How I Rebuilt My Work Life From Bed Using Only My Voice
For eleven months, I couldn't figure out how to work while lying down. Early on, everyone had the same advice: "Just use voice input." It took me almost a year to understand why that advice was both correct and completely insufficient.
I'd spent years in corporate HR. My entire career was built on sitting at a desk, writing documents, running meetings, reviewing policies. When a herniated disc made sitting physically impossible, I didn't just lose my job — I lost the ability to do any job the way I'd always done it.
This is the story of how I finally built a system that works from bed, and why it took eleven months instead of eleven days.
Why "Just Use Voice Input" Doesn't Work
Everyone says it. It sounds so simple. Talk into your phone. Words appear on screen. Done.
Here's what actually happens. You dictate a paragraph. The voice recognition gets half the words wrong. Punctuation is random. There are no line breaks. Homophones are misidentified everywhere. What you get isn't a paragraph — it's a mess that needs to be manually cleaned up.
And "manually cleaned up" means tapping on a small phone screen. Selecting text. Backspacing. Rearranging sentences. For someone with a herniated disc who can't sit at a computer, this editing process was physically painful in its own right. I developed calluses on my fingers from scrolling. My neck ached from looking down at the screen for hours.
The input was easy. The editing was hell. And nobody warns you about the editing.
The Eleven-Month Detour
In mid-2025, I tried Google Docs voice typing — widely considered the best free option at the time. The accuracy didn't match my speech patterns. Every dictated paragraph needed extensive correction, and correction meant screen time, which meant pain.
Here's where I made my biggest mistake: I concluded that voice input simply wasn't for me. Instead of questioning the workflow, I went back to forcing myself to sit at a desk. One day of sitting meant two to three days of increased pain afterward. But I was trapped in a belief that felt like a fact: "Real work requires sitting at a computer."
I kept paying for that belief with my body.
What I didn't realize was that I was treating a process problem as a tool problem. I kept searching for a "better voice app" when what I actually needed was to rethink the entire workflow — not just the input method, but everything that happens between speaking and publishing.
The Two Shifts That Changed Everything
The solution came from two changes that, individually, would have been insufficient. Together, they rewired how I work.
The first was lowering my standards. After years in corporate HR producing polished documents, the idea of publishing imperfect text felt deeply unprofessional. But "professional quality with a broken body" was producing exactly zero output. "Imperfect quality from bed" was producing actual work. I decided to publish voice-dictated text without obsessive corrections — raw, real, good enough. The moment I accepted that, the pressure disappeared.
The second was a leap in AI speech recognition. In early 2026, I switched to SuperWhisper running a top-tier speech model. The misrecognition rate dropped to near zero. The thing that had been my biggest source of frustration — spending hours correcting transcription errors — essentially vanished overnight.
Neither shift alone would have been enough. Lowering my standards without better accuracy would have produced unreadable text. Better accuracy without lowering my standards would have kept me in the perfectionism trap. It took both.
The System I Use Every Day
Here's what my daily workflow looks like now, working entirely from bed.
I start with voice capture. Lying on my back, I talk into my phone. Not in polished sentences — just fragments. Ideas, observations, half-formed arguments, stories I want to tell. Whatever's in my head gets spoken aloud. I don't worry about structure or grammar or whether it makes sense. I just talk.
That raw voice text then goes to an AI assistant — usually Claude. I ask it to organize the fragments into something coherent: suggest headings, identify logical flow, point out gaps, arrange the pieces into a structure that a reader can follow.
The AI produces a draft. I review it, make final decisions on tone and content, adjust anything that doesn't sound like me. Done.
| Before | After | |
|---|---|---|
| Voice tool | Google Docs (free) | SuperWhisper + premium speech model |
| Editing | Manual phone screen tapping | AI handles restructuring |
| Quality bar | 100% polished or nothing | Good enough to publish |
| Devices | Phone only | iPhone + MacBook |
| Workflow | Dictate → manually fix → publish | Dictate → AI organizes → review → publish |
This article was made using this exact process.
The key insight is simple but took me eleven months to internalize: voice input alone isn't a workflow. Voice input plus AI processing is.
Why It Took So Long (And What That Teaches)
If someone had handed me this system on day one, I could have started producing work immediately. Instead, I spent nearly a year cycling through the wrong approach. The reasons are worth examining, because they apply to anyone adapting to new physical constraints.
I was solving the wrong problem. "Find a better voice app" was the wrong question. The right question was "how do I redesign the path from thought to published text so that sitting at a desk is no longer required?"
I couldn't let go of old standards. A career in HR had trained me to produce carefully formatted, error-free documents. Publishing anything less felt like admitting defeat. But the real defeat was producing nothing at all.
And I underestimated what equipment investment could do. I eventually bought a MacBook with enough RAM to run local speech-to-text software, and upgraded my phone for a better microphone. During recovery, spending money felt irresponsible. But reframing it as "investing in a workspace that doesn't require sitting" made the decision obvious. I've written about those specific purchases in detail.
This Applies to More Than Herniated Discs
Here's where my background in HR becomes unexpectedly relevant. I'd spent years working on process optimization — helping teams produce the same output with fewer resources, streamlining workflows, eliminating bottlenecks. What I was doing for myself was exactly the same thing: my body was the constrained resource, and I needed to redesign the process around it.
This voice-to-AI workflow works for anyone operating under constraints that limit conventional desk work. Chronic pain or disability that restricts screen time. Parenting, where hands are occupied and time comes in fragments — voice memos during naps, batched for AI processing later. Caregiving, where you're anchored to one location with only short windows of availability. Mental health challenges that make sustained screen focus difficult — short voice bursts are less draining than extended typing sessions.
The people who benefit most from this system aren't beginners. They're people who already have knowledge, experience, and things worth saying — but have lost the physical ability to output through conventional means. Voice input plus AI doesn't create expertise. It unlocks expertise that's trapped inside a body that won't cooperate.
The Part About Philosophy
I studied philosophy in university. Spent years thinking about how to live, what gives work meaning, whether there's an optimal way to exist. I never found answers. I took a corporate job because philosophy doesn't pay rent.
Now, lying in bed with a herniated disc, I'm asking those same questions from a position I never anticipated. And this time I've learned something the classroom couldn't teach: answers come from doing, not from thinking about doing. Eleven months of wrestling with voice input taught me more about adapting to constraints than four years of philosophy courses. Not because the philosophy was wrong — but because the answers only become visible when you're actually moving.
What You Can Try Today
If something is limiting your ability to work the way you used to, here's a five-minute experiment.
Talk into your phone. Don't aim for sentences. Just speak whatever's on your mind — ideas, frustrations, observations, things you know that other people might not. Save the text.
Hand that raw text to an AI. Ask it to organize your thoughts into a structure. You'll likely be surprised at how much coherent, publishable material is buried inside messy dictation.
Then ask yourself one question: what knowledge or experience do I have that I could share with the world, if I had a way to get it out of my head without sitting at a desk?
Eleven months of trial and error went into building this system. Use this article to skip most of that time.
Written by Ryo — years in corporate HR, herniated disc survivor. This article was structured with AI assistance (Claude). All content reflects my own experience and is not medical advice.
Read next:
- I Spent $4,100 on Recovery Gear — Only 3 Things Were Worth It
- I Spent 1,500 Hours Talking to AI From a Hospital Bed