# Intercom Just Got Killed. Don’t Pay a Subscription Anymore Source: https://www.youtube.com/watch?v=Ud9Cr_TnQgE Recap page: https://rapidrecap.app/video/Ud9Cr_TnQgE Generated: 2025-11-23T15:33:17.199+00:00 --- ## Quick Overview The video demonstrates how to use an AI agent (Composer 1, utilizing Sonnet 4.5) within a development environment to rapidly design and implement an Intercom-style help center for the Thumio application, going from initial planning and folder structure creation to generating functional code and fixing build errors in one continuous workflow, ultimately showcasing the speed and capability of AI-assisted development. **Key Points:** - The presenter uses an AI agent (Composer 1 with Sonnet 4.5) to plan and build a modern, Intercom-style help center for the Thumio application, aiming to replace a costly Intercom subscription. - The AI agent successfully generated a detailed folder structure including pages for Index, Home, Category, and Article views, along with necessary components like Headers and Footers. - The agent outlined key decisions, such as hardcoding article content in TypeScript/JSX files initially and using simple client-side filtering for search functionality. - A build error occurred because the agent improperly referenced a file extension (using .ts instead of .tsx), which the agent immediately fixed by renaming the files and updating imports. - The agent quickly generated the content for several FAQ articles, including details on pricing, haybales consumption, and refund policies, demonstrating its ability to populate content based on the application's context. - The presenter emphasizes that this entire complex development task, which would normally take weeks, was accomplished in a single, fast-paced session using the AI agent. - The video contrasts the AI-assisted development speed with the previous cost of using Intercom, suggesting the AI tool eliminates the need for expensive third-party services for this functionality. ![Screenshot at 01:11: The developer navigates the file structure in the coding environment, showing the agent's initial plan to structure the new help center pages and components within the existing Thumio project.](https://ss.rapidrecap.app/screens/Ud9Cr_TnQgE/00-01-11.png) **Context:** The video features a software developer demonstrating the use of an advanced AI agent environment (likely an AI coding assistant like Cursor or similar) to build a complex feature: a fully functional, Intercom-style Help Center for an existing application called Thumio. Thumio is an AI-powered YouTube thumbnail creation tool, and the developer is attempting to replicate the high-quality support experience of Intercom to avoid paying their subscription fees, showcasing the agent's ability to handle planning, file structure creation, code generation, and error correction. ## Detailed Analysis The developer initiates the process by prompting an AI agent (Composer 1, using the Sonnet 4.5 model) to create an Intercom-style help center for the Thumio application, with the goal of replacing an expensive subscription. The agent first generates a comprehensive plan, including a file structure for the help center under pages/Public/HelpCenter/ and defining key components. The plan addresses critical architectural decisions like content storage (initially hardcoded TypeScript/JSX), routing structure (/help, /help/:categoryId, etc.), and navigation links (footer only). Upon execution, the agent encountered a build error due to incorrect file extension referencing (.ts instead of .tsx), which it immediately corrected by renaming files and updating imports across 16 modified files, achieving success in the build phase. The agent then proceeded to generate content for specific FAQ articles, such as 'What are haybales?' and 'What is your refund policy?', populating the content structure derived from the initial plan. The developer is highly impressed with the speed, noting that this entire process would have taken weeks manually, and concludes that AI development is incredibly fast and powerful, even joking about avoiding future corporate sponsorships. ### AI Agent Setup & Planning - Agent Composer 1 using Sonnet 4.5 was instructed to design an Intercom-style help center for Thumio; the agent produced a detailed plan covering folder structure (pages/Public/HelpCenter/), data layer (articles.ts), and component breakdown. ### Content & Architecture Decisions - Key decisions included hardcoding content in TypeScript/JSX files initially, using simple client-side text filtering for search, and opting for footer-only navigation links. ### Automated Debugging - An initial build failed due to a file extension error (.ts instead of .tsx), which the agent immediately identified and fixed by renaming files and updating imports across multiple files. ### Content Generation & Verification - The agent populated content for specific FAQ articles (e.g., 'What are haybales?', 'Billing & Plans'), showing structured data in TypeScript/JSX format for things like free plan limits and export formats. ### Final Outcome & Impression - The entire process, from planning to resolving errors and generating functional UI content structure, was completed rapidly, which the presenter states saved weeks of manual work and eliminates the need for expensive third-party customer service software like Intercom. ![Screenshot at 00:00: The presenter introduces the Thumio AI thumbnail generator platform, setting the stage for demonstrating AI-assisted development.](https://ss.rapidrecap.app/screens/Ud9Cr_TnQgE/00-00-00.png) ![Screenshot at 00:17: The developer switches to the agent interface, inputting the request to build an Intercom-style help center, showing the initial context provided to the AI.](https://ss.rapidrecap.app/screens/Ud9Cr_TnQgE/00-00-17.png) ![Screenshot at 01:13: The developer is inside the VS Code environment, showing the file structure where the agent deleted the old 'HelpCenter' folder, which caused immediate build errors.](https://ss.rapidrecap.app/screens/Ud9Cr_TnQgE/00-01-13.png) ![Screenshot at 02:59: The agent generates a plan, including a detailed folder structure and suggesting the use of Sonnet 4.5 for planning the complex help center implementation.](https://ss.rapidrecap.app/screens/Ud9Cr_TnQgE/00-02-59.png) ![Screenshot at 06:32: The agent outputs the detailed 'Intercom-Style Help Center Implementation' markdown plan, including structure, key files, article categories, and uncertain topics marked with '!!'.](https://ss.rapidrecap.app/screens/Ud9Cr_TnQgE/00-06-32.png)