Opportunities › Voice-first local AI assistant for knowledge work › Proposed app
Hushdesk
Private voice help for your documents, email and calendar, on your device
- Score
- 43
- confidence 54
- Evidence
- 6
- items on the page
- Rivals
- 2
- named in sources
- First version
- 45 d
- estimate
- Type
- Desktop app
- PWA: runs in the browser, installs on Mac and Windows
- Market since
- 2025
- New market · AI estimate
ScreensConcept
Desktop app: Knowledge workers use voice assistance while working through documents, email drafts, and calendars at a desk, requiring hands-free help during focused desk work. Sample screens on a computer, in the proposed brand. Their numbers are sample data.
Contract review
Nothing leaves this deviceListening
About this appAI proposal
Hushdesk is a voice assistant for people who work in documents, email and calendars. Open a file, a draft or a day's schedule in the app, then ask about it out loud: summarise this page, reply to this email, move the meeting. Speech recognition and the language model run on your own device, so nothing you say or open is sent to a cloud service.
It is made for professionals and for people outside tech who have not found voice assistants useful. There is no prompt writing and no typing needed. Hushdesk knows what you have open and what you have selected, so short, natural requests work. Choose from several speech and language models to fit your machine and language.
Unlike cloud assistants, it works offline and keeps your work private. Unlike a screen-reading desktop tool, it works through files and items you bring into it, so it runs in a browser tab on any laptop, and you decide what it can see.
- Tagline
- Talk to your work. It stays on your device.
- Voice
- calmplainprivatehelpful
- Why this name
- Hush says quiet and private: a helper you speak to softly, that keeps your words to itself. Desk says everyday work. Two plain words, easy to say aloud.
In the first version
Members say which features they would use. It tells PWAS what people want; it changes no score.
Hands-free, voice-first
Hold a key or say a wake phrase, then speak. Answers come back as text and read-aloud speech, so your eyes and hands stay on the task.
No member yet
Runs on your device
Speech recognition and the language model run locally in the browser. Works offline once models are downloaded. No audio or file leaves your machine.
No member yet
Knows what you have open
Ask about "this paragraph" or "that email". Hushdesk uses the document, thread or selection you have in focus as context.
No member yet
Documents, email and calendar
Import PDFs, text, notes, mail files and calendar files. Ask questions, draft replies and propose schedule changes, all by voice.
No member yet
Brand identityAI proposal
Check the name against trademark registers before using it.
Palette
#2F6B5EPine · primary#F6F1E7Oat · surface#1F2A2EInk · text#E0883AAmber · accent, listening state#4D6275Slate · data, secondary text
Typography
Aa
Hushdesk
ABCDEFGHIJKLMNOPQRSTUVWXYZ 0123456789
Talk to your work. It stays on your device.
PricingAI proposal
The only price in the material is a public claim of $5 for a compiled version of a similar tool. No competitor pricing page is quoted. These tiers are a starting point to test, not a verified market price.
Free
Free
- Small speech and language models
- 20 voice requests a day
- Documents and text editing
Desk
$5 once
- No daily limit
- Email and calendar import
- Choice of several models
- Markdown export
Pro
$6 a month
- Larger, more accurate models
- Priority support
- Sync of settings across devices
- Early access to new features
What it could be worthEstimate
What a software business making this much sells for today, from real deals. Pick a monthly revenue.
If it makes
Worth about
$194,000
Sale range $132,000 to $263,000
$120,000 a year in revenue
From 87 software businesses listed and sold on Empire Flippers in the last year, revenue vetted by the broker: they sell for about 19× their monthly revenue, most between 13× and 26×.
Measured on profit, the middle deal is 32× monthly net profit: about 2.7 years of profit.
An estimate, not an offer or financial advice. A real price depends on profit, growth, churn, how much of the work is the owner's, and the buyer.
How it would growAI proposal
How the product would reach new customers through its own use, and how the first ones would be found. A plan to test, not a result.
- Built-in loop
- Shared outputs. Users send replies, summaries and meeting notes made by voice. An optional footer line, "Drafted with Hushdesk", links to the app. Users can switch it off, so the effect is modest.
- Do they become customers?
- Recipients of a shared reply or summary only see a small footer; most are not the buyer. Test whether footer clicks lead to installs among colleagues who also draft in voice, not just visits.
- First customers
- Independent consultants and small-firm professionals who handle confidential client documents and email, and want hands-free help without sending files to a cloud service.
- Where to reach them
- Hacker News and Show HN: Post a demo of a fully offline voice session on a PDF and an email draft. Ask for feedback from people who use local models.
- Independent consultant communities: Offer the free tier in forums and newsletters for consultants and lawyers, who care about client privacy.
- Notes-app user groups: Share Markdown export workflows in outliner and note-taking communities.
- First milestone
- Fifty free installs and ten paid $5 purchases within eight weeks from one community, with at least five weekly active users.
Why people already pay for this
- VerifiedHacker News post received 21 points and 8 comments
- Public claimUsers outside tech circles still do not use voice AI assistants much, indicating low adoption despite technical feasibility
- Public claimPriced at five dollars for compiled version, indicating monetization model
- Public claimProduct adds value through context awareness of user's screen, documents, and active applications
- Public claim
Tech stack
Every proposal builds on the same PWA base, installable from the browser on any phone or computer. Dashed boxes are what this app adds.
- AppNext.js and TypeScript
- PWAManifest and service worker
- UITailwind CSS
- Data and authSupabase Postgres
- PaymentsStripe
- EmailResend
- HostingVercel
- This app · On-device speech to textOpen speech model run in browser via WebAssembly/WebGPUTranscribes locally so audio never leaves the device and the app works offline.
- This app · On-device language modelSmall open-weight model run in the browser with WebGPUAnswers questions and drafts text locally. Several sizes let users match their hardware.
RisksAI inference
- PlatformA browser app cannot read other applications' screens. The product depends on imported files and focused items, which may disappoint users who expect full desktop awareness.
- TechnicalRunning speech and language models in a browser needs WebGPU and several GB of downloads. Older laptops and some browsers may be slow or unsupported, hurting first impressions.
- CommercialEvidence says people outside tech rarely use voice assistants, and the signal is small (21 points, 8 comments). Demand beyond early adopters is unproven.
- AISmall local models can misread a request or draft wrong text. Acting on email and calendar by voice needs read-back confirmation, and mistakes could erode trust.
- CommercialA $5 one-time price brings little revenue per user and may not cover support. Free local alternatives made by hobbyists may appear.
Everything marked AI proposal or AI inference was written by a model from the evidence on the opportunity page. The proposed name must be checked against trademark registers before it is used.