Entry 001 was a record of calls I had already made in public. This one is different. These are new claims, written down here first, while they are still uncomfortable — which is the only honest time to enter a prediction.
The call, in one sentence:
Three years from now, the primary way people operate their businesses will not be clicking and typing.
Why the keyboard is the bottleneck
Typing is the slowest part of knowledge work, and we have stopped noticing because it is the part we were trained in. A thought arrives fully formed. Then it waits — for a laptop to be opened, an app to load, a cursor to be placed, and a hand to translate it at a fraction of the speed the thought occurred.
Clicking is worse, because clicking is not even expression. It is navigation. It is the tax you pay to reach the place where expression becomes possible.
The interface has been the constraint the whole time. We built software around the keyboard because the keyboard was what we had, then mistook that constraint for the natural shape of work.
Speech runs several times faster than typing for most people, requires no surface, no posture, and no device retrieval. And it now survives contact with machines well enough to be an instruction rather than a transcript.
What this actually looks like
I want a thought to become a published change without a computer entering the story.
Standing on a golf course, I should be able to say into a watch that the seasonal promotion on the home page needs to change, and have it change — drafted, checked against the brand, staged for approval, and published. On a beach, I should be able to talk out a blog post, have it written in my voice, and have it live before I have finished the walk.
Not "open an app and dictate into a text field." That is still typing with your mouth. I mean voice as the operating layer: voice fed, voice analyzed, voice activated. The system hears intent, understands the business it is attached to, decides what has to happen, and executes — then reports back in the same channel.
The architectural call — and this is the contrarian half
Most of the industry assumes voice-first computing arrives as a hardware story: better chips, on-device models, new gadgets, local compute powerful enough to run everything in your hand.
I think that is the wrong bet, and it is the part of this entry I most expect to be argued with.
The unlock is not silicon. It is orchestration — voice control routed through pipelines and browser-control layers, where the work happens in systems that already have context about your business, and the device in your hand only has to capture intent and carry a response back.
If that is right, the watch is already sufficient. It does not need to be smart. It needs to hear you and reach something that is. The capability stops being a function of what you are holding and becomes a function of what your business is connected to — which is a very different thing to invest in, and a very different thing to sell.
It also means the winners will not be the companies with the best device. They will be the companies whose systems can be driven by intent, because they were built with the pipes exposed.
We already bet the company on it
This is not a stage observation. When we designed Hydra, we built on the premise that within a year our own operators would not be typing to run client sites — that the platform would be driven by instruction rather than by forms and dashboards.
That premise shaped real architecture decisions: control surfaces that a pipeline can drive, browser-level automation instead of humans clicking through admin screens, and a knowledge layer the system can consult so an instruction like "change the promotion" resolves into the right edits without someone specifying every field.
We spent the money on the assumption before the market agreed with it. That is the same posture as the 2008 search bet and it is the loop described in 'Light Years Ahead' Is Not a Talent — you place it early, in production, and let the evidence grade you.
On the record as of August 2026
| # | The call | Where I said it | Review |
|---|---|---|---|
| 006 | Clicking and typing stop being the primary way businesses are operated. By 2029, for the routine operating work of a small or mid-sized business — content changes, approvals, status checks, publishing, reporting requests — voice becomes the dominant initiating interface, with keyboards reserved for dense authoring and code. | This entry | August 2029 |
| 007 | Voice-first computing arrives through orchestration, not hardware. The unlock is pipelines and browser-control layers driven by systems that hold business context — not on-device CPUs, local models, or a new class of gadget. Wearables win as capture-and-return endpoints precisely because the intelligence is not local. | This entry | December 2028 |
| 008 | The device stops determining the capability. What an operator can do from a watch converges with what they can do from a laptop for the operating tasks above, because capability becomes a property of what the business is connected to rather than what the person is holding. | This entry | December 2029 |
| 009 | Software built only for human clicking becomes a liability. Platforms whose only control surface is a human-operated dashboard get displaced by platforms that can be driven by instruction — the same displacement pattern as the agency model, one layer down in the stack. | This entry | June 2029 |
What would prove me wrong
A prediction without a failure condition is a slogan, so here are mine.
- Call 006 is wrong if, in 2029, operators still reach for a keyboard by default for routine changes — not because voice is unavailable, but because it is not trusted or not faster in practice.
- Call 007 is wrong if the capability lands primarily as on-device intelligence, and the meaningful difference between operators becomes which hardware they bought rather than which systems they connected.
- Call 008 is wrong if a real capability gap persists between wearable and laptop for these tasks in 2029 — if the small screen turns out to be load-bearing for approval and judgment rather than incidental.
- Call 009 is wrong if click-only platforms hold their position because switching costs and habit matter more than interface speed. This is the one I consider most likely to bite me, because I have underestimated inertia before.
And the honest hedge on the internal bet: our own one-year premise was aggressive. Some of it landed and some of it is still being built. The architecture was the right call. The timeline was a stretch, and I would rather say that here than let it quietly disappear.
Where this is heading
Every interface shift has looked the same from the inside. The old input feels normal until the new one becomes available, and then the old one feels like an obstacle nobody had named. Command lines to windows. Keyboards to touch. Touch to speech.
The businesses that win the next stretch will not be the ones with the most screens. They will be the ones whose systems can be operated at the speed of thought, from wherever the thought happens.
Written down, dated, and scheduled for grading. The hub for every entry is The Prediction Ledger, the reason the dates exist is here, and the long-form version of where I think all of this goes is chapter twelve of Hands Up.