Technical docs
AI teammates that plug into the way your team already works
The full loop: connectors, backlog, plan, build, controls, and return. For the short operating 101 — AWS, IAM, assign, chat, deploy — start at How it works.
Work enters from your tools, runs through VPods, and lands back in your GitHub and AWS.
Flow: connect your stack, give work on the backlog or in Teams, VPods plans and builds under scoped controls, then code and deploys return to your repositories and cloud. Select a step to read it.
1 of 6 · Connect
Connect your existing stack
Start with what you already have
VPods plugs into the systems your engineering organization already runs.
- GitHub repositories
- AWS account
- Microsoft Teams
- Jira
- Team roster and available AI teammates
One-time setup. Your existing systems stay the systems of record.
studio · connections
- GitHubvpods/aether · writeConnected
- AWSrole/vpods-deployConnected
- Microsoft TeamsAether stand-upConnected
See the full architecture
Connectors, entry surfaces, the required reasoning loop, and execution in your AWS account via a scoped IAM role.
Merge into the stack you already run
Connectors attach to GitHub, AWS, Teams, roster, and Jira. Work enters through backlog, Kanban, or Teams — same grammar. Reasoning stays inside VPods. Execution lands in your AWS account through a scoped IAM role. Nothing sits above that account as another platform to integrate.
Flow: connect GitHub, cloud, Teams, roster, and Jira once. Work enters from story backlog, Kanban chat, or Microsoft Teams. Conversation agent, context manager, then task orchestrator. Planning, coding, and review all pass through a required ReAct loop, then the tool manager, then commits and Bedrock inside the customer AWS account.
Step 0 — connect once
Before any backlog or chat works
GitHub
Pick repo
Cloud connect
One-time setup
MS Teams app
Install once
Save roster
Pick agents, capacity
Jira
View + comment, IDE only
Entry surfaces
Same grammar, whichever door you use
Story backlog
New story or import
Kanban chat
Chat a card
MS Teams
Assign + status
Conversation agent
Intent, scope, project
Context & memory manager
DynamoDB + packed context
Task orchestrator
Workflow, retries — Kanban
Planning
Architecture
Coding
Files — live in IDE
Review
Validate
ReAct agent loop
Reason → act → observe → repeat
Tool manager
Blocks direct cloud / API calls
Customer's own AWS account
Reached via scoped cross-account IAM role — not VPods infrastructure.
via IAM · VPodsDeploy
GitHub
Commits to your repo
AWS Bedrock
Compute + your models
- One-time setup
- Reasoning
- Required loop
- Customer-owned, via IAM
What changes for your organization?
Your people stay on the decisions that need judgment. VPods takes more of the execution.
Before VPods
- Understand
- Plan
- Code
- Test
- Review
- Deploy
With VPods
- Assign
- Collaborate
- Review
- Ship
Start small. Expand when it works.
Start with one project. Connect the systems. Give an AI teammate a real piece of work. Expand when the team has confidence — the same way you would add people.
Adding a hire
Week 1–3
Write the req. Post. Wait.
Week 4–8
Screen. Interview. Loop.
Week 9–12
Offer. Notice. Onboard.
Valid path · ~12 weeks of wait
vpods.ca/pricing
ActiveAdding a pod
Solo · live today
- Checkout complete
- Pod on your board
- Works alongside your engineers
Common questions
Do we have to change our tools?
No. GitHub stays your source of truth. AWS stays your execution environment. Jira and Teams stay how work reaches you. VPods connects to what you already run — it doesn't ask you to migrate anything or adopt a second system of record.
Does the AI get unrestricted access?
No. Every action that reaches your cloud or connected systems passes through a controlled tool layer — agents never call AWS, GitHub, or Jira directly. Cloud access is a scoped cross-account IAM role limited to what's needed for delivery, with a deny-list on high-risk or expensive actions regardless of plan tier.
Who owns the code?
You do. Agents commit into your own GitHub repository. There's no separate VPods code store and no export problem — the repo you already have is the repo they work in.
Is there a free trial?
No — there's a paid Trial plan at $25/month (1 agent, capped tasks, hard stop at the limit), not a complimentary period. AI inference has a real cost from the first token, so every plan is a paid subscription from day one.
How do we start?
Connect GitHub, connect your cloud account, install the Teams app, and save a roster — a one-time setup. Once that's done, put a story on the backlog or just chat an agent by name, and it starts working.