Quick Overview
In this build-along video, Matt Wolfe demonstrates the development and configuration of an open-source productivity dashboard named Control Center. Created using OpenAI Codex within ChatGPT, the video explains the architecture of the dashboard and provides step-by-step instructions for installing and customizing the tool.
Key Points
- 1.Matt Wolfe created Control Center, a modular open-source personal dashboard designed to centralize industry news, brand mentions, newsletters, audience metrics, tasks, and reminders.
- 2.The application was constructed using OpenAI Codex within ChatGPT to write and debug the Next.js codebase across several iterations.
- 3.Control Center connects directly to a dedicated Gmail inbox through Google OAuth to ingest, deduplicate, and summarize newsletters across industry sources.
- 4.Users can run the application entirely offline using local language models via Ollama or LM Studio, or connect cloud API keys from OpenAI, Anthropic, Google Gemini, or xAI.
- 5.The dashboard code is freely available on GitHub and can run locally or deploy online through ChatGPT Sites hosting.
Summary
Matt Wolfe begins by explaining his shift in focus toward building bespoke tools and customized software rather than relying solely on off-the-shelf generative media models. He showcases a personal command dashboard called Control Center, which aggregates the daily operational tasks of his business into a unified workspace. To demonstrate how accessible custom development has become, he sets out to build an open-source version of this system from scratch.
Using the Codex feature inside the ChatGPT desktop application, Wolfe initiates a local Next.js project called Control Center. In his initial prompt, he outlines the necessary functional components: an executive summary page, an industry news aggregator, a brand mention tracker, a reading queue shelf, audience growth monitoring, a newsletter inbox reader, and a task list. Codex generates the functional structure, after which Wolfe requests UI refinements such as persistent dark mode and resolves broken API integrations over several development iterations.
The resulting dashboard contains several interconnected modules. The Industry tab scans monitored lab sites, RSS feeds, and key topic terms, pulling in recent developments and ordering them by chronological freshness or algorithmic importance. The Mentions tab searches Google News, Bing News, and web sources for personal and business brand names, using identity-aware filtering rules to discard false matches from namesakes such as athletes or actors. The Reminders tab provides a quick-capture interface for saving online articles, videos, and tweets along with contextual notes for future reference.
The Audience view allows users to enter profile URLs for platforms including YouTube, X, Instagram, Facebook, LinkedIn, Threads, and TikTok, displaying aggregate follower counts, platform mix distributions, and individual growth trajectories over time. The Newsletters tab connects to a dedicated Gmail account via Google OAuth client credentials. It scans incoming newsletter issues, extracts unique stories, deduplicates stories covered by multiple publications, and links directly to the original sources.
Wolfe demonstrates how the application can operate with various AI backends. In Settings, users can choose built-in heuristic sorting, connect cloud API keys from OpenAI, Anthropic, Google Gemini, or xAI Grok for enhanced summarization, or run completely offline using local models via Ollama or LM Studio. Finally, Wolfe demonstrates how to clone the public GitHub repository into Codex or deploy it online through ChatGPT Sites hosting to access the dashboard across multiple devices.
Dashboard Architecture and Purpose
Matt Wolfe demonstrates his custom dashboard concept, designed to eliminate daily tab clutter by consolidating business intelligence into a single interface. The application contains dedicated views for daily executive briefs, industry newsfeeds, brand monitoring, task management, link reminders, audience metrics, and newsletter processing.
Building the Application with Codex
The entire Next.js application was built from scratch inside ChatGPT using the Codex environment and OpenAI reasoning models. Wolfe defined core specifications in an initial prompt, verified functional prototypes, added features like dark mode, and iteratively debugged API routes and OAuth integrations over multiple coding sessions.
Core Modules and Automated Intelligence
The Industry and Mentions tabs scan configured websites, RSS feeds, and search engines, using identity-aware filtering to isolate genuine brand mentions from irrelevant namesakes. The Newsletter tab connects to Gmail via OAuth credentials to extract news stories across dozens of newsletter subscriptions, deduplicate overlapping reports, and rank them by importance.
Deployment Options and Local LLM Integration
The project is distributed on GitHub under an open-source repository that can be cloned and run locally in under three minutes. Users have the flexibility to run the intelligence features without cloud dependencies using local runtimes like Ollama and LM Studio, or deploy a remotely accessible instance via ChatGPT Sites hosting.
The Bottom Line
The video demonstrates the complete process of building, configuring, and deploying a personalized business dashboard called Control Center using AI coding tools. It establishes that automated news monitoring, newsletter deduplication, brand tracking, and social metrics can be self-hosted locally or deployed in the cloud without third-party SaaS subscriptions. While the application provides a functional baseline for immediate use, ongoing feature enhancements and individual customization are left to the open-source community.
FAQ
What is Control Center and what are its primary dashboard features
Control Center is an open-source, Next.js personal business dashboard that aggregates industry news, brand mentions, email newsletters, audience analytics, task lists, and web reminders into a single interface.
How does the newsletter module in Control Center process incoming emails
It connects to a dedicated Gmail inbox via Google OAuth, scans incoming newsletters, extracts individual news stories, deduplicates stories covered by multiple senders, and generates concise summaries.
Which local language model runtimes are supported by the Control Center application
Control Center supports local offline processing through LM Studio and Ollama, allowing users to curate and summarize data without cloud API connections.
How can users deploy the Control Center dashboard to ChatGPT Sites
Users with paid ChatGPT pro plans can prompt Codex in the project repository to build the project onto Sites, which deploys the app to ChatGPT servers and generates a secure URL accessible behind login authentication.
How does identity-aware filtering function in the Mentions tab of Control Center
Identity-aware filtering uses context rules and specific exclusions to separate mentions of the user or brand from unrelated entities who share the same name.
Worth watching for
Developers, content creators, and solopreneurs seeking to streamline daily news aggregation, audience tracking, and task management into a customizable, AI-assisted dashboard.
- ai-dashboard
- codex
- chatgpt
- open-source
- productivity-tools
- local-llms