OCR, document classification and dataset labelling with Jev: builds that read a large collection and return one typed answer per item.
36 builds · page 1 of 2
TypeSafe AI
@typesafeai
we announced Jev 1.14 on 10.14 at 1:14pm PT 🔥
a new and improved model, along with experimental features for our dreamers:
• images
• a new box type
• longer context
surprise, Jev was already able to see you! 👁️
BUT our bar for production-grade intelligence is quite high, so builders beware, these are preview features.
enter the ~*~ danger zone ~*~ at your own risk. 🚨
DocJev is the fastest way to classify and split complex document packets ⚡️. (and the default is fully free and OSS!)
I made a sick teaser video below. TY Opus 5.5 🙏
check it out: github.com/jerryjliu/docj…
I used Jev to classify 1,018 AI research papers.
The result: $0.08 total cost and 256ms median end-to-end latency per paper.
The pipeline was:
1. Summarize each paper with DeepSeek V4 Flash
2. Send the title + summary + 24 possible topics to Jev
3. Use Jev to classify each paper
4. Visualize everything on http://1kpapers.com
The summaries cost $3.99 on @togethercompute. The classifications cost $0.08 on @typesafeai.
So for just over $4 of inference, I ended up with a pretty useful way to explore the top AI research papers from the past year.
I think this is where things are heading: different models for different parts of the workflow, instead of using one model for everything.
I’m running evals on the Jev classifications before replacing the current ones, but the site is already live: http://1kpapers.com
One card in the feed, every twelve, on every page of it. $100 a week, cancel any week.
Milind S
@milindlabs
Okay so Jev can actually do computer use really well
Without any screenshots, or LLMs and no Pixels leave my mac
I dont even read the Dom elements
A local CoreML model segments every button and UI element on screen.
On-device OCR reads the labels. That text is all Jev gets.
It returns a probability across those elements and tells me the best one to click.
Then it clicks, re-runs detection, and decides again. In a loop until the goal is done.
~90ms per decision. Faster than any LLM computer use I've tried.
Blazing fast computer use, without any latency
@typesafeai is building something really interesting
I made a DuckDB extension where you can use @typesafeai 's Jev to do quick classification of rows in any csv/parquet file or duckdb table
about 10sec for 1k rows ~ better than using an LLM, way more ergonomic than a classifier
game-changing for data analysis!
JEV IS INSANE.
I gave it 80 real exam questions and 297 practice ones. In 80 seconds, it told me which ones are most likely to appear on the real exam and which ones aren’t.
All for $0.0256.
Can't stop playing with @typesafeai 😁
Built a construction plan-set classifier with Jev.
Proq turns civil and building plan sets into bills of materials using an LLM pipeline we built on GPT-4.1. Jev classified an entire 26-sheet plan set in 2.9 seconds for $0.0052.
It matched GPT-4.1 and GPT-6 Astra on 100% of sheet-level classifications while running 17–21x cheaper and 5x faster than our production pipeline.
I used Jev to solve a problem every agent eventually runs into: reading logs.
22.8M lines, and running an LLM on every one would've cost $1,120.
Tocsin groups them into 11,812 repeating patterns, then asks Jev about each pattern once.
6 minutes, 64 cents, 123 patterns that actually needed to be looked at.
http://github.com/TPAteeq/tocsin
The paging policy is just a prompt.
You tell it what should wake someone up at 3 am and what's just another log line.
1,000 AI papers sorted into 24 topics for $0.0585. Then Opus 5 graded the labels.
@nutlope's Jev paper map went viral, but the pipeline never shipped and the eval was "still running". So I rebuilt both and opened them.
The first judge run came back empty: Opus spent its whole budget thinking and answered nothing. Reasoning off, second run: it agreed with Jev on 85 of 100 papers, at 153x the cost and 1.9s against 57ms per paper.
The 15 misses are not random. One number Jev already returns tells you which labels to recheck.
Cheap models sort. Expensive models audit only what the cheap one flags.
Repost if you classify anything at scale, because the eval rows are public and anyone can rerun them with their own judge in one command.
Code in the reply.
Nitpicky, an AI generated photo detector powered by jev
AI generated photos can be told from nits. That's why we build something to zoom into every detail:
faces, fingers, characters, numbers, poses, where common senses fall apart, judged by jev
Classifying geopolitical news in real time with Jev.
Ddimakrivolap
Movez
@0xMovez
Jev Founder, Diogo Amogo, just released 12-page PDF on building a Jev Harness for coding agents
this is a 10-step blueprint on how to make your coding agents 200× faster and 400× cheaper:
step 1 → meet Jev: the LLM writes, the harness executes, Jev decides what each turn sees, where it routes and whether it runs
step 2 → ask the question that breaks every agent: how would you design one if LLMs had no KV cache?
step 3 → stop routing blind: Opus → Sonnet → Opus costs 6.19 vs 4.15 for pure Opus, because handing back reprocesses the whole context
step 4 → follow the tokens: reading and searching take 56.2% of tool turns and 46.5% of tokens. Writing code is under 10%
step 5 → score every chunk per query: hide, short summary, long summary or full. Compress after the question, not before
step 6 → disclose tools in tiers: one-line snippets for 100s of tools, schema on demand, docs for one-off queries. Batteries stop costing context
step 7 → load instructions by condition: touching *.tsx loads the style guide, billing/ loads its gotchas file, and compaction can't erase either
step 8 → route by trust, not just difficulty: secrets and infra stay on first-party frontier models, public docs go to the cheapest one
step 9 → share one retrieval pass: cross-model review, eval generation, ELI5 explainers and live progress pages all run read-only in the background
step 10 → gate every command: programmable allow / ask / deny policies that read a script before it runs, not just its name
Send this PDF and the article below to your Claude Code or Codex instance and start shipping 200× faster.
JEV can’t see images. So how do you give it eyes?
What’s possible? What can run locally? How far can you get for free?
The answers (and the catches) are in the full breakdown below:
Jev for RAG
in almost all cases you trust the semantic matching capability of Jev more than dot product similarity
very useful as the direct similarity metric in small data cases
and a great reranker with big data
Another insane Jev use case!
Traditional database filters need precise, predefined conditions. But many questions are semantic:
- Is this article mainly about software engineering?
- Which topic best describes it?
- How technically deep does it appear?
These usually require moving rows into application code, invoking a model, parsing its output, and writing the result back.
pg-jev is an open-source Postgres extension that exposes Jev through SQL functions.
- jev() works inside WHERE
- jev_prob() returns a probability
- jev_choice() selects a label
- jev_score() ranks rows across ordered levels.
It needs no vector column or embedding index.
Under the hood, pg-jev batches rows, sends them to Jev, and returns typed answers that SQL can filter, sort, group, and combine with exact predicates.
The recording below runs it against real Hacker News stories stored in Postgres.
It first shows the rows, then asks Jev to find software-engineering stories, classify them by topic, and rank them by technical depth. The final query reports requests, tokens, cost, and cache hits from the database session.
GitHub repo: https://github.com/realZachi/pg-jev
(don't forget to star it ⭐)
Postgres keeps the data and controls the query, while Jev handles the part SQL cannot express as a deterministic condition.
If you want to understand what Jev is doing under the hood, I also wrote a hands-on guide to building a Jev-style model with open models, entirely locally.
Read it below.
