Why Christiaan Huizer’s eulogy for the “Excel Killer” is right, but missed the true cause of death.
I have a nasty habit of arriving at the funeral before the body is even cold.
This week, Christiaan Huizer published “The Ceiling and the Floor: How the ‘Excel Killer’ lost its way,” a piece that reads less like a tech analysis and more like a breakup letter to a lover who promised to change the world but ended up just changing the font.
It’s a stellar must-read article co-starring drama and sadness. Christiaan asserts that Coda failed because it focused so obsessively on the “High Ceiling” (the Makers, the Architects, the wannabe-engineers) that it neglected the “Floor” (the Editors, the consumers, the people who just want the damn data). We can all agree there’s ample truth in this observation.
On every point, he’s right, of course. But he’s also polite.
Huizer treats Coda’s absorption into the new “Superhuman” suite (the Frankenstein monster stitched together from Grammarly, Coda, and Superhuman Mail) as a “strategic consolidation”. I call it what it is: a salvage operation.
When I wrote “Coda’s Future Demise” back when the valuation was still unicorns and rainbows, I warned that the unit economics of a “Maker-first” platform were fundamentally broken. You cannot build a billion-dollar business on a tool that requires a Computer Science degree to “write” a document. Huizer notes that the “Editor” role was effectively a “Maker-lite” which blurred security lines and frustrated enterprise adoption. He sees this as a UI/UX failure.
I see it as an architectural dead end.
The “Superhuman” Pivot: Pretty, But Vapid?
Let’s look at the new reality. Coda is no longer a standalone “doc as powerful as an app.” It is now the backend [content] database for the Superhuman Suite.
If you look at the new pricing page, Coda is bundled alongside “Superhuman Go”—an agentic layer that promises to “work in every app you use”. This validates my critique in “Coda’s New Look: Pretty, But Not for Power Users”. The rebranding wasn’t just a fresh coat of paint; it was a demolition of the “Maker” interface. The power users—the people who built those complex Packs and schemas—are being pushed into the basement to keep the boiler running, while the front door is widened for the “AI consumer” who just wants Grammarly to rewrite their emails.
Christiaan mourns the loss of the “Excel Killer”. But Coda was never going to kill Excel. Excel is a cockroach; it will survive the heat death of the universe and the frozen vacuum that follows a supernova event, generative AI. Coda’s mistake was trying to be better at being a doc, a grid, a chart, rather than leaning into what it actually was: an operating system for information APIs.
The Agentic Framework: The Ghost in the Machine
This brings me to the one thing everyone seems to be missing, and the subject of my piece “How Packs Could Evolve into an Agentic Framework”.
Huizer points out that Generative AI “commoditized” the core capabilities of smart docs. Why write a complex Filter() formula when you can ask ChatGPT to “show me sales from Q3”?
But look closely at the “Superhuman Go” beta features: “Connector and partner agents”. What are those? They are Coda Packs.
My thesis has always been that Coda’s “Packs” were the real product. The “Doc” was just a canvas. By acquiring Coda, Grammarly/Superhuman didn’t buy a document editor; they bought a runtime environment for agents. They purchased the plumbing that connects Salesforce, Jira, and Slack. Whether they can actually execute on this technical strategy remains to be seen.
The “Demise” of Coda as a document surface was inevitable. But its rebirth as the “headless” brain of the Superhuman Agent is the only logical exit ramp. Huizer rightly complains that the “High Ceiling” was too high for users to reach. The irony is that in this new Agentic world, the user isn’t supposed to reach the ceiling at all. The AI floats up there, pulling the levers the Makers built, while the user stays safely on the floor, typing into a chat box or, soon, speaking to the invisible high ceiling behind the curtain.
The Verdict
Christiaan Huizer is mourning the death of a dream: the dream that normal people want to be software engineers. They don’t. They want magic.
Coda spent ten years building a workshop for magicians (Makers), only to realize that the audience (Editors) just wanted to see the rabbit. The Superhuman acquisition is the final admission that the workshop is closed.
The “Excel Killer” is dead. Long live the Agentic Backend.
