Summon the software
We need to stop thinking of software (or apps) as things we “release” and start thinking of them as something we “summon.”
Increasingly, I believe we’ll summon our own software to help us work through the myriad of day-to-day tasks we tackle on any given day. Historically, for ad-hoc use cases such as data analysis or visualization, we’ve relied on solutions like spreadsheets, presentations and docs. I call these the “open canvas” tools. These are highly-unopinionated tools upon which you can build almost anything.
However, I am already at the point where I find it faster and more effective to “summon” some basic software, than try use existing solutions like a document or a spreadsheet. I spin up bespoke software every few days to work on presentations, explore product usage data or brainstorm ideas.
Critics say this is ridiculous, that “normal people” don’t want to make software. Of course normal people don’t want to deal with database migrations and race conditions! But this is predicated on a fixed definition of what it means to make software.
Consider the following prompt, which maps perfectly to a typical “task” at work.
“I’m trying to model our new tiered pricing. Create a calculator app for me. I want sliders for ‘Number of Seats’ and ‘Add-on Usage’ so I can visually see how the margin changes for the customer. Add a button that generates a ‘Compare to Competitor X’ table based on the current slider positions.”
Is it easier (or faster) to try and figure this out in a spreadsheet and desperately remember how to use VLOOKUP, or just pass the data to Claude Code and say, build me a simple app. What took a day, now takes minutes.
Now extrapolate. Look around the corner. As these tools get better and better, it’s clear that users will start to tackle more ambitious projects. Software that starts to interface with their customers, and with their data. Enterprise software will be fine for a long time (governance, trust, security, reliability, compliance), but to me it seems obvious that software is the new spreadsheet.
This future is already here, it’s just not evenly distributed yet.


I agree. This becomes a challenge when building something new. Deciding between API, CLI, MCP, actual frontend app. The uncertainty I face right now is high. But I tend to agree with your direction.
I also take it one step further. I think what Google is doing with multi modal is very interesting. I’d like to deliver more in audio and video as opposed to text. I see no reason why we have to read everything by default.