Jeremy Morrell on Extensible Software and LLMs

On 19th August 2026, Simon Willison shared a quotation from Jeremy Morrell discussing the potential for extensible software on the web. Morrell argues that large language models (LLMs) significantly reduce the cost of creating extensions, while modern sandbox primitives lower deployment costs and offer strong security boundaries. This combination allows applications to maintain a solid, accountable core while enabling users to safely extend functionality in multiple directions, with LLMs filling gaps. Morrell envisions this approach empowering users with enhanced capabilities. The quote was published on Willison's blog, highlighting a forward-looking perspective on software architecture.