xllify

Copilot's chat interface and tools like Claude for Excel are useful for one-off work inside a workbook: writing formulas, generating VBA, analysing data. Their output stays in the workbook: formulas, LAMBDA names, or VBA that anyone with the file can see and edit. There's no way to package the result as a standalone add-in.

Microsoft also offers a =COPILOT() worksheet function that calls an AI model directly from a cell. It's genuinely useful for text tasks (summarising, classifying), but Microsoft document it as non-deterministic: results may change on recalculation. It requires an internet connection, a paid Copilot licence, and has per-hour usage limits. It's not suited to deterministic calculations.

xllify produces a standalone add-in file, not workbook-embedded code. The Assistant helps you write and refine the logic; once built, it runs identically every time, locally, with no internet connection or per-use cost.

There are many good ways to extend Excel. VBA and LAMBDA are built in and require nothing extra. Excel-DNA prioritises raw performance for .NET developers, producing native XLLs with deep Excel integration. PyXLL and xlwings bring the Python data ecosystem to Excel with rich feature sets spanning custom functions, task panes, and more. Microsoft's own Python in Excel runs Python remotely in the cloud via Anaconda, and like Copilot, requires a paid Microsoft 365 subscription. xllify covers far less ground: it focuses solely on custom functions, removes all setup and tooling from the picture, and once built, everything runs locally.

xllify focuses on one thing: turning function logic into a distributable add-in. Write code directly, paste in a formula or VBA, or describe what you want. The output is a single file, either a web add-in or a native XLL, with nothing for the end user to install or configure.

Office.js is Microsoft's official platform for building cross-platform Excel add-ins using web technologies (HTML, CSS, JavaScript/TypeScript). It works on Windows, Mac, and the web, and supports custom functions, task panes, and ribbon commands. It's the direction Microsoft is investing in.

Building with Office.js means setting up a Node.js project, configuring a manifest, hosting your add-in on a web server, and dealing with the custom functions runtime model. The developer experience has improved but remains involved, especially for someone who just wants to ship a few functions. Distribution is also non-trivial: you need to host the add-in, manage the manifest, and either sideload or publish through the Office Store.

xllify uses Office.js under the hood for its web add-in output, but abstracts away all the setup. You get a working add-in without touching manifests, webpack, or deployment. For native Windows performance, xllify also produces XLLs, which are single-file downloads anyone can open. Hosted add-in distribution is on the roadmap. xllify is currently focused on custom functions, with async and RTD in development.

The C API is the lowest-level way to build XLLs. It gives you maximum performance and full control over Excel integration, including worksheet functions, menu commands, and dialog boxes. It's the foundation that xllify's own runtime is built on.

The trade-off is complexity. You're writing C or C++, managing memory manually, and working with Excel's XLOPER data structures directly. Registration, type conversion, and error handling are all on you. Code is extremely verbose and error-prone, documentation is quite old, and the last SDK release, although stable and perfectly functional, was over twelve years ago. Build tooling is non-trivial and Windows-only.

xllify handles all of that infrastructure. You write in Luau (or describe what you want in plain English), and xllify compiles and packages the result as a single-file XLL.