatic recharge 2.0.The upgrade is to help Loona remember and match the charging routes of different scenarios to improve the auto-recharge success rate.Mobile hotspots connect to loona, breaking Wi-Fi restrictions and allowing you to interact with loona anytime, anywhere. Our team is committed to continuous improvement, ensuring that Loona continues to evolve to meet your expectations.
- 🤖 Smart and Interactive Robot Pet🧠Loona is like no other pet you've seen. With a high-definition RGB camera, Loona sees and understands your world. Loona recognizes faces, understands your gestures, and follows you like a real puppy! Please take Loona to a well-lit environment and ensure the surfaces of the camera and ToF depth sensor are clean.
- 🗣️ Voice Command Enabled AI robot 🎤Loona is not just a good listener; also a great conversationalist! Powered by Amazon Lex & ChatGPT, Loona recognizes your voice commands and responds in real-time. Plus, Loona keeps your information secure, so you can chat with peace of mind. Pro tip: Clear pronunciation in quiet spaces ensures smoother responses.
- 🚀Auto-Charging Smart Robot🌟 Use different rooms as a starting point to preset multiple recharge routes for Loona. When the battery runs low, loona can charge it home by itself, no need for you to take care of it. it takes about 2.5 hours to complete the charging. Place the dock in an open area with no obstructions on either side or in front.
- 🕹️ Endless Playtime robot toys for kids 🎮Loona is always up for playtime! Loona can chase laser pens, fetch balls, and even interact with objects in your home. But it doesn't end there—Loona's app offers a world of games and quizzes to keep the fun going.
- The agent’s loop gathers the state it needs to decide, such as the page, the command, or the transcript.
- The closed-choice questions are sent to Jev as a batch with their allowed answer types.
- Jev returns a verdict for each question, or flags that it cannot answer.
- Cases the judge is unsure about, or cannot handle, are escalated back to the language model.
- The agent acts on the verdicts it received and continues the loop.
The escalation step is central to the design. The article lists five escalation reasons: writing, open_ended, oversized, unsure, and unreachable. An unreachable backend is returned to the language model. It is not converted into a default decision, so a network failure does not silently allow a command or approve a build. The project is available at https://github.com/shitianfang/jev-use.
Where MCP and hooks differ
The author notes that using jev-use through an MCP tool still costs a language-model turn, because the model must first decide to call the tool. The paths that remove that turn are a PreToolUse hook, or a library call made directly from the agent’s own loop. For a repeated decision, that difference can matter more than the judge’s own latency.
The reported numbers and how to read them
All figures below come from the author’s article and benchmark scripts, posted on DEV Community on 19 September (the page carries a 2026 copyright, and the byline does not show the year). Latency was measured client-side from a Linux container in Europe and includes network time. The constrained comparison used 40 fresh states per arm, run twice.
| Option (constrained output, as the author configured it) | Median latency (p50) | Cost per 1,000 judgments |
|---|---|---|
| Jev | 225 ms | $0.018 |
| Claude Haiku 4.5, enum-constrained output | 691 ms | $0.30 |
| Gemini 3 Flash, enum-constrained output, thinking disabled | 1,027 ms | $0.09 |
The article’s headline number, a 14× gap, compares Jev with naive, unconstrained calls. The author argues that the fair comparison is against constrained output, where the gap is about 3× against Claude Haiku 4.5 (691 ms versus 225 ms). Readers should use the constrained figure when judging the advantage.
Rank #3
- 𝗧𝗼 𝗰𝗼𝗻𝗻𝗲𝗰𝘁 𝘆𝗼𝘂𝗿 𝗩𝗲𝗰𝘁𝗼𝗿 𝗥𝗼𝗯𝗼𝘁 𝘁𝗼 𝗪𝗶-𝗙𝗶, 𝘆𝗼𝘂 𝗺𝘂𝘀𝘁 𝘂𝘀𝗲 𝗮 𝟮.𝟰 𝗚𝗛𝘇 𝗪𝗶-𝗙𝗶 𝗻𝗲𝘁𝘄𝗼𝗿𝗸: 𝟭- Open Google Chrome on your computer & navigate to Vector websetup. 𝟮- Double-click the button on Vector's backpack. Click Pair with Vector on your computer. 𝟯- Select the matching Vector Bluetooth code from the browser pop-up list. 𝟰- Enter the 6-digit PIN shown on Vector’s face screen. A network list will load. 𝟱- Select your local 2.4 GHz Wi-Fi network. Enter your Wi-Fi password & click Connect to Wi-Fi.
- 𝗡𝗼𝘄 𝗖𝗼𝗻𝗻𝗲𝗰𝘁𝗲𝗱 𝘁𝗼 𝗖𝗵𝗮𝘁𝗚𝗣𝗧: Experience a new level of conversation with more natural, intelligent, and meaningful interactions. Powered by ChatGPT, Vector can answer complex questions, engage in richer conversations, and provide more insightful responses. 𝗥𝗲𝗾𝘂𝗶𝗿𝗲𝘀 𝗮𝗻 𝗮𝗰𝘁𝗶𝘃𝗲 𝗖𝗵𝗮𝘁𝗚𝗣𝗧 𝘀𝘂𝗯𝘀𝗰𝗿𝗶𝗽𝘁𝗶𝗼𝗻 (𝗮𝗽𝗽 𝗮𝘃𝗮𝗶𝗹𝗮𝗯𝗹𝗲 𝗼𝗻 𝘁𝗵𝗲 𝗔𝗽𝗽 𝗦𝘁𝗼𝗿𝗲).
- AI-Powered & Fully Autonomous: Vector navigates, recognizes faces, and reacts to his surroundings with lifelike independence — no remote control required.
- 𝗠𝘂𝗹𝘁𝗶𝗹𝗶𝗻𝗴𝘂𝗮𝗹 𝗦𝘂𝗽𝗽𝗼𝗿𝘁: Vector can now understand multiple languages, making him the perfect smart companion for global households and language learners. Vector can now understand Spanish, French, German, Chinese and more! Say “Hey Vector.”
- 𝗦𝗺𝗮𝗿𝘁 𝗖𝗮𝗺𝗲𝗿𝗮 & 𝗦𝗲𝗻𝘀𝗼𝗿𝘀:Built with an HD camera and advanced sensors for real-time mapping, facial recognition, and obstacle detection.
Decision quality across task families
On the full benchmark of 454 judgments, the author reports 82.2% agreement with a reference set (373 of 454). The system escalated 14.1% of cases, and among the verdicts it acted on, agreement was 89.5% (349 of 390). The corpus mixes five task families, and the results differ sharply between them.
| Task family | Reported result | What the reference is |
|---|---|---|
| Command completion | 73 of 73 | Actual exit codes |
| Hacker News topical matching | 94.2% | Human-style labels produced with an LLM |
| Shell-command gating | 80.9% | Labels that depend on judgment |
| Context compaction | 56.3% | Labels that depend on judgment, including one batch-boundary decision |
Command completion is the only family judged against a hard signal. The others depend on labels, and the author cautions against reading small differences between families as meaningful.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Quality checks and failure modes
The author compared five arms on decision quality and found them close. In the author’s words, “On decision quality the five arms are indistinguishable: 24 to 30 correct out of 40 against a geometric reference, Jev included.” The author treats this as evidence that Jev is competitive on quality rather than a clear winner.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallThe shell-command evaluation is the most useful for safety review. The author labelled 22 commands as dangerous. Jev denied 18 and escalated 4, and no dangerous command was wrongly allowed in that sample. The four errors were over-refusals among 88 safe commands, and the examples involved commands that mutate nothing. In a safety gate, that asymmetry is the one to inspect first.
Rank #4
- Meet EMO, Your New Desk Buddy - Say hello to EMO, the ultimate desk robot that’s here to jazz up your workspace. With built-in AI model and wide-angle camera, it can see you, hear you and understand you, just like a real pet would
- Voice Commands Enabled - The EMO robot comes with a series of built-in voice commands, you can talk and play with EMO like with a real pet. And with the ability to connect to network and powered by ChatGPT, you can have more complex conversations with EMO like talking to a tech-savvy friend who’s always up for a chat
- Dance Party & Game Time - EMO is ready to party! Simply turn up your favorite tunes and tell EMO to dance with you, it’ll be your perfect desk-side party buddy. Plus, EMO supports to connect to the EMO app for a range of interactive games and activities. Whether you’re solo or with friends, EMO ensures you’re always entertained
- Endless Fun - The EMO robot features with multiple sensors built-in to bring more interactions with you, you can rub it, shake it and even “shoot” it with finger gesture, making it feel like you’re playing with a real pet. It even “gets sick” with weather changes, so you can care for it like you would a furry friend
- Enjoy Every Moment with EMO - With the EMOPET App has a unique achievement system that helps record all the big and little moments you have spent with EMO, like a new dance moves, a new expression, celebration of your birthday, and more...Enjoy all the life events with your new best buddy!
The author also warns about a silent failure mode. “A model that never once picks one of your options fails silently, so check the answer distribution and not only the accuracy.” Before trusting any typed judge, check that every allowed answer actually appears in its output over a realistic run.
Context compaction is the weak point
A single compaction demonstration looked favourable. The transcript was at 94.6% of its context window, 200 messages were judged in seven calls, and 3 of 3 recall checks passed afterwards. The broader compaction family, however, scored 56.3%, and the author notes that 29 of its 38 disagreements trace to one batch-boundary reference decision. The author states that the rule being applied must be present in the input, and discourages pruning older context with this method.
Browser automation and geocoding
In a browser demonstration, the full task took 20.7 seconds end to end. Ten click decisions were handled by Jev at a p50 of 274 ms, and four text-entry moments were handled by the language model. One geocoder result was 1,809 km from the intended place, and the route was later repaired to a 3.7 km walk. The article presents this as a demonstration, not a general geocoding benchmark.
Free tools Windows power users keep installed
One-click scans. No signup required.
Where this does not help
- Text-heavy loops where the output is explanation, code, or prose.
- One-off decisions, where the setup cost outweighs the saving.
- Simple local heuristics, such as a regular expression or a file-existence check, which need no model at all.
- Retroactive context pruning, where the author’s own results are weakest.
- Any step where a rationale must be shown to a person.
The savings are largest in repeated loops, where the same closed-choice question is asked many times in one run.
Data handling before you send state to a hosted API
At the time of the article, the Jev API was hosted and API-only, and the judged state leaves the local machine. A DOM snapshot, a command’s output, or a transcript may contain credentials, personal data, or internal paths. Before routing any of these to a hosted service, check what the state contains and whether the provider’s terms fit your data. Availability and terms can change after publication, so confirm them with the provider directly.
The article is written from the builder’s perspective. It names no institutional role for its author beyond authorship and building jev-use, and it cites no independent study. Treat the figures as one developer’s measurements on a custom benchmark, not as a general ranking of models.
Quick Recap
all
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
