A design tool that goes back to the future

I still miss Macromedia Fireworks

Man, I loved that app. It revolutionized design software, and it still doesn’t get credit for how much it shipped first:

• Pixels and paths in one canvas. Bitmap editing and vector drawing together in 1998, when every other tool made you pick a side.
• PNG as the source file. A brilliant hack: the document was a PNG with the editable source inside, so every file was its own preview.
• Live effects. Shadows, glows and bevels that stayed editable on vectors and text. Nothing ever flattened.
• Symbols. Reusable instances with a library, fifteen years before Sketch symbols and Figma components.
• The web-export workflow. Slices, per-slice compression, side-by-side optimize previews. Everyone has this now; Fireworks had it first.
• The property inspector, applied to drawing. Fireworks MX took Dreamweaver’s invention and made it feel native to graphic design.

I was at Google back in the day, and most of the design team used it every day for years. Gmail, Google Maps, Docs, Spreadsheets, Presentations, Sites, and more were all designed in Fireworks.

But then Adobe bought Macromedia and Fireworks was discontinued in 2013. I kept it on my machine as long as I could. There was nothing to replace it with.

Until…

A wild Sketch appears

Sketch came out of nowhere with 2.0 in 2012 and picked up the torch. 1.0 was a little-used, vector-only tool. But 2.0 was a small, fast, native Mac app that was the closest thing Fireworks ever got to a successor. Vector-first but with basic bitmap editing. File-based, no bloat. It won an Apple Design Award that year, and for a good stretch it was where nearly every app interface and icon got drawn.

Sketch earned a credit list of its own, shorter but real:

• Exporting, reinvented. Mark any layer exportable, get @1x and @2x assets in one click. No slices, no Save for Web. Every tool since has copied it; the retina era would have been misery without it.
• Pixel-perfect vectors. Drawing snapped to the real pixel grid, so interface work never landed on half pixels.
• The infinite canvas. Artboards scattered across one endless page, a whole app visible at once. Illustrator had artboards first; Sketch made them the way apps get designed.

In 2017, Sketch was the primary tool for about 70% of interface designers. But a year earlier a little-known browser app had launched that was about to change the game.

Figma enters the chat

Figma launched in 2016. It ran in the browser. A whole team could work in the same file at once. And it was cheap! App designers especially moved over almost overnight. Design became a team sport, for better and for worse, and Figma was built for exactly that.

The wild part is that an app like that shouldn’t have been possible in a 2016 browser. The web couldn’t render a serious design canvas, so Figma didn’t ask it to. Evan Wallace wrote a rendering engine from scratch in C++: text rasterization, masking, blurs, blend modes, the whole graphics stack. He pointed it straight at the GPU through raw WebGL and used none of the browser’s own drawing machinery. When WebAssembly arrived in 2017, Figma was one of the first serious apps on it and load times dropped 3×. Everyone remembers Figma as the multiplayer company. Under the hood it was a graphics-engine company that happened to run in a tab.

The takeover was total. By 2024, Figma was the primary tool for 82% of interface designers. Sketch, which owned the field seven years earlier, was under 2%.

Today Figma keeps getting more expensive and complicated, and Sketch keeps chasing it. In that chase for design-system and multiplayer parity, Sketch drifted from what made it great. Every recent release has been less reliable for me. I had to make a move.

Accomplice is my attempt at a modern day lovechild of Fireworks and Sketch 2.0. Built for one person. Pixels and paths in one canvas. One file on your disk. Native and quick. The gaps between a solo tool and Figma get filled a different way: local AI and Claude Code integration working right beside the canvas, not a platform wrapped around it.

Yes, it’s vibecoded

Accomplice was design directed, tested, and put through its paces by me, but written by Claude Code. I’m not going to be coy about that, it says it right there in the repo.

Here’s what I’d want to know in your position. I use this every day. It’s the tool I make my own work in, so I hit the bugs before you do, and I fix them the same day because the alternative is living with them. Features get added when I need them, which is a narrower filter than a roadmap and a better one. Nothing ships that I haven’t used on something real.

The usual worry about software built this way is that it’s either crap or will be abandoned in a month. The tell isn’t how the code got written. It’s whether anyone actually uses the thing. I do, daily, and I’d have to stop working to stop maintaining it.

And the format is the safety switch. Every document is a PNG with the source riding inside it. Every page ships as plain SVG in the archive. If I write a bad release, your artwork is still sitting there in a format anything can open.

Accomplice is the tool I've always wanted but couldn't build until today. It's built with love and care and I hope you find it useful.

Software is about to get weird

Software as a service is dead. There's no going back.

Figma won't be able to keep charging per seat. Sketch won't be able to keep charging a yearly subscription. Like everybody else, they’ll have to get creative.

You can do what I did and build your own alongside Claude Code. Accomplice is open source. Use it as a starting point or a reference and go nuts. Fork it and let me know what you build.

If you’d rather start with well-polished defaults, give Accomplice a try and tell me what you think.

Thanks for reading and thanks for being here. Let's get weird.

Adam Howell

Most of what you need. None of what you don’t.

This tool is aimed at one person making one file. Being clear about the edges is the point.