VxCloud
Chat System · for delivery and response

Where ops happens — and stays.

Operations-first chat with channels, threads, AI copilots, and a CLI / TUI front door. Deploy events, alerts, and AI suggestions live alongside the humans deciding — with a full audit trail your auditors actually like.

0 room
shared operator context
0s
deployment updates
0 silos
cross-team gaps
deploy-studio
7 members · production room
LIVE
#deploy-studio · Friday 3:42 pm · 7 members
JW
Joel3:41p

Pushing v2.14.0 to staging now. Anyone with eyes on the canary?

Talks to the model providers and tools your room already runs on

Anthropic
OpenAI
Vertex AI
GitHub
GitLab
Kubernetes
Docker
OpenTelemetry
PostgreSQL
HashiCorp Vault

One platform · four room types

Built for the rooms ops teams actually live in

Not "another Slack." Rooms come with deploy + incident + on-call + support primitives baked in — AI copilots, structured CI events, runbook hooks, and audit chain-of-custody included.

Deploy rooms

Every release gets a thread. Build, canary, and rollback events stream in next to the people deciding.

Incident war rooms

Page-to-mitigate in one channel. AI pulls traces and PRs into the thread; the timeline writes itself.

On-call handoffs

Day-shift drops the tape; night-shift picks it up. AI summarises threads, links to runbooks, and pings the right human.

Customer support copilot

Tier-1 conversations live alongside engineering. AI drafts replies grounded in your docs, ready for one-click send.

#deploy-studioDEPLOY ROOM

Deploy rooms

Every release gets a thread. Build, canary, and rollback events stream in next to the people deciding.

🤖
CI Bot

✓ build 4812 · 6/6 jobs green

JW
Joel

Promoting to 25% canary.

✨
vx-aiAI

No new error signatures since canary started.

Deploy rooms · 1 / 4 · auto-advance every 4.5s

Two surfaces · one operator brain

Chat from your terminal — or live the whole stack inside a TUI

Ops engineers don't want to leave the keyboard.vxcli chatstreams multi-provider answers; vxcli tuidrives every deploy, session, and AI thread from a single full-screen Bubble Tea dashboard.

vxcli chatmulti-provider · markdown + diffs · pipe-friendly
vxcli chat — bash · session 8e2aSTREAMING
claude-opus-4-7·provider: anthropic↑↓ history · / palette · ctrl-c quit
$ vxcli chat --provider anthropic --model claude-opus-4-7
→ Connected. Type your message, or `/sessions` to switch.
you:why is studio-api restarting on prod-eu?
claude:
0 chars · 184 tokens · $0.0021
/sessions · /model · /export

Anthropic, OpenAI, Google, OpenClaw — switch with /provider. Sessions persist under ~/.vxcloud/chat/ and resume instantly via the picker.

vxcli tuiBubble Tea · keyboard-driven · 60fps
vxcli tui — Bubble Tea82×24
vxcloud
node1.prodxcloud.com
● auth · joel
us-east-1
ServiceEnvStateAge
studio-apiprodrunning4m 12s
studio-workerprodrunning4m 12s
refund-botstagingbuilding0m 38s
mailcowprodrunning6d 4h
icodebase-idxprodrunning2d 11h
ml-trainspotqueued0m 04s
↑↓ nav⏎ open/ paletteq quit
Bubble Tea · 60fps

Deploys, sessions, doctor diagnostics, and embedded AI chat — all keyboard-driven. Press/for the fuzzy palette across rooms, AI, and resources.

Drive it from CLI · TUI · webhook · SDK

Every room is automatable, scriptable, and audited

Post structured CI/deploy/incident events from any service. Spawn AI copilots scoped to a room. Subscribe to AI suggestions and gate them behind your existing approvals.

  • vxcli chat — multi-provider, multi-session, terminal-native
  • vxcli tui — full-screen Bubble Tea dashboard, keyboard-only
  • POST /rooms/<channel>:messages — structured webhooks from CI
  • Python SDK — async streams, scoped agents, room-scoped events
1# Multi-provider AI chat in your terminal — no GUI required.
2# Switch providers and models mid-session; sessions persist locally.
3
4vxcli chat \
5 --provider anthropic \
6 --model claude-opus-4-7
7
8# Slash commands inside the session
9/sessions # picker for prior conversations
10/model haiku # hot-swap to claude-haiku-4-5
11/provider openai # hot-swap provider
12/files add ./PR-491.diff # ground the next answer in a file
13/export markdown ./out.md # dump the transcript to disk
14
15# Pipe-friendly: read stdin, emit answer to stdout.
16git log --oneline -20 | vxcli chat --stdin --one-shot \
17 "summarise these commits as release notes"
18
19# Use it in CI: post the digest into a chat room.
20vxcli chat --one-shot "weekly perf summary from $(date -d '7 days ago')" \
21 | vxcli rooms post --channel '#weekly-ops'
✨

Try it without leaving the terminal

vxcli tui --tab chat --provider anthropic — multi-provider AI chat, embedded in the dashboard.

rooms · v1 · prod
vxcli chat · vxcli tui · webhook · SDK

Built for production ops

The chat your platform team would build — if they had time

Operations-first channels

Per-deploy, per-incident, per-service rooms with structured event types — not just message walls.

Embedded AI copilots

Spawn scoped agents into a room — they read the thread, your runbooks, and recent incidents before replying.

Live deploy + CI events

Build, canary, rollback, and CI status arrive as first-class messages — with one-click drill into logs.

Incident timelines

Threads automatically become a postmortem-ready timeline with actor, action, and outcome captured.

Voice + screen share

WebRTC voice calls and screen share built in — no separate Zoom link, no separate Slack huddle.

Semantic search

Search across every room, file, and pinned doc with semantic match — find that thread you swear was last quarter.

Webhooks + SDK

Post events from any service in any language. Subscribe via async streams. Scoped tokens per room.

IdP-scoped + audited

SSO, MFA, role-based room access, and an append-only audit log streamed to your SIEM. SOC 2 ready.

CLI & TUI native

vxcli chat for terminal-only AI; vxcli tui for the whole stack — same auth, same audit, same rooms.

Real workloads, measured

Production traffic — not synthetic chat tests

Aggregated from 30 days of customer rooms. Median across deploy, incident, on-call, and support channels in production.

Audit trail captured on 100% of messages
AI suggestions cite the runbook they came from
Message delivery p5084ms
AI first-token p50184ms
Page → first-message34s
Suggestion accept rate71%
Audit log coverage100%

Where chat earns its keep

From release war rooms to weekly ops digests

Release war rooms

Every deploy gets a thread. CI events, canary metrics, AI summaries, and human decisions live next to each other.

Incident response

Page lands, room opens, AI pulls suspects from logs and PRs. Timeline is automatic; postmortem writes itself.

Customer support copilot

Tier-1 conversations with AI-drafted replies grounded in your docs. One-click send when an engineer approves.

On-call handoffs

Day-shift drops the tape; AI summarises overnight noise; night-shift gets a one-screen briefing.

Weekly ops digests

vxcli chat --one-shot piped into vxcli rooms post — every Monday, the team gets a grounded weekly summary.

Codebase Q&A in chat

Ask grounded questions about your repos right in the room. Citations link to the file in GitHub or GitLab.

Open it in your browser. Or your terminal. Or both.

One room. Every event. Every operator.

Open a workspace, post the first deploy event from CI, and watch your team converge — with AI copilots and an audit trail along for the ride.