What we build.

Production Telegram systems — bots, Mini Apps, and automated channels. Every project ships in its own Docker container with CI/CD, monitoring, and a deployment pipeline. Fixed price, fixed timeline, full code ownership.

9 bots shipped grammY + TypeScript Telegram Stars AI integration Docker + CI/CD

bot

Custom Telegram Bots

Your customers interact inside the app they already use. No downloads, no onboarding friction, no app store approval. We build bots that handle payments, answer questions, manage inventory, send notifications, and automate workflows — all inside a Telegram chat thread. Every bot is a standalone TypeScript application with its own database, error handling, and admin interface.

Payments

Telegram Stars integration for frictionless in-app purchases. Invoice creation, pre-checkout validation, refund handling. No external payment processor required.

AI Integration

Claude Vision for image processing, Claude Haiku for classification and enrichment, structured output with Zod schemas. Cost-controlled with daily caps and fallback policies.

Conversation Flows

Multi-step onboarding, form collection, interactive menus. Session management with persistent storage. Graceful timeout and error recovery.

Notifications

Event-driven alerts, scheduled digests, threshold monitoring. Rate-limited delivery that respects Telegram’s flood control. Silent and priority modes.

Admin Controls

Built-in admin commands for monitoring, broadcasting, user management, and diagnostics. Error alerts sent directly to your Telegram chat.

Localization

Full i18n support with per-user language preferences. RTL-aware formatting. Currently shipping bots in English and Polish.

Who this is for
  • Businesses that want to reach customers inside Telegram without building a native app
  • SaaS products adding a Telegram interface to existing services
  • E-commerce operations that need order management, support, or payment processing in chat
  • Teams that need internal automation — monitoring, alerts, approvals, data collection
grammY TypeScript Docker SQLite Telegram Stars Claude AI GitHub Actions EC2
Common questions
How long does a bot take to build?
A focused bot ships in 1–3 weeks. Multi-service systems with AI, payments, and microservices take 4–8 weeks. We define the timeline during scoping — you know the delivery date before we start.
Do I need my own server?
No. We handle deployment to cloud infrastructure (AWS EC2). The bot runs in a Docker container with automatic restarts, health monitoring, and CI/CD. You get full access to the server and can manage it yourself if you prefer.
Can you add features to an existing bot?
Yes. We review the existing codebase, assess the architecture, and scope the additions. If the code quality is solid, we extend it. If not, we recommend a targeted rewrite of the affected components.
Discuss your bot project

mini app

Telegram Mini Apps

Full web applications running inside the Telegram interface. Product catalogs, booking systems, admin dashboards, interactive tools — anything you would build as a web app, but accessible without leaving the chat. Users authenticate automatically through Telegram. No login forms, no password resets, no account creation friction. The WebApp SDK bridges the gap between web and native — haptic feedback, theme matching, and seamless data sharing with the bot.

Web Frontend

Responsive web interfaces built with modern frameworks. Server-rendered or static, depending on the use case. Optimized for Telegram’s embedded browser.

Telegram WebApp SDK

Native integration with Telegram’s WebApp API. User authentication, theme synchronization, haptic feedback, main button control, back button handling.

Bot + App Hybrid

Mini Apps work alongside a companion bot. The bot handles notifications and quick commands; the app handles complex interactions that need a full interface.

Payments

In-app purchases through Telegram Stars or external payment providers. Checkout flows, order management, receipt generation — all within the Mini App.

Who this is for
  • Businesses that need more than a chat interface but less than a native mobile app
  • Product teams building storefronts, booking systems, or dashboards inside Telegram
  • Existing bots that need a visual interface for complex data or configuration
  • Internal tools where Telegram authentication eliminates login management
Telegram WebApp SDK TypeScript HTML/CSS REST APIs Docker nginx
Common questions
What is a Mini App exactly?
A web page that opens inside Telegram’s built-in browser. It looks and feels like a native screen but is built with standard web technologies. Telegram passes the user’s identity automatically, so there is no login step.
Can a Mini App work without a bot?
Technically yes, but practically no. The bot is the entry point — it launches the Mini App, handles notifications, and provides a chat-based fallback for simple interactions. We build them together.
Discuss your Mini App

channel

Channel Automation

Telegram channels that run on autopilot. We build the full content pipeline — source aggregation, deduplication, AI-powered enrichment, branded formatting, and scheduled publishing. You define the niche and the editorial standards. The system curates, processes, and publishes without daily intervention. One non-negotiable rule: nothing posts without AI processing. No raw RSS dumps. Every post gets enriched or it stays in the queue.

Source Aggregation

Multi-source RSS polling, API integrations, web scraping. Configurable polling intervals per source. Automatic deduplication by URL and title similarity.

AI Enrichment

Every article processed through Claude for summary, classification, key point extraction, and relevance scoring. Structured output with retry logic and failure isolation.

Scheduling

Cron-based batch processing. Randomized intervals between posts for organic cadence. Priority publishing based on AI relevance scores. No blocking loops.

Cross-Promotion

Multi-channel networks where each channel drives subscribers to the others. Automated cross-promotion modules. Shared audience growth across your channel portfolio.

Analytics

Source performance tracking, post engagement metrics, subscriber growth monitoring. Data stored in SQLite, accessible via admin bot commands.

Cost Control

Daily AI enrichment caps, configurable batch sizes, efficient prompt design. Running three channels costs under $5/month in API calls.

Who this is for
  • Media companies and publishers that want Telegram distribution without manual effort
  • Niche communities that need a curated content feed — jobs, news, deals, research
  • Marketers building audience channels as lead generation or brand presence
  • Anyone who has tried RSS-to-Telegram tools and found the output quality unacceptable
grammY TypeScript Claude AI rss-parser SQLite Docker cron
Common questions
How much does it cost to run a channel?
Server costs are shared across services (a t3.medium EC2 instance handles multiple channels). AI API costs depend on volume — processing 100 articles per day costs roughly $2–5/month. Total operating cost per channel is well under $20/month.
Can I control what gets published?
Yes. Relevance thresholds filter out low-quality content automatically. You can also set up an approval queue where articles wait for manual review before publishing. Most clients start with approval and switch to full automation once they trust the pipeline.
What happens if AI enrichment fails?
The article stays in the pending queue and gets retried on the next cycle. Nothing posts without AI processing. This is a hard rule — the channel either publishes enriched content or it publishes nothing.
Discuss your channel

Ready to build?

Tell us what you need. We respond within 24 hours with a scoping proposal.

Start a project