Opportunities › Local-first AI agent desktop applications › Proposed app
Hearthdesk
A private AI agent workspace that runs on your device with the models you choose
- Score
- 48
- confidence 41
- Evidence
- 5
- items on the page
- Rivals
- 1
- named in sources
- First version
- 50 d
- estimate
- Type
- Desktop app
- PWA: runs in the browser, installs on Mac and Windows
- Market since
- 2026
- New market · AI estimate
ScreensConcept
Desktop app: AI agent workspace for filing, chats, memory, and permission logs; users interact with agents on their own device for sustained work sessions. Sample screens on a computer, in the proposed brand. Their numbers are sample data.
New task
Ready to run- TaskSort invoices
About this appAI proposal
Hearthdesk is an installable AI agent workspace that keeps your files, chats and agent memory on your own device. Connect a model running on your computer, a model in your browser, or a provider you pick, and switch between them per task.
It is for developers, knowledge workers and small teams who want agents to read documents, draft, sort and run routine jobs without sending everything to a cloud service or being tied to one model maker.
Every step an agent takes appears in a plain activity log. Agents ask before they touch a file, and you can read, edit or remove each skill they have learned.
- Tagline
- Your agents, your models, your machine.
- Voice
- CalmPlainTrustworthy
- Why this name
- A hearth is a place at home where work gets done close to you. It suggests warmth, ownership and data that stays in the house.
In the first version
Members say which features they would use. It tells PWAS what people want; it changes no score.
Local by default
Chats, memory and documents stay on your device. Nothing leaves it unless you connect a remote model and allow it.
No member yet
Bring any model
Connect a local model server, an in-browser model or your own API key. Set a different model for each agent or task.
No member yet
Readable skills
Agents save what they learn as short skill files you can read, edit, disable or export. No hidden prompts or memory.
No member yet
Permission prompts
Agents ask before reading a folder, writing a file or calling a remote model. Grants can be one-time or saved per agent.
No member yet
Brand identityAI proposal
Check the name against trademark registers before using it.
Palette
#B4532AHearth Orange · primary#F6F1E7Linen · surface#23201BCharcoal · text#2F6B5EPine · accent#D9A441Amber · data highlight
Typography
Aa
Hearthdesk
ABCDEFGHIJKLMNOPQRSTUVWXYZ 0123456789
Your agents, your models, your machine.
PricingAI proposal
The material gives no competitor prices. These tiers follow the monetization idea in the opportunity (premium features, commercial use, enterprise licensing) and are a starting point to test.
Personal
Free
- Local and bring-your-own models
- Activity log and permissions
- Up to 5 custom skills
Pro
$12 a month
- Unlimited skills and agents
- Skill pack import and export
- Commercial use
- Encrypted backup and device sync
Team
Contact us
- Shared policy file
- Approved skill library
- Audit log export
- Setup support
Would you pay Pro · $12 a month?
Your answer tells PWAS which prices people accept. It changes no score.
No answers yet.
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 export skills or skill packs as files or links and post them in developer and privacy communities. Recipients open a preview page that shows what the skill does and can install it in their own Hearthdesk.
- Do they become customers?
- Recipients who install a skill are already likely users, but a preview page alone is just exposure. Test whether preview-page visitors install the app and whether any move to Pro.
- First customers
- Independent developers and small consultancies of 2 to 10 people who handle client documents and cannot send them to cloud AI tools.
- Where to reach them
- Open-source communities: Post in GitHub discussions and forums around local models, with a free skill pack and a demo of the activity log.
- Privacy-focused forums: Share a short guide on running a document-sorting agent with no network access.
- Direct outreach: Contact small consultancies about a policy file pilot and offer free setup in return for feedback.
- First milestone
- Ten paying Pro users or two paid team pilots from small consultancies within 60 days of launch.
Why people already pay for this
- VerifiedGitHub repository gained 392 stars in 215 days
- Public claimThere is an established market for AI agent desktop applications, evidenced by the existence and adoption of proprietary solutions like Claude Cowork.
- AI inferenceThe repository has active adoption and community interest, with 392 stars and 90 forks indicating user demand for this type of tool.
- Public claimUsers value privacy and local execution, as the project emphasizes 'local-first' and 'privacy-first' as core features in contrast to cloud alternatives.
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 · In-browser inferenceWebGPU model runtimeLets small models run inside the PWA with no server and no install beyond the app.
- This app · Local model bridgeConnector to local model serversLets the PWA call models already running on the user's computer or network.
RisksAI inference
- PlatformBrowser limits on file access, background work and GPU memory may make agents less capable than a native desktop app, especially on Safari and mobile.
- TechnicalLocal models vary in quality and tool use. Agents may fail on multi-step jobs unless tasks are kept narrow and tested per model.
- CommercialFree open-source desktop projects already cover this need. People may not pay for a PWA version unless the policy and skill-sharing features matter.
- AIImported skills could carry prompt-injection or unsafe instructions. Permission prompts and a sandbox must hold up against them.
- CommercialThe material shows interest in the form of repository stars, not willingness to pay. Pricing is unproven.
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.