Opportunities › Desktop automation orchestration for AI integrations › Proposed app
Deskhand Runs
Run legacy Windows desktop automations as code, with video replay
- Score
- 47
- confidence 57
- Evidence
- 8
- items on the page
- Rivals
- 2
- named in sources
- First version
- 60 d
- estimate
- Type
- Desktop app
- PWA: runs in the browser, installs on Mac and Windows
- Market since
- 2001
- Mature market · AI estimate
ScreensConcept
Desktop app: Integration teams and engineers configure automations, review video replays and debug desktop software runs at a desk. Sample screens on a computer, in the proposed brand. Their numbers are sample data.
Runs
Last 24 hours94.2%
About this appAI proposal
Deskhand Runs lets integration teams and AI agents connect to Windows desktop software that has no API. You write each automation as code, and it runs the same way every time on a managed virtual machine. Triggers come from an API call, a Slack message or a person approving a step.
The platform handles the parts that usually break: cloning machines, queueing, parallel runs and two-factor prompts. Every run is recorded as video with step logs, so a failure shows you the screen at the moment it went wrong.
It is built for AI companies and integrators who must ship a connection to an old system, such as a medical records client, and keep it running. It is not a drag-and-drop tool for office staff.
- Tagline
- Old desktop software, run like an API you can watch.
- Voice
- PlainPreciseCalm
- Why this name
- A deskhand is the helper who works at the desk for you. Runs says the product is about jobs that execute and can be replayed.
In the first version
Members say which features they would use. It tells PWAS what people want; it changes no score.
Automations as code
Define each workflow in a typed script that can be reviewed, versioned and tested. An AI agent can draft it and a developer approves it.
No member yet
Deterministic runs
Steps replay the same way each time. If a screen differs from what the script expects, the run stops and reports it instead of guessing.
No member yet
Video replay with step logs
Every run is recorded. Tap a log line to jump to that moment in the video and see what the desktop showed.
No member yet
VM pools and cloning
Start from one prepared Windows image. Clone it for parallel runs and retire machines that drift.
No member yet
Brand identityAI proposal
Check the name against trademark registers before using it.
Palette
#1F2A33Deep Slate · primary#F6F4EEPaper · surface#161C21Ink · text#E3A020Signal Amber · accent#2C8C84Teal · data
Typography
Aa
Deskhand Runs
ABCDEFGHIJKLMNOPQRSTUVWXYZ 0123456789
Old desktop software, run like an API you can watch.
PricingAI proposal
Verified: UiPath's pricing page lists one paid plan at $25 a month plus a free plan; Free and Team follow it. Scale and Critical have no price in the material and are starting points to test.
Free
Free
- 1 VM, 1 run at a time
- 200 runs a month
- 7-day video retention
Team
$25 a month
- 2 concurrent VMs
- 5,000 runs a month
- Slack and API triggers
- 30-day video retention
Scale
$249 a month
- 20 concurrent VMs
- 50,000 runs a month
- Human-in-the-loop and 2FA
- Failure triage
- 90-day video retention
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
- No loop in the product. Reaches customers through sales, developer content and example automations. Runs happen on customers' own back-office systems, and no output is shown to outside people.
- Do they become customers?
- Not a viral product. Test whether developer-led content and a free tier convert integrators into paying teams, and whether one customer's first win leads to more systems.
- First customers
- Small AI companies and integration agencies of 2 to 20 people that must connect to Windows-only healthcare software with no API.
- Where to reach them
- Hacker News and developer forums: Post a worked example of automating a legacy Windows app, with a run replay. Reply to threads on desktop RPA pain.
- Direct outreach: Contact AI startups building healthcare agents and offer to build their first legacy integration with them.
- Open example library: Publish sample automation scripts and a free-tier demo so developers can try a run in minutes.
- First milestone
- Five paying teams in healthcare integrations, each running at least one live automation for 30 days.
Why people already pay for this
- VerifiedHacker News post received 105 points and 64 comments
- AI inferenceDesktop RPA projects commonly fail at rates exceeding 30%, and at scale can generate thousands of support tickets monthly due to cascading failures and lack of observability.
- Public claimBuilding desktop automations at scale is extremely difficult across three dimensions: scripting complexity, orchestration infrastructure, and debugging and observability.
- AI inferenceA customer was willing to commit to a contract contingent on solving a desktop RPA integration with a legacy medical records system, indicating high business value for solving this problem.
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 · Windows VM hostingCloud Windows VMs with image cloningRuns need real Windows desktops that can be cloned, pooled and shut down per job.
- This app · Job queueDurable queue and worker schedulerNeeded for retries, concurrency limits and pausing runs for a person.
RisksAI inference
- TechnicalKeeping desktop runs deterministic on varied legacy apps is hard. Pop-ups, slow screens and updates may still break scripts, so the failure rate may not drop as hoped.
- CommercialUiPath and Blue Prism are established. A small team must win a narrow niche of AI companies and integrators before competing on breadth.
- RegulatoryMedical records and similar systems carry privacy rules. Video recordings may capture patient data, so redaction and data handling may be required before sales.
- PlatformSome legacy software licences forbid automation or running on cloned VMs. Customers may be blocked by vendor terms.
- AIAgent-written scripts may be wrong in subtle ways. Human review and test runs are needed before they touch live systems.
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.