devvy
Log inGet started
Firecracker VM infrastructure

Turn messages into
production code.

Ask Devvy from Slack. Get a pull request back. Ephemeral Firecracker VMs handle the work — isolated, fast, and safe.

See how it works
devvy/demo·No credit card required
How it works

From chat to
merged PR.

Every request runs in a clean, isolated Firecracker VM — no shared state, no contamination between tasks or tenants.

01

Request work

Ask Devvy to implement a feature or fix a bug directly from your team's chat. No tickets, no context switching.

SlackTeamsDiscordWhatsApp
02

Devvy plans the work

Devvy analyzes the repository and generates a step-by-step plan before touching a single file.

Plan
1.Update login handler
2.Add retry logic
3.Add integration tests
03

Ephemeral VM executes

Devvy spins up an isolated Firecracker microVM to safely execute the task. The environment is destroyed after execution — no residual state.

$ firecracker boot
✓ VM ready — 182ms
isolated · ephemeral · destroyed on exit
04

Pull request created

Devvy opens a pull request and responds to review comments automatically — no hand-off needed.

GitHubGitLabAtlassian
Any Repo. Anywhere.

Works with the tools
your team already uses

Connect Devvy to any Git repository or messaging platform. No migration, no new tools — just results.

Messaging
Slack
chat trigger
Microsoft Teams
chat trigger
Discord
chat trigger
WhatsApp
chat trigger
D
Devvy
orchestration layer
Repositories
GitHub
PR + review
GitLab
MR + review
Atlassian
PR + review

More integrations coming soon. Request yours at hello@devvy.dev

Run Dozens of Tasks at Once

Built for parallel
development.

Run multiple Devvy workflows simultaneously without blocking your team. Devvy spins up isolated environments for every task — automatically.

Active workflows 0 running
#engineeringSlack
alex.kim
@devvy ▋
#backendSlack
priya.s
@devvy ▋
#dev-opsDiscord
tom.w
@devvy ▋
#platformTeams
sarah.l
@devvy ▋
D
Devvy
auth-service182ms
billing-service178ms
api-gateway191ms
mobile-backend174ms
+18 more workflows running
Devvy spins up isolated Firecracker microVMs for every task.
Run 1 workflow or 100 — without slowing down your team.
Secure by Design

Secure by
default.

Security is the foundation — not an afterthought. Designed for teams that cannot afford a breach.

Powered by Firecracker — AWS Lambda's VMM

Isolation

Every task runs inside a dedicated Firecracker microVM — not a container, not a sandbox. Hardware-level separation between every task and every tenant.

Ephemeral environments

VMs are automatically destroyed after each execution. Zero residual state, zero credential exposure. No attack surface persists between jobs.

Audit logs

Every trigger, every run, every result — logged with timestamps and caller identity. Full transparency for compliance and incident review.

Least privilege

Devvy only receives the repository access it needs for the task at hand. No standing permissions, no broad OAuth scopes.

Customize Devvy

Devvy adapts to your engineering workflow.

Configure how Devvy plans, executes, and reviews code using a devvy.yaml file in your repository.

devvy.yaml
1version:1
2 
3workflow:
4  planning:
5    require_plan_approval:true
6 
7  execution:
8    environment:firecracker
9    install_dependencies:true
10 
11  linting:
12    run:true
13    tools:
14      - eslint
15      - prettier
16 
17  testing:
18    run:true
19    command:pytest
20 
21  pull_request:
22    reviewers:
23      - backend-team
24    auto_respond_to_comments:true
Execution flow
Plan
Waiting for approval…
GATE
Execute task
Firecracker VM · isolated env
Run linting
eslint + prettier
Run tests
pytest
Open pull request
backend-team notified
Respond to review
Auto-addresses feedback & pushes
Devvy executes workflows as a finite state machine — ensuring deterministic, observable automation.
⌥
Linting & formatting
Run ESLint, Prettier, Black, or any custom tool.
✓
Test execution
Define test commands that must pass before a PR is opened.
⏸
Approval gates
Require human sign-off before Devvy implements changes.
↩
Review responses
Automatically address pull request feedback and push updates.

Every repository can define its own Devvy workflow.

Bring Your Own AI

Bring your own model.

Use OpenAI, Anthropic, or your own API keys. Devvy orchestrates the workflow while you control the models.

Devvy orchestrates the workflow — you control the models.
Switch providers anytime without changing your workflow.
API keys are encrypted at rest and never logged.
Works with any OpenAI-compatible endpoint, including local models.
OpenAI
GPT-4o, o1, o3
Anthropic
Claude 3.5, Claude 4
Your own API keyBYOK
Any OpenAI-compatible endpoint
More providers supported on request
Devvy — live run
Try Devvy Live

See Devvy in action.

Trigger a real Devvy workflow on our demo repository. Watch it plan, boot a VM, write code, and open a pull request — live.

Runs in an isolated sandbox. No account required.
Pricing

Simple, transparent
pricing.

Starter

For small teams getting started.

$0
Free forever
Start free
  • Up to 3 engineers
  • 50 VM runs / month
  • Slack integration
  • GitHub PR comments
  • Community support
Most popular
Team

For teams that ship every day.

$79/ mo
Start 14-day trial
  • Up to 25 engineers
  • Unlimited VM runs
  • Custom VM snapshots
  • Audit logs
  • Priority Slack support
  • SSO (Google / GitHub)
Enterprise

Custom contracts and SLAs.

Custom
Talk to us
  • Unlimited engineers
  • Unlimited VM runs
  • SAML / SCIM
  • Dedicated support
  • Custom SLA
  • On-prem option
FAQ

Common questions.

Still have questions? Email us

Let Devvy
write the code.
You ship.

No credit card required. Free plan available.

182ms
avg VM boot time
100%
isolated environments
0
context switches