Eido wants to merge 
Tested on a Windows 365 Cloud PC
Clicks through the change in Edge or a desktop app, with screenshots on the work item.
Product How Eido Build works,
How Eido Build works,
inside Azure DevOps
From work item to pull request, within your budgets and approvals.
Agentic engineering,
inside Azure DevOps.
Work comes in from Azure Boards and goes back to Azure Repos. Nothing new to roll out.
Assign it to Eido
Assign a work item to Eido, like anyone on the team, or press Send to Eido. Or put it on autopilot.
Your workflow, your gates
Your stages, the model and budget for each, and where a person approves.
Built, tested, and shown
Agents build on a branch, run your tests, and try the change on a Windows 365 Cloud PC. Watch it live.
A pull request on Azure Repos
Linked to the work item, with each stage’s summary. Your pipelines and branch policies decide what merges.
Review comments land on the line, with a fix you can apply.
Where Eido lives in Azure DevOps
- Work items The Eido tab
- Boards Runs, costs, outcomes
- Pull requests Reviews and /lesson
- Project settings Workflows and policy
- Organization settings Models, budgets, audit
Access follows your Azure DevOps permissions. No Eido roles to manage.
A budget on every run.
A price on every work item.
Every model call priced against its work item. No run can overspend.
Four budgets. The tightest one wins.
- Contoso, this month $1,158 left of $3,000
- Contoso Billing, this month $396 left of $1,500
- This run $36.14 left of $40.00
- Its Build stage $12.90 left of $15.00
Cap handed to the agent $12.90
Owners are alerted at 50%, 80% and 100%.
-
Never goes over
Every model call is checked against the cap before it is sent.
-
Paused, not failed
Raise the budget and it carries on where it stopped.
-
Waiting is free
At an approval, the sandbox is handed back.
-
Every call priced
At Azure list prices, less your discount.
One governed route to AI,
for every engineer
The same gates, models and limits for every team. All of it audited.
-
Gates your groups own
Your Azure DevOps groups approve. Eido never can.
-
Who may send work
An Azure DevOps group decides who can assign to Eido.
-
Approved models only
Allowed for the organisation, narrowed per project.
-
Agents on a short leash
Tool and command limits. Pushes only to their own branch.
-
No secrets in the sandbox
Credentials stay in Azure Key Vault.
-
An audit log that only grows
Every call, approval and change. Export it to Sentinel.
Agents that keep watch, test and learn
Always-on agents, finding bugs and ideas
They read Zendesk, ServiceNow, Application Insights, Sentinel and more, and raise work items with the evidence attached. Your board is the inbox.
Read-only, each with its own model and budget.
Code review in the pull request
One comment per finding, on the line, with a fix you can apply.
Lessons your team approves
Reply /lesson. Once your team approves it, every later run follows it.
Built on what your teams already govern
It runs in your Azure subscription. Built for the Microsoft stack, not by Microsoft.
Azure DevOps
Its only interface, built from Azure DevOps’ own components.
Microsoft Foundry
Your own model deployments, chosen per stage.
Microsoft Entra ID
A managed identity. No passwords or keys.
Windows 365 for Agents
Cloud PCs for testing, watched live.
Azure Key Vault
Every secret in your vault. Agents never see one.
Log Analytics and Sentinel
An optional copy of the audit log, for your SIEM.
Questions
Do our engineers have to learn a new tool?
No. It is an Azure DevOps extension, and Azure DevOps is its only interface.
Does it work with GitHub?
Today it works with Azure Boards, Azure Repos and Azure Pipelines. If you also have code on GitHub, talk to us.
Where does it run?
In your own Azure subscription, with models from your own Microsoft Foundry. Each model call records where it was hosted.
Which models can it use?
Any your organisation deploys in Foundry and approves, such as OpenAI, Claude, DeepSeek, Mistral and Llama, chosen per stage.
We already use GitHub Copilot. Do we have to choose?
No. Copilot helps in the editor; Eido takes work items to pull requests. It reads your copilot-instructions.md too.
What does it cost?
From $30,000 a year, for a minimum of 50 users at $600 a user. A user is anyone who orchestrates tickets with Eido: developers, QA and product.
Put AI to work in Azure DevOps
Your process, your approvals and your budget come with it.