Opportunities › Self-hosted AI coding agents platform › Proposed app
Yardhand
Run coding agents on your own servers, any model, clear cost per task
- Score
- 45
- confidence 53
- Evidence
- 5
- items on the page
- Rivals
- 1
- named in sources
- First version
- 45 d
- estimate
- Type
- Desktop app
- PWA: runs in the browser, installs on Mac and Windows
- Market since
- 2024
- New market · AI estimate
ScreensConcept
Desktop app: Development teams configure agents, monitor task progress, review pull requests, and manage permissions at a desk as part of core development workflows. Sample screens on a computer, in the proposed brand. Their numbers are sample data.
New task
Ready to start- Repoacme/billing-api
About this appAI proposal
Yardhand lets a development team run autonomous coding agents inside its own cloud account or on-premise servers. You assign a task from a ticket or a chat, an agent works in an isolated workspace, and it opens a pull request for a person to review.
It is built for engineering leaders and teams who want agents but need code to stay in their own infrastructure, and want to see what each task costs. You choose the model behind each agent, such as Claude, Codex or another LLM, and can switch without changing your workflow.
The core is meant to be open and self-hostable. A managed option exists for teams that would rather not run it. Spend, logs and permissions are visible to the team instead of hidden behind a seat price.
- Tagline
- Agents that work in your yard, on your terms.
- Voice
- PlainPracticalTransparent
- Why this name
- A yardhand is a helper who works on your own land. It suggests agents running on your own infrastructure, doing steady work you can see.
In the first version
Members say which features they would use. It tells PWAS what people want; it changes no score.
Runs where your code lives
Deploy workers in your cloud account or on-premise. Repos and secrets stay inside your network.
No member yet
Pick the model per agent
Connect Claude, Codex or another LLM with your own keys. Change the model without rewriting tasks.
No member yet
Cost shown per task
Every run lists model tokens, compute time and total cost, with a monthly cap per team or project.
No member yet
Isolated task workspaces
Each task gets a fresh container with only the repo access you grant. It is deleted when the task ends.
No member yet
Brand identityAI proposal
Check the name against trademark registers before using it.
Palette
#2F5D46Pine · primary#F6F1E7Oat · surface#1F2A24Soot · text#D9822BMarigold · accent#3B6E8FSlate blue · data
Typography
Aa
Yardhand
ABCDEFGHIJKLMNOPQRSTUVWXYZ 0123456789
Agents that work in your yard, on your terms.
PricingAI proposal
Devin's pricing page (verified) lists paid plans from $20 to $200 a month, a free plan and prices on request. Cloud and Team tiers mirror that range; Enterprise is contact-us. All are starting points to test.
Community
Free
- Open-source core, self-hosted
- Bring your own model keys
- Unlimited workers you run
- Community support
Cloud
$20 a month
- Managed workspaces, one project
- Per-task cost view and caps
- Run logs kept 30 days
- Usage of compute billed at cost
Team
$200 a month
- Managed workspaces, many projects
- Roles and approval rules
- Log export to your storage
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. Yardhand runs inside a team's own infrastructure, and its output is pull requests in private repos. Little of it is seen by outsiders, so growth comes from open-source visibility, content and direct sales.
- Do they become customers?
- Open-source users who star or install it can become paying cloud customers, but that is untested. Test how many self-hosters convert to managed hosting or support.
- First customers
- Engineering leads at software companies of 10 to 50 developers that restrict where code can be sent, such as fintech or health software teams.
- Where to reach them
- GitHub and open-source community: Publish the core, with a clear install guide and a cost-per-task benchmark against paid agents.
- Developer forums and newsletters: Share write-ups of self-hosted agent setups and real per-task costs on Hacker News and engineering communities.
- Direct outreach: Contact engineering leads in regulated industries and offer a guided pilot on one repo.
- First milestone
- Five teams in one regulated vertical each pay for managed hosting or support after a 30-day pilot.
Why people already pay for this
- VerifiedGitHub repository gained 385 stars in 46 days
- Public claimThere is demand for open-source alternatives to proprietary AI coding agents
- Public claimDevelopment teams value the ability to self-host and control their infrastructure
- Public claimThe market supports multiple LLM backends for coding agents
- Verified
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 · Agent workersContainerised runner for customer serversTasks must run in the customer's own cloud or on-premise, not on our hosting.
- This app · Job queuePostgres-backed queueDispatches tasks to workers and tracks retries without adding another service.
RisksAI inference
- TechnicalSandboxing agents safely on customer servers is hard. A weak isolation or permission flaw could expose code or secrets, and installs across different infrastructure can be costly to support.
- CommercialIf the core is open source, teams may self-host and never pay. Demand for managed hosting and support over the free core is unproven and must be tested.
- CommercialAgent quality depends on the chosen model. If results trail closed products, teams may not switch even at lower cost.
- PlatformModel providers and Git hosts can change API terms, rate limits or prices, which would change per-task costs and how integrations work.
- Intellectual propertyAn existing open-source project already describes itself as an open alternative. Differentiation, licence choice and name clearance need checking before launch.
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.