Jev Wrapped

Product Introduction

1. Definition: Jev Wrapped is an open-source, AI-powered Telegram channel analytics tool. Technically, it is a web application that leverages a large language model (LLM) named Jev to perform granular content analysis on public Telegram channels.
2. Core Value Proposition: It exists to provide automated, transparent, and data-driven insights into the content composition and quality of any public Telegram channel. Its primary value is in auditing Telegram channel content, analyzing social media sentiment and tactics, and benchmarking channel quality without requiring user registration or a Telegram account.

Main Features

1. Automated Post-by-Post Categorization: Jev Wrapped processes up to 1,500 of a channel's most recent posts individually. For each post, the Jev model (version jev-1.13) classifies it into one of ten predefined content types (e.g., news, jokes, announcements) and evaluates three key qualitative dimensions.
2. Multi-Dimensional Content Scoring: Beyond categorization, the model performs three specific binary assessments per post: identifying paid advertising, detecting the use of clickbait headlines, and flagging content that leverages fear or anger (FUD) for engagement. This provides a nuanced view beyond simple metrics.
3. Interactive, Transparent Visualization & Reporting: The tool generates a visual "Wrapped" card displaying the monthly mix of content types. Crucially, it links directly to the posts that scored highest in each qualitative dimension (e.g., "top clickbait"), allowing for immediate manual verification and audit of the AI's judgment, ensuring model transparency.

Problems Solved

1. Pain Point: The lack of objective, automated tools to assess the substance and health of Telegram channels. Users must manually scroll through feeds to gauge if a channel is news-heavy, ad-spammed, or relies on manipulative rhetoric.
2. Target Audience: Digital marketers researching channels for advertising; journalists and researchers investigating disinformation or media trends; investors and analysts monitoring community sentiment in crypto/tech niches; general users wanting to vet a channel's credibility before subscribing.
3. Use Cases: A marketing manager vetting potential Telegram channels for partnership can instantly see its ad density. A researcher studying conflict-related channels can quantify the use of fear/anger language over time. A crypto investor can check if a project's official channel is primarily substantive announcements or clickbait.

Unique Advantages

1. Differentiation: Unlike simple Telegram analytics bots that track subscriber counts or views, Jev Wrapped analyzes content semantics and intent. Unlike manual review, it processes a year of data in seconds with consistent, documented criteria.
2. Key Innovation: The application of a fine-tuned LLM (jev-1.13) as a consistent "decision model" to perform specific, reproducible classifications on social media content at scale. Its open-source nature and the feature linking to source posts for verification are critical innovations in AI transparency and trust.

Frequently Asked Questions (FAQ)

1. How does Jev Wrapped analyze Telegram channels without an API? It scrapes publicly available content from Telegram's web interface (t.me). Since it only accesses public channels and requires no user login, it operates within typical terms of service, providing a no-sign-in Telegram analysis tool.
2. Is Jev Wrapped accurate in detecting clickbait and ads? The Jev model provides probabilistic classifications based on its training. Accuracy is demonstrated through the tool's core design principle: transparency. Every high-scoring post is linked, allowing users to audit and judge the model's calls themselves, making it a tool for assisted analysis rather than an absolute authority.
3. What are the ten kinds of posts Jev Wrapped identifies? While the specific list is part of the model's fine-tuning, based on the description, categories likely include: News, Paid Advertisement, Joke/Humor, Opinion/Rant, Announcement, Educational Content, Fear/Anger-based content, Clickbait, and other common Telegram post types. The exact taxonomy is defined in the open-source Jev model code.
4. Can I use Jev Wrapped for private channels or groups? No. The tool is strictly for public Telegram channel analysis. It cannot access private channels, groups, or direct messages, as it does not require or support user authentication.
5. What is the "Jev" model and who created it? Jev is a decision-model LLM developed by TypeSafe. It is specifically fine-tuned to make structured, consistent judgments on text, such as classifying content and answering specific yes/no questions about its properties, which makes it ideal for this automated content moderation and audit task.