On-scene capture, fast story drafts, working against deadline
The shape of the day
Journalism has always been a job of two halves: being somewhere, then writing it up. The gap between those two activities is where most of the friction lives - and where the job has gotten harder.
Newsrooms are smaller than they were. Pew Research puts U.S. newsroom employment 26% lower in 2020 than in 2008 - newspaper newsrooms specifically lost 57% of staff over that period. Nearly 15,000 media jobs were cut in 2024 alone. The reporters still standing are producing more: a Columbia Journalism Review survey found 46% of small-market journalists were filing more stories than two years prior, in newsrooms that had already shrunk. Nearly two-thirds work more than 40 hours per week, and 79% regularly work outside standard hours.
The smartphone has reshaped fieldwork. The "mobile journalist" isn't a trend anymore - it's the baseline. Reporting trips that used to require a separate recorder, notebook, and camera now run entirely off one device in a pocket. That phone is also how you file, how you edit, how you stay connected to the desk. The affordances are real. So are the limits: trying to compose anything substantial on glass is a constraint that doesn't disappear however fast you swipe.
The BLS characterizes the journalist's working state bluntly: "pressure or stress when trying to meet a deadline or cover breaking news" is a structural feature of the job, not an occasional event. Academic researchers have described journalism as "institutionalized heuristic decision making" - fast thinking isn't an accident of culture, it's the actual operating mode. You're not slow-crafting prose. You're making a hundred small decisions at speed, in sequence, with a hard stop at the end.
Where typing slows you down
There are three places in a typical reporter's day where the keyboard (or lack of one) costs the most time.
In the field, mid-scene. You're standing at a protest, outside a courthouse, walking through a neighborhood. You're watching something. You have thirty seconds before the moment passes. Opening an app and tapping out a sentence means stopping, looking at the screen, losing peripheral awareness of what's happening around you. The observation you meant to capture becomes a three-word fragment that makes no sense when you read it two hours later.
The re-entry write-up. You've spent four hours out. You come back with fragments in a notes app, a voice memo you can't remember the context for, and impressions you haven't written down yet. You know what you saw. Getting it into draft form requires reconstructing context that was completely obvious at the time and is now fading. This is where the clock starts running.
The rough-to-ready gap at the laptop. Story drafts are rarely written once. There's a rough version, then a round of restructuring, then a line-edit. The time-consuming part isn't producing the first sentences - it's getting enough connected text on screen that you can see the story's shape and fix it. Dictating a rough first pass can give you that scaffold faster than typing it cold.
The keyboard constraint is different at each stage, but it adds up across the day. Typing speed data from an internal New York Times study of reporters and editors ranged from 35 to 134 WPM - a 4x variance within a single professional group. Even at the high end, speaking runs at about 150 WPM (per the National Center for Voice and Speech). On a phone, you're working with a fraction of that.
How voice fits this workflow
Voice isn't useful for everything in a journalist's day. The clearer you are about where it fits, the less time you waste trying to force it where it doesn't.
Where it helps:
Field observation notes. Standing somewhere, noticing something - speak it. A sentence or two. What you smell, what the crowd looks like, what the building's state is, what the source said as they walked away. These don't need to be polished. They need to exist, with context, so you can use them later. Voice capture is faster than typing and keeps your attention on the scene rather than the screen.
Quick sourcing notes. After a call or a brief encounter - not a formal interview, just a moment of contact - speak a few lines to yourself. Who it was, what they said, what you still need to verify. Before you've had time to forget the details.
First-draft scaffolding at the laptop. You know the story. You know the lede, the nut graf, the ending. Speak it out. AICHE removes the filler words, adds punctuation, and hands you a rough document. It's not a finished draft - nothing is, after one pass - but it's material to work with rather than a blank page.
Capturing a thought mid-edit. You're reviewing your draft and you see what the next paragraph should do. Before you find the cursor and start typing, press the hotkey and say it. Two sentences. Then go back to editing.
Where voice is not the right tool:
Verbatim quotes. Don't use AICHE to capture a source's exact words. The AI cleanup removes filler and smooths speech, which is exactly what you don't want when accuracy to the source's phrasing is legally and professionally essential. Use a dedicated recorder for anything you'll quote directly.
Interview transcription. If you're recording a full interview and want a searchable transcript, that's a different problem requiring a different tool (more on this below).
On-record attribution capture. Any statement you might use attributed in print should be on a recorder where you control the unedited file. AICHE's cleanup pipeline is good for your notes; it's not appropriate for a record you might have to defend.
AICHE specifically for journalists
These are the features that matter most for the reporter's workflow, in plain terms.
Capture-first mobile app. On iPhone, iPad, and Android, AICHE is built around the idea that you're recording a memo to process later. Tap the mic, speak, save. The note lands in AICHE cleaned up and searchable, not as a raw audio file you have to replay. This fits the field workflow: you're capturing in the moment, processing at the laptop.
Offline recording with automatic queue. Coverage often happens in places where connectivity is bad or absent - basements, rural areas, transit tunnels, venues with spotty service. AICHE saves any recording made offline to an encrypted local queue. When signal returns, the queue processes automatically. You don't have to remember to retry it. The observation you captured in a parking garage with no signal shows up clean on your laptop a few minutes later. The technical details: audio is encrypted on disk, round-trips to Groq for transcription when connectivity allows, and is discarded immediately after processing, within 1 second.
AI cleanup, not raw transcription. Raw voice recognition produces garbage that takes time to edit: filler words, stutter, phantom phrases, missing punctuation. AICHE runs a post-Whisper cleanup pipeline - filler word removal, hallucination filtering, punctuation and paragraph placement, custom vocabulary enforcement - and the result is text you can actually work with. A 10-minute field observation note comes back in about 3 seconds, not 40 seconds, and it's already close to usable prose rather than a transcript you need to clean by hand.
Custom vocabulary. Source names, neighborhood names, organization acronyms, local terminology - teach AICHE these once (up to 50 entries) and they get spelled correctly every time. If you cover a beat where the same proper nouns appear constantly, this pays off immediately.
Cross-device sync. What you capture on your phone at 2pm should be on your laptop by the time you sit down at 4pm. AICHE's sync is end-to-end encrypted (you set the passphrase; AICHE cannot read the content). Notes move automatically in the background. The workflow is phone captures in the field, laptop drafts at the desk, and no manual transfer step in between.
Apple Watch recording. For situations where pulling out a phone is awkward or conspicuous, you can tap your wrist and speak. The recording transfers to iPhone and then syncs to your other devices. Useful when you want to capture a quick note without the social signal of raising a phone.
Desktop hotkey for laptop drafting. On Mac or Windows, Ctrl+Alt+R (or ⌃+⌥+R on Mac) starts recording in whatever app has focus. Press again to stop. The cleaned-up text inserts at your cursor. At the laptop drafting stage, this means you can dictate a rough paragraph, let AICHE clean it, and keep working without switching apps or contexts.
Honest tradeoffs
AICHE doesn't fit every part of the journalism workflow. These are the places where it's the wrong tool.
Not an interview transcription tool. If you recorded a 45-minute interview and want a searchable transcript for quotation pulling, use Otter.ai or Trint. These are purpose-built for that job - timestamped transcripts, speaker separation, word-level search. AICHE is not designed for that workflow and does not produce that output.
Not for verbatim quote capture. The AI cleanup pipeline is a feature for your notes, but it's a problem for anything you need to be verbatim. AICHE's post-processing removes filler words and smooths speech. That's exactly wrong for capturing a source's quoted words. Use a recorder where you control the unedited audio file.
Source confidentiality requires a cloud round-trip. AICHE's transcription goes through Groq, a named cloud provider. For sensitive sources - anyone you're protecting by not naming, anyone in a situation where their voice being on a server even briefly is a risk - this is not the right tool. The audio is discarded immediately after processing, within 1 second, and the sync is end-to-end encrypted, but "quickly" and "encrypted" are not the same as "never left the device." For situations that require true local-only processing, AICHE isn't it. Know the difference between your observation notes (low risk, cloud is fine) and your source-protection workflows (evaluate separately).
Mobile is capture-first, not an inline keyboard replacement. On iPhone and Android, AICHE works as a memo-recorder that produces cleaned text. It's not a replacement for the iOS keyboard dictation button or Gboard's voice input. If you want to tap into a notes app and dictate inline text character-by-character, use Apple Dictation for that. AICHE's mobile model is: record, process, result lands in AICHE - then you copy it where you need it.
Cloud transcription requires a connection to process. The offline queue captures the recording, but cleanup doesn't happen until connectivity returns. If your deadline is in 20 minutes and you're underground, you're waiting. Plan around this: get above ground, or use wifi before the clock is critical.
Desktop UI is English only. If you report in a language other than English, AICHE accepts voice input in 99 languages and transcribes accurately - but the app interface on desktop is English. Mobile apps are available in 28 languages. This is only a practical issue if you need to navigate the desktop app in a non-English language.
What to try first
You don't need to overhaul anything. Pick one part of your existing workflow.
Experiment 1: Capture three field observations on your next assignment. Before you put the phone away after any moment you'd normally try to type a note, open AICHE instead. Speak two or three sentences describing what you're seeing, what was just said, what the atmosphere is. Do this three times in one outing. When you sit down to write, read back what came through. The observation density compared to typed notes is usually obvious.
Experiment 2: Dictate a rough lede before you start typing. On your next story with a clear angle, before you open the draft document, record 30-60 seconds saying what the story is and how you'd explain it to someone. Let AICHE clean it up. Use that output as the starting material for your lede, even if you rewrite it entirely. The act of saying it out loud often surfaces the framing faster than staring at a blank document.
Experiment 3: After-call sourcing notes. The next time you hang up from a source call, immediately record a 90-second voice note: who you spoke to, the two or three things they said that matter, what you still need to check. Speak it before you touch your notes app. See whether the output is more complete than what you'd have typed.
Try AICHE
Seven-day free trial. No credit card required. Personal plan starts at $3.99/mo on annual billing.
See plans and start the trial at /pricing
The trial covers the full feature set. If you're testing this for field capture, try the offline queue on day one - find somewhere with no signal and record a note. That's the case where it earns its place.