Build an AI Agent in Telegram with Vercel, Cursor, Composio

Build an AI Agent in Telegram with Vercel, Cursor, Composio

by Shawn Esquivel

Applied Sciences & Engineering AI — News & Applications UYComputer scienceUYQArtificial intelligence (AI)
🎙 Shawn Esquivel 👥 11.8M 📅 June 19, 2026 ⏱ 67 min 👁 31K 📄 tutorial 🧭 2026-08-03
Available in: English (current) Français

Keywords

AI agenttool callingOAuthmemorydeployment

Summary

This course teaches how to build a production-ready AI agent from scratch, inspired by OpenClaw. The agent is built using the Vercel AI SDK chatbot template, which provides the UI and database setup. Composio is integrated to give the agent access to over 1000 apps via OAuth, enabling actions like fetching emails and drafting replies. Supermemory is added for long-term memory, allowing the agent to learn from past conversations. The agent is made accessible via Telegram, and scheduled automations are implemented using Vercel Cron. The course covers core concepts such as tool calling, context engineering, OAuth, memory, authentication, database migrations, multi-channel interfaces, and deployment. The instructor, Shawn Esquivel, is a developer relations engineer at Composio and a Cursor Ambassador. The course is hands-on, with the instructor using Cursor to generate code, and emphasizes understanding and reviewing AI-generated changes. The final product is a deployable agent that can be used in real-world scenarios.

152 words

Critical Evaluation

The course provides a comprehensive, step-by-step guide to building an AI agent, leveraging modern tools and platforms. The value lies in its practical approach: it demonstrates how to combine Vercel AI SDK, Composio, Supermemory, and Telegram to create a functional agent. The instructor’s expertise is evident, and the use of Cursor to generate code makes it accessible even to those with limited coding experience. The argumentation is solid, with clear explanations of each component’s role. However, the course is heavily promotional, as Composio provided a grant, and the instructor is a developer relations engineer at Composio. This introduces a potential bias towards Composio’s ecosystem, and the course does not critically evaluate alternatives. The scientific rigor is moderate: while it covers technical details, it lacks in-depth discussion of limitations, security considerations, or ethical implications. The sources cited are primarily official documentation and templates, which are reliable but not independent. The title accurately reflects the content, and the course delivers on its promise. Overall, it is a valuable resource for developers interested in building AI agents, but it should be complemented with more critical perspectives.

183 words

Title / Content Match

The title accurately reflects the content: building an AI agent for Telegram using Vercel, Cursor, and Composio.

Quality & Reliability

8/10

The course is a practical tutorial by a developer relations engineer at Composio, with a clear structure and hands-on demonstrations. It leverages official templates and tools, and includes transparent cost information. However, it is promotional in nature (Composio grant) and lacks deep theoretical rigor or independent verification of claims.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

Concurring Sources

Contribution & Novelties

The course provides a practical, end-to-end guide to building a production AI agent, integrating multiple services (Vercel AI SDK, Composio, Supermemory, Telegram, Vercel Cron). It emphasizes using AI tools like Cursor to accelerate development, making it accessible to non-experts. The novelty lies in the combination of these tools and the focus on real-world deployment.

Pour aller plus loin :

120 words

Radar Profile

The radar profile shows high scores in information quantity and quality, reflecting the comprehensive and practical nature of the tutorial. The technical level is moderate, suitable for developers with some experience. The reliability is high due to the use of official tools and templates, though the promotional aspect slightly reduces the overall score.

Reliability 8/10