Everything You Need
To Protect Your Community
Spam protection, fair moderation with AI-reviewed ban appeals, anti-raid defense, analytics, a content scheduler, and a support mode — built on inspectable signals (our own trained model, the CAS blocklist, an AI classifier, keyword rules, OCR, and image hashing), not one opaque black box.
Complete Feature Set
15 tools that cover the full life of a Telegram community — from the first spam message to daily content
AI Moderation Copilot
Describe what you want in plain language — the copilot proposes exact settings and rules, you review and apply.
3-Level AI Detection
Rule-based filters catch obvious spam, a local AI model handles routine cases, and a cloud model reviews the complex ones.
AI Ban Appeals
Punished members appeal in one tap; an AI arbiter reverses mistakes in seconds — automatically.
Anti-Raid Shield
Join-surge detection with automatic storm mode: lockdown, captcha, owner alert, self-restore.
CAS Integration
Connected to Combot Anti-Spam with millions of known spammers. Auto-ban repeat offenders across all groups.
OCR & QR Detection
Extract text from images and QR codes. Catch spam hidden in pictures that humans would miss.
Advanced Moderation
Complete moderation toolkit with warnings, kicks, bans, and customizable rules for your community.
Real-Time Analytics
DAU, messages, member growth, activity heatmaps. Understand your community with data, not guesses.
Auto Triggers
Set up automatic responses to keywords, commands, and events. Build custom workflows without code.
Message Scheduler
Post at an exact date and time in your group's timezone, or on a recurring schedule — daily, weekly, biweekly, monthly, weekdays or weekends only, custom intervals with an end date. AI drafts the posts, RSS feeds supply the content.
FAQ Assistant
Answers repetitive questions instantly — strictly from your own knowledge base, never inventing.
CAPTCHA Protection
Verify new members with customizable CAPTCHA. Block bots while welcoming real users.
Support Mode
Turn your group into a help desk. SLA tracking, ticket management, AI response suggestions.
Smart Notifications
Get alerted on important events - spam waves, member milestones, unusual activity patterns.
Content Filters
Block links, media, forwards, and specific content types. Fine-tuned control over what's allowed.
Anti-Spam Shield
Multi-layer spam protection combining AI, blocklists, and behavioral analysis.
DM Broadcasts
Message your bot's subscribers directly — audience funnels, delivery stats, quotas from 2,500/mo.
Fake Admin Detection
Catch scammers impersonating admins by name, homoglyphs and profile photo — auto ban, mute or notify.
Leaderboard & Gamification
XP, levels and a top-3 podium built from real activity — recognition for the members who make your group.
White-Label Bot
Your own bot's name and avatar on the full Telm engine — our brand stays invisible to your members.
Welcome Messages
Personal greetings with inline buttons and auto-deletion — a clean first contact for every newcomer.
Audit Log & Live Feed
Every ban, mute and deletion recorded with reasons — searchable, streamed live and exportable to CSV.
Public REST API
Check texts with the production engine, manage settings and read the journal — from your own code.
Webhooks
Moderation events pushed to your server in real time — HMAC-signed, deduplicated and retried.
How Telm Compares
Manual moderation, typical keyword bots, and Telm side by side.
| Feature | Manual Moderation | Typical Keyword Bot | Telm |
|---|---|---|---|
| Response Time | Minutes to hours | Seconds | Seconds |
| Detection Method | Admins read everything | Keyword lists | Own trained model + AI + CAS + OCR + rules |
| Understands Context | |||
| False-Positive Handling | Nothing is logged | Out-of-context matches | Logged, reviewable, tunable |
| 24/7 Coverage | |||
| Image Spam Detection (OCR) | |||
| Monitoring Mode Before Enforcement | |||
| Safe Test Mode for Admins | |||
| Ban Appeals with AI Arbiter | DMs to admins — if noticed | ||
| Learns From Your Group's Verdicts | |||
| Anti-Raid Shield with Storm Mode | Manual lockdown | ||
| Cross-Group Protection (CAS) | |||
| FAQ Auto-Answers from Knowledge Base | Admins reply by hand | ||
| AI Content Writer + RSS Autoposting | |||
| Analytics Dashboard | |||
| Cost per Month | Moderator time | From $0 |
How Our AI Works
Three layers of protection working together
Rule-Based Filters
Fast regex and keyword patterns catch obvious spam — crypto scams, adult content, and known bad links — before the heavier AI checks run.
Local AI Model
A fast text classifier runs on our servers to score each message on patterns, context, and user behaviour — no per-message external API call, so checks stay quick.
Cloud AI Analysis
Borderline cases escalate to a large language model for nuanced judgement, catching disguised or context-dependent spam that simple rules miss.
What Telm does when it finds spam
Every action is logged in Monitoring Mode first — you review before Telm enforces anything.
Monitor first
New groups start in Monitoring Mode: Telm records what it would do — warn, mute, ban, or delete — without touching a single member, so you can trust its calls before enforcement.
Warn
Borderline or first-time offenders get a warning, keeping good-faith members in the group instead of banning on sight.
Mute
Repeat or higher-confidence cases are muted for a set duration, stopping the spam without permanent removal.
Delete
Spam messages, scam links, and flagged images are removed automatically so the group stays clean.
Ban
Confirmed spammers and CAS-listed accounts are banned and, optionally, kept out across all of your groups.
Frequently Asked Questions
Direct answers about what Telm can do for your Telegram group or channel
What features does Telm include for Telegram moderation?
Telm combines 15 tools in one bot: AI spam detection with its own trained model, ban appeals with an AI arbiter, an anti-raid shield, CAS blocklist integration, OCR detection of spam inside images, moderation commands with an escalation ladder, real-time analytics, smart triggers, a message scheduler with an AI writer and RSS sources, an FAQ auto-answer assistant, join captcha, a support mode for member requests, owner notifications, content filters, and layered anti-spam with newcomer quarantine.
How does Telm catch spam that keyword bots miss?
Instead of one keyword list, Telm runs every message through layers: fast rules, our own model trained on real Telegram traffic (97.9% of spam caught with 0.04% false positives), and an AI classifier for the hard cases. OCR reads text inside images, image hashing catches re-posted pictures, verdict memory instantly blocks reworded duplicates of known spam, and signals are shared across groups — so a spammer banned in one community can't simply move to the next.
Can I see what the bot would do before it acts on my group?
Yes. Telm starts in Monitoring Mode: it runs every message through the full pipeline and logs what it would do — without touching anyone. When you're ready, switch on enforcement. Admins can also enable test mode, where the bot moderates their own messages as if they were regular members, so you can feel out the strictness on yourself.
What happens if the bot punishes someone by mistake?
The member sees a short, polite notice in their own language (it disappears after 60 seconds) with a private appeal link. An AI arbiter reviews the appeal against the actual messages and the group's context: clear mistakes are reversed automatically within seconds, the ban is lifted, and the member's reputation is restored. Every appeal gets a trackable case number.
When can scheduled messages be sent?
At any exact date and time — interpreted in your group's own timezone — or on a recurring schedule: daily, weekly, biweekly, monthly, weekdays only, weekends only, or fully custom intervals with an optional end date. Scheduled posts support media and buttons, can be edited or cancelled before they go out, and the built-in AI writer or connected RSS feeds can supply the content.
Does Telm work in languages other than English?
Yes. Spam detection is multilingual, including CJK scripts and transliterated text. The dashboard is available in 25 languages, and member-facing notices — like punishment notifications and appeal replies — are sent in the member's own language (13 languages supported).
How much does Telm cost?
There's a free plan for smaller communities, and paid plans as your groups grow. Every feature can be evaluated in Monitoring Mode before the bot takes a single action, so you can see the real value on your own traffic before paying anything.