Amnon Shashua’s Next Act: An AI That Writes Code From Plain English

Amnon Shashua, the Israeli computer scientist who co‑founded Mobileye and AI21 Labs, is quietly preparing an AI‑powered coding product that could turn the dream of “build an app by describing it” into reality. According to The Washington Post’s AI & Tech Brief, the tool is designed to let people create working mobile apps and websites by describing their ideas in plain language—no coding skills required. Shashua’s track record in translating human intent into machine‑readable output (think Wordtune for writing) makes this a natural, if ambitious, next step.

The Washington Post’s technology desk has been running a series called AI & Tech Brief, spotlighting how recent advances in artificial intelligence allow non‑programmers to build functional apps and sites just by typing out a few sentences. One social post promoting the coverage put it bluntly: *“Recent advances in artificial intelligence allow anyone to build working mobile apps and websites just by describing an idea in plain language. See how creating a working app or website can take just minutes — no coding skills required.”* A companion interactive feature, “Anyone can code now. What will you build?”, focuses on AI tools that generate and run code from natural‑language instructions.

Amnon Shashua is the central figure in the brief. He is best known as the co‑founder of Mobileye (the computer‑vision company acquired by Intel) and as co‑founder and chairman of AI21 Labs, a language‑AI company that has already shipped several products aimed at bridging the gap between human intent and structured output. AI21 Labs has raised $64 million in a Series B round (valuation $664 million, total funding $118 million) led by Ahren Innovation Capital Fund, with participation from Shashua himself, Pitango Ventures, TPY Capital, and investor Mark Leslie. The company’s stated mission is to “change the way people read and write” using AI.

For years, “no‑code” has been a buzzword more than a reality. Platforms like Bubble or AppSheet lowered barriers but still required logical thinking and some understanding of data flow. Meanwhile, AI coding assistants like GitHub Copilot and Microsoft Copilot have shown that LLMs can generate functional code snippets, but those tools still assume the user knows what code they need. Shashua’s approach aims to remove that assumption entirely: you describe the app you want, and the AI builds it.

Shashua’s own research supports this vision. In a recent talk at Hebrew University’s “Celebrating AI Revolution” conference, he discussed “the emergence and dangers of reasoning in LLMs,” specifically referencing in‑context learning, prompt engineering, and AI systems that “trade on code.” He noted that improved reasoning abilities in LLMs “will change everything,” moving AI beyond simple pattern matching to complex structured problem‑solving—including coding. This is the same technical underpinning that powers Copilot, but Shashua’s product appears to target the end‑user directly, rather than the professional developer.

Existing AI21 products already map informal language to structured outputs. Wordtune, described as a “deep‑tech writing assistant,” understands context and meaning, not just grammar. It helps users rewrite or refine text, and it integrates with Gmail, Slack, WhatsApp, and Outlook. Wordtune Read lets users speed‑read documents by summarizing long texts and extracting key information. If Shashua can apply that same language‑understanding engine to code generation, the result could be a genuinely intuitive app factory.

For non‑technical founders and product managers, a tool that turns English into executable code could dramatically reduce time‑to‑prototype. Instead of hiring a developer or learning React, a founder could describe a restaurant‑review app or a booking system and get a working version in minutes. That lowers the barrier to entry for the next wave of consumer and small‑business tools.

Shashua’s company AI21 Labs is already set up for this kind of scale. The $118 million in total funding provides runway for product development and distribution. The existing Wordtune user base (integrated into major platforms) gives AI21 a distribution channel that competitors like Bubble or Adalo lack. If the coding product works as advertised, it could be offered as a standalone product or even bolted onto Wordtune as a “build from your words” feature.

There’s also a competitive angle. Shashua has explicitly mentioned Microsoft Copilot in his public remarks, positioning his product in the same category. But while Copilot is tightly integrated into Visual Studio and Microsoft 365, Shashua’s tool may aim for a broader, more general‑purpose web interface—anyone with a browser could describe their idea and hit “build.”

As of the latest reports, several critical details remain unknown: the product’s name, exact launch date, pricing model, and which programming languages or frameworks it supports. The Washington Post coverage frames Shashua’s tool as part of a broader trend rather than a finished product announcement. However, Shashua’s recent social posts about Double.ai—another AI project under his leadership that topped Nvidia’s benchmarks—suggest he is actively pushing the envelope on model performance.

Given the pace of AI development, a public beta or early‑access program could appear within months. For now, the most concrete takeaway is that a proven entrepreneur with deep language‑model expertise is betting that the future of software development is not writing code, but describing it.