1,284

You vibecoded the app.Vibeops the rest.

DunOps connects to your stack and runs the deployment work — DNS, certs, rollbacks, credentials — through a chat thread. You describe the outcome. It handles the ops.

thread / connectworkspace ready · 14 connectorsauto-demo
iasimnaveed/docs-sitedocs.dunops.comAdd context
Dun 1.0auto
Watching:

Talks to the stack you already run on

VercelCloudflareAWSGitHubGitLabDatadogSentryPagerDutySlackLinearNotionRenderFly.io1PasswordVaultVercelCloudflareAWSGitHubGitLabDatadogSentryPagerDutySlackLinearNotionRenderFly.io1PasswordVault

Repo to green padlock, in one sentence.

Point DunOps at a repository and a domain. It reads your deploy, creates the DNS record, issues the cert, and verifies HTTPS — in the time it takes to write the message. No tab-switching. No propagation prayer.

The old way
  • Vercel → import → 6 tabs of waiting
  • Cloudflare → A record → refresh, refresh, refresh
  • Re-issue cert → SSH in → curl, curl, curl again
  • Paste the URL in Slack. Pray.
0 tabs
Opened during the run
~15s
Message to green padlock
Connect my docs-site repo to docs.dunops.com
Run plan · run_4f2a
Connect docs-site to docs.dunops.com
Active
Detect framework & build settings
Next.js 14 · pnpm · output: .next
2.1s
Verify DNS authority for dunops.com
Cloudflare zone · 4 nameservers OK
0.8s
Propose record changes
CNAME → cname.vercel-dns.com
Preview · Cloudflare zone
3 changes proposed · 0 destructive
Dry run
OpTypeNameValueTTL
addCNAMEdocscname.vercel-dns.comAuto
addTXT_vercelvc-domain-verify=…Auto
keepA@76.76.21.21Auto
Approval · cloudflare:dns:write
Apply DNS changes & issue certificate
Awaiting you
Will modify dunops.com zone and request a new Let’s Encrypt cert for docs.dunops.com. Rollback is one click — your zone snapshot is taken automatically.
requested · just now

Prod deploys without the clicky-clicky.

Connect your CI and your host. DunOps watches for green builds and promotes on your signal — or on a schedule you set. Every promotion is logged, reversible, and requires no one to find the right button in the right dashboard.

The old way
  • Open Vercel → squint at build hashes → click “promote”
  • Ping someone with prod access. Wait. Ping again.
  • Stare at Datadog for 20 minutes hoping nothing pages
  • Slack “shipped 🚢”. Refresh. Pray.
0 dashboards
Tabs opened to ship
2 approvals
Per prod deploy, recorded
Promote claw-code-parity staging build to production
Gate checks · staging build 9c4a7b1
9 checks across CI, perf, errors, & cost
Running
CI · 248 tests · 0 failures
GitHub Actions · run #4128
2m 4s
Performance · p95 latency 217ms (≤ 300ms)
Datadog · /api/* over 1h
14s
Error budget · 0.04% (≤ 0.1%)
Sentry · last 24h, staging
3s
Cost delta · +$3/mo projected
Lambda invocations · vs last 7d
1s
+ 5 more · linter, types, a11y, bundle size, SBOM
all green
8s
2-key approval · production
Promote 9c4a7b1 to claw-code-parity · iad1
1/2 approved
ix · approved 14:11 waiting on dev or sre on-call
posted to #deploys · 14:10

Pager at 2am? Don't wake up. Just text it.

Describe what's wrong. DunOps reads your recent deploys, identifies the last stable build, and rolls back — or tells you exactly why it can't. No war room. No runbook. No one has to own the 3am shift.

The old way
  • 3 PagerDuty pings · Sentry on your phone, half-blind
  • Squint at Vercel: which one was last green again?
  • Promote the old build. Pray it's compatible.
  • Write the Slack message yourself. Badly. Half-asleep.
10s
From “go” to healthy
Confirmed
Postmortem drafted & signed
Signal · sentry + datadog
api-gateway error rate 4.2% (baseline 0.05%)
Degraded
Spike started 14:02 UTC, ~3min after deploy a91f3b. Suspected cause: changed middleware ordering in src/auth.ts. Suggested: roll back to e210cc8 (last healthy).
Roll back the last deploy of api-gateway and tell Slack why
Rollback plan · 4 steps
Revert api-gateway → e210cc8
Confirm to apply
Promote e210cc8 on Vercel (api-gateway)
prod · iad1 + sfo1
Drain stale nodes & warm cache
/health · /status · /v1/ping
Verify error rate < 0.5% for 60s
Datadog query · auto-monitor
Post incident summary to #incidents
draft + offending diff attached
requested · 14:06 · proposed by signal
New · Playbooks

Save the chat. Re-run it forever.

Any DunOps thread can become a Playbook — a typed, parameterized graph you run from a button, a webhook, or another chat. Same steps. Same approval gates. Same outcome, every time.

  • A graph you can actually readApprovals, branches, parallel apply — all visible, all editable on the canvas.
  • Parameterized inputs, not hardcoded valuesRepo, domain, region, scope — every step takes inputs at run time, not edit time.
  • Triggered from chat, webhook, or scheduleSlack command, cron, HTTP webhook, or another thread — pick the trigger that fits.
  • Approval gates preserved per stepEach node carries its scope and approver. The audit trail tags along automatically.
playbooks / release-claw-code-parity.pbklive
Trigger
Slack / cron / chat
every Mon · 09:00 UTC
Preflight
Verify build is green
github · vercel
Plan
Compute DNS + cert diff
cloudflare · vercel
Approval
2-key required
on prod scope
Apply
Promote + DNS + cert
parallel · 3 hosts
Verify
p95 latency & errors
datadog · 5m window
Notify
Slack #releases summary
with diff + run link
Integrations

Talks to the tools you already run on.

Connect Vercel, Cloudflare, GitHub, AWS, or 40+ other providers in under a minute. DunOps reads everything. It writes nothing until you approve the plan.

New connectors weekly

Hosting & Runtime

  • Vercel
  • Cloudflare
  • AWS
  • Render
  • Fly.io
  • Kubernetes

Code & CI

  • GitHub
  • GitLab
  • CircleCI
  • GitHub Actions

Observability

  • Datadog
  • Sentry
  • PagerDuty
  • Grafana

Notify & Secrets

  • Slack
  • Linear
  • Notion
  • 1Password
  • Vault
  • Doppler
MCPBring your own connector via MCP.Anything with an HTTP API — internal services, niche providers, your own platform — works in minutes.Don't see your tool? Request it
Alternatives

We're honest about the alternatives.

Scripts break silently. AI assistants hallucinate permissions. IaC is powerful but it isn't a chat. Here's where each one actually sits.

What you need
DunOps
Scripts + dashboards
The status quo
Generic AI
ChatGPT · Cursor · Copilot
IaC
Terraform · Pulumi
Time to first working action
From idea to a real change applied, on day one.
Minutes
Hours, if it works
Code, not action
Days of scaffolding
Understands natural language
Describe the outcome, not the syntax.
Native
Bash flags
Native
HCL / TypeScript
Shows diff before running
DNS, env vars, alias changes — visible up-front.
Every action
Inline maybe
Code diffs only
terraform plan
Enforces approval gates
Apex DNS, prod aliases, credential rotation pause.
Per-scope policy
If you remembered
Not the model's job
Via PR review
Full audit trail
Every plan, tool call, approval, signed and searchable.
Built-in
Shell history
Chat history only
tf state + git
Reusable without re-writing
Same flow, drift-free, parameterized at run time.
Playbooks
If you wrote it
Not a concept
Modules
Manages complex multi-resource state
State files, drift detection, large-scale graphs.
Per run, no state file
DIY
DIY
Core competency
Runs entirely offline / air-gapped
No network, no model calls, no audit upload.
Cloud product
Bash on a laptop
Cloud product
CLI runs local
Who it's for

Built for people who ship a lot.

Not built for DevOps teams with ticketing systems. Built for the people who are the DevOps team.

Solo founders

You're building three things at once. You don't have time to remember which Cloudflare account holds which domain. DunOps holds that context for you.

Small platform teams

Two reviewers on every prod DNS change. One audit trail your team can actually read. No more “wait, who deployed this?” in Slack at 9pm.

Agencies & freelancers

Twelve client workspaces. One interface. No credential sprawl, no missed renewals, no emergency Sunday morning.

Who it’s for

Built for people who ship a lot

Solo founder

I ship 3 side projects a month. Dun keeps them straight.

user · what's live and what's stale

dun · 4 live · 1 paused · stale: weekend-app · indie-news pinged 12d ago

Adi · acme-blog · indie-news · weekend-app · slow-cal

Agency lead

13 client workspaces. One audit log per client, no missing.

user · audit acme · 7 days

dun · 31 changes · 0 unapproved · 2 reverted · CSV ready

Mira · 13 workspaces · audit ready

Startup CTO

Two reviewers required for prod DNS — agent enforces it.

user · drop CNAME api.acme.com

dun · blocked — prod DNS needs 2 approvers · invited @theo

Theo · seed-stage · prod safety on

Pricing

Priced for engineers, not committees.

Start on the free plan. Upgrade when your shipping cadence demands it — not before.

Hobby

$0/ forever

Get started. No card.

  • 1 seat
  • 20 workflow runs / mo
  • 3 connectors
  • Save up to 3 Playbooks
  • Single-key approvals
  • 7-day audit retention
  • Standard model routing
  • Community support
Start free
Most engineers

Pro

$24/ seat / mo

For active shippers.

  • Up to 5 seats
  • 500 workflow runs / mo
  • All connectors
  • Unlimited Playbooks
  • 2-key approvals on prod scope
  • 90-day audit retention
  • Smart model routing
  • Email + Slack support · 1 biz-day SLA
Start Pro

Team

$72/ seat / mo

For teams with standards.

  • Unlimited seats
  • Unlimited workflow runs
  • All connectors + custom MCP
  • Workspace Playbook library
  • Custom approval policies
  • 1-year audit retention
  • Premium model routing + BYOK
  • Dedicated channel · 4-hour SLA
Start Team

Questions about volume or enterprise? Talk to us

FAQ

Questions we've actually been asked.

What's a Playbook, exactly?

A typed, parameterized graph behind a chat thread. Action nodes, decision gates, retry edges, parallel fan-outs — saved, named, and re-runnable from a button, a webhook, or another chat. Eleven ship with every new workspace; you can save your own once canvas authoring lands.

How does DunOps know what to do?

It plans first. Every request becomes a typed graph of steps you can read top-to-bottom before anything runs. Tools are restricted by workspace scope; risky steps (apex DNS, prod aliases, credential rotation) always pause for a human go. No autonomous mutations, ever.

Does DunOps ever store my credentials?

Provider tokens are encrypted at rest and scoped per workspace. We never log, ship to a model, or surface them in chat — they live inside the connector adapter that calls the provider on your behalf. You can rotate or revoke any token from the integrations panel.

What's actually free, and what isn't?

The Hobby plan is free forever: 1 workspace, 1 user, 20 workflow runs / month, 3 connected providers. No card. You hit a paid plan when you bring teammates in, want more runs, or need 2-key prod approvals — not before.

How do I know if my stack is supported?

First-class connectors today: Vercel, Cloudflare, AWS, Render, Fly.io, Kubernetes, GitHub, GitLab, CircleCI, GitHub Actions, Datadog, Sentry, PagerDuty, Grafana, Slack, Linear, Notion, 1Password, Vault, Doppler. If it has an HTTP API, MCP bridges it. If it doesn't yet — request it, we add new connectors weekly.

What if I don't like a change after it's applied?

Every mutating run takes a snapshot first. One click reverts — DNS records, alias swaps, credential rotations, env vars. The rollback is part of the same audit trail, signed and reversible for 24h.

Do you replace my DNS provider, host, or CI?

No. DunOps drives the dashboards you already pay for. You keep your accounts, your billing, your OAuth scopes. We're read-only until you approve a change.

Shipping shouldn’t be the scary part.

You already have an AI that writes the code. Give yourself one that ships it.