Vibe Coding Revolution: How AI Built a macOS Presentation App in 45 Minutes
In a striking demonstration of AI-assisted development, developer Simon Willison created a fully functional macOS presentation app called Present.app in under an hour using only natural language prompts. The tool, which leverages Swift and SwiftUI to render web-based slides with remote phone control, highlights the accelerating shift toward agentic engineering in software creation.

Vibe Coding Revolution: How AI Built a macOS Presentation App in 45 Minutes
summarize3-Point Summary
- 1In a striking demonstration of AI-assisted development, developer Simon Willison created a fully functional macOS presentation app called Present.app in under an hour using only natural language prompts. The tool, which leverages Swift and SwiftUI to render web-based slides with remote phone control, highlights the accelerating shift toward agentic engineering in software creation.
- 2In a quiet corner of Mountain View last week, software engineer and AI observer Simon Willison delivered a keynote at Social Science FOO Camp that may go down as a watershed moment in the evolution of developer tools.
- 3Using nothing but a series of natural language prompts and an LLM-powered coding assistant, Willison built Present.app — a lightweight, full-screen macOS presentation tool — in 45 minutes.
psychology_altWhy It Matters
- check_circleThis update has direct impact on the Yapay Zeka Araçları ve Ürünler topic cluster.
- check_circleThis topic remains relevant for short-term AI monitoring.
- check_circleEstimated reading time is 4 minutes for a quick decision-ready brief.
In a quiet corner of Mountain View last week, software engineer and AI observer Simon Willison delivered a keynote at Social Science FOO Camp that may go down as a watershed moment in the evolution of developer tools. Using nothing but a series of natural language prompts and an LLM-powered coding assistant, Willison built Present.app — a lightweight, full-screen macOS presentation tool — in 45 minutes. The app, which loads and navigates between web pages as slides, automatically saves state, and can be remotely controlled via a mobile browser, is not just a novelty; it is a functional prototype of what agentic engineering looks like in the wild.
Willison, known for his annual deep dives into the state of large language models, has long advocated for the democratization of software development. His latest project, documented in detail on his personal blog, leverages Swift and SwiftUI to create a 76KB compressed application that eliminates the fragility of traditional web-based slide decks. Unlike Keynote or Google Slides, Present.app treats each slide as a live URL, rendering content directly from the web — a feature ideal for data-rich presentations that require real-time updates. Crucially, the app auto-saves all URLs and recovers gracefully after crashes, a feature that addresses a longstanding pain point for presenters who rely on browser tabs.
What elevates this project beyond a clever hack is its integration of remote control functionality. Willison augmented the app with a minimal HTTP server, exposed via Tailscale, allowing him to control the presentation from his smartphone. The mobile interface includes large navigation buttons, a touch-enabled scroll bar, and font-size controls — all implemented without external libraries. According to Willison’s own code walkthrough, the server was built using raw socket programming, with a hand-rolled HTTP parser that routes /next and /prev requests to update slide state. While this approach introduces potential CSRF vulnerabilities, Willison acknowledges the trade-off: "For this particular application I don’t really care."
The implications of this "vibe coding" — a term Willison coined to describe rapid, intuition-driven AI-assisted development — extend far beyond presentation tools. As highlighted in a recent analysis by Breached.Company, AI agents are now being deployed to autonomously probe and breach enterprise firewalls, demonstrating that the same prompt-driven automation can be weaponized or democratized depending on intent. Willison’s app is the peaceful counterpoint: a tool built not to exploit, but to empower. His experience underscores a growing trend in software engineering: the rise of the "augmented developer," who leverages AI not as a code generator, but as a collaborative partner in problem-solving.
While some may dismiss Present.app as a gimmick, its elegance lies in its minimalism. It does one thing — presenting web-based slides — and does it without bloat, dependencies, or complexity. Willison, who admits he had no prior experience with Swift, was able to build a production-grade macOS application because the language’s native support for web views and system integration made it the ideal vehicle. This mirrors findings in Engineering journal, which notes that "the convergence of domain-specific languages and generative AI is accelerating the prototyping lifecycle across disciplines, reducing barriers between concept and execution" (ScienceDirect, 2026).
As cybersecurity threats grow more sophisticated — with NSO Group and others reporting an uptick in AI-powered phishing and automated exploit generation — the ability to rapidly build secure, purpose-built tools becomes a critical skill. Willison’s project doesn’t replace professional developers; it expands their reach. "I still used a whole bunch of my own accumulated technical knowledge," he writes. "Someone who knew what they were doing could have built a far better solution in the same amount of time." But now, anyone with a clear idea and a prompt can begin to build.
Present.app has been open-sourced on GitHub with the disclaimer: "This app was vibe coded [...] I make no promises other than it worked on my machine!" Yet, its very existence signals a new era — where software creation is no longer the exclusive domain of those who speak the machine’s language, but of those who can articulate their intent clearly to the machines that now listen.

