Engineering Newsletter Built with Grok Bot
This custom engineering newsletter tracks 23 blogs, providing daily digests of new posts. It includes top-tier sources like Uber, Netflix, and Google, all organized in Notion.
view details →research use cases
48 published use cases filed under research.
Grok Creates Private Lists from Replies and Mentions
Grok built four private lists including Inner Circle and AI Builders by analyzing 60 days of replies and mentions. It verified each handle and provided transparency on any unverifiable information.
view details →Grok Bot for Stock Insights
Grok Bot provides updates on the most-mentioned stock tickers and their movements, along with market-moving news, distinguishing confirmed information from rumors.
view details →Custom Legal Jev Model for Trademark Classification
Developed a custom Jev model to classify trademarks under the Abercrombie spectrum, achieving a 75.8% accuracy on LegalBench's Abercrombie test after 7 minutes of training on Google Colab.
view details →Automated RSS Feed Curation Process
This process involves handpicked RSS feeds and an automated curator that evaluates date-bound feeds. A custom set of heuristics is used to score and rank posts, with top selections identified through a classifier.
view details →Decision Language Models for Research Paper Discovery
A tool has been developed using decision language models to assist scientists in quickly finding relevant research papers. The author also benchmarks Jev and OpenAI's Decisions API in a related article.
view details →Automated Research Bot for X
This bot runs research on X three times daily, searching for relevant posts and filing them as PRs. It compiles a morning digest email to keep you updated.
view details →DeepSeek Harness and Jev Code Repository
Access the code, prompts, and run records for results recomputation built on DeepSeek Harness and Jev. Feedback and questions are encouraged.
view details →Institution-Level Invention Counts and Prize Associations
Analysis of 277 institutions shows a correlation between invention counts and science prizes, including Nobels and Turing awards, with a log-log slope of 1.84 and R² of .41.
view details →AI Creates Explainer Video from Research Paper
Claude read a research paper, wrote a script, animated, and narrated a 6-minute explainer video in 4K.
view details →Keynote Intro Alert for Event with Varun Rana
The event features Varun Rana giving the keynote introduction, not a separate stand-up set. This detail ensures attendees are informed and avoid disappointment.
view details →Researching Family History with Advanced Tools
In 2017, I began researching my family history and have since adopted various tools and methods. Recently, I overcame a research barrier using a combination of archival records, DNA triangulation, and probabilistic proof calibration.
view details →Jevland: A Comprehensive Knowledge Base
Jevland is a curated collection of over 646 entries, including 131 open repositories, designed to provide insights into various use cases. Each entry is sourced, featuring official work, integrations, and new projects as they emerge.
view details →Top Use Cases for Muse / Instinct
Explore various applications of Muse / Instinct, including submitting FOIA requests, creating spend-limited Privacy cards, and completing visa forms. Discover how it can streamline tasks like finding travel details and making reservations.
view details →Comparison of Jev and GPT-5.6 for Book Predictions
A test comparing Jev and GPT-5.6 for predicting book ratings using 1,000 Goodreads ratings revealed that Jev was slightly more accurate, 53 times cheaper, and 25 times faster.
view details →Improved Binary Intent Classification Results
Simulation of 100 generations for binary intent classification showed 3600 LLM calls with inconsistent quality. Using Jev, unusable results were eliminated, achieving 100% parseable quality for the next pipeline.
view details →NYC Fraud Watch: Daily Company Research
A daily automated research tool identifies potential fraud-related companies in New York City and State, utilizing public open data. The tool cross-checks contracts and shared identities to flag entities for further investigation.
view details →Jev Model Performance on NOAA Dataset
The Jev model outperformed NOAA's quality layer by identifying 35 issues in climate records compared to 3 caught by hand-written QC rules. It operates at a cost of $0.02 per 1,000 records with a full audit trail.
view details →Pentagon Releases 71 UFO Files Today
The Pentagon has released 71 UFO files, including details on stargates, 480 mph orbs, and a secret $22M program. A case from Albuquerque is noted to be similar to a personal sighting.
view details →Document Extraction with Jev and docsignals
Introducing docsignals, a document extraction workflow that combines DWS with Jev for precise typed answers and exact citations.
view details →AI Feedback System Test Results with Jev
Jev was tested in an AI customer feedback system using 1,040 GitHub issues, proving to be 98% cheaper and 84% faster than Claude Sonnet 4.6 while maintaining higher accuracy.
view details →Pottery Studio Inquiry via Muse Call
A user utilized Muse to call a pottery studio, asking four questions without using their phone. Within two minutes, they received all answers along with a full transcript.
view details →Fine-tuning GLiNER 2.5 Outperforms Jev
Fine-tuning the open-weight GLiNER 2.5 model on a local CPU improved accuracy by ~18pp, surpassing Jev by ~9pp. The process took about 51 minutes and resulted in an 8-10x speed boost for specific tasks.
view details →Evaluation of Jev Model Against LLMs
Early experiments compare the Jev model from Typesafe AI with LLMs like GPT-OSS and GLM. Results indicate Jev may be suitable for production workloads, offering competitive pricing and performance.
view details →