Managed AI workspace

Your AI workspace,
always on.

A persistent AI assistant across Telegram, Discord, Slack, and the web. When the job needs code, spawn a sandboxed agent that ships a PR and stops the meter. One subdomain, one bill, bring your own LLM key.

Nebula Deck - Managed AI workspace | Product Hunt
In the wild

Same deck, every screen you reach for.

Desktop console for inspection and orchestration. Same session, same memory, on your phone — chat anywhere, including with yourself across devices.

// live shot · compare moltis to openclaw and hermes agent 35.2K in · 1.3K out · 20.9 tok/s · 3m ago
🔒you.nebuladeck.dev/sessions/compare-moltis
Nebula Deck desktop console showing a comparison session with model usage, sandboxed runtime, and session sidebar.
Nebula Deck mobile chat — assistant fetching latest info from the web via web_fetch tool.
/ web · tool calls
// one session Start on desktop, finish on mobile. The deck holds the conversation, the tool calls, and every artifact.
// transparent runtime See the sandbox label, the model route, the token count, the speed. No hidden middle layer.
// real tools, real output web_fetch, shell, MCP — the deck shows the call, runs it, returns the trace. Markdown, tables, reasoning blocks render in place.
Pricing

$7 flat for the deck.
Heavy lifting, on demand.

The deck handles chat, coding, scripting, web search, and automation natively — included at $7/mo. When a job needs its own isolated environment, spawn a worker container. It runs, delivers, and shuts down. LLM tokens billed at your provider's rates, no markup.

The Deck
$7/ month

Always-on AI workspace. Chat, code, automate. Bring your own LLM key.

Subscribe
  • Dedicated sandboxed container, 24/7
  • Persistent hybrid memory (vector + FTS)
  • Shell execution, coding, scripting, cron jobs
  • Web search (unlimited, no API key)
  • Telegram · WhatsApp · Discord · Slack
  • Matrix · Nostr · Teams · web
  • MCP tools, self-extending skills
  • Voice in / voice out
  • Encrypted credential vault
  • BYOK — 20+ LLM providers
// worker containers (coding agents + browser) On-demand from $0.05/hr · per-second billing, 60s · Or choose a plan below
Developer
$15/ month

Solo dev. Occasional PRs and browser automations.

Concurrent workers3
Compute credit$10
Includes deck$7 value
Included worker hours200
Subscribe →
Studio
$39/ month

Parallel workflows.

Concurrent workers5
Compute credit$35
Includes deck$7 value
Included worker hours700
Subscribe →
Observatory
$99/ month

Full throughput for small teams.

Concurrent workers10
Compute credit$100
Includes deck$7 value
Included worker hours2000
Subscribe →
When a task needs full repo cloning, autonomous multi-file engineering, or browser automation, your deck spawns an isolated worker container. The worker does the heavy lifting in its own sandbox so your workspace stays clean. Workers are completely optional — most tasks run inside the deck itself.
FAQ

Questions?

Who is this for?

Developers, freelancers, and small teams who want an always-on AI workspace with persistent memory — plus coding agents when they need them — without running their own infrastructure.

Do I need a compute tier?

Not necessarily. The $7 deck handles chat, coding, scripting, web search, cron jobs, webhooks, MCP tools, and voice natively. Compute tiers add isolated worker containers for autonomous multi-file engineering (via OpenHands) and browser automation (isolated Chromium). Many users never need one.

What does "bring your own key" mean?

Your deck calls LLM providers directly — Anthropic, OpenAI, Gemini, DeepSeek, OpenRouter, and 20+ others. The key sits in your encrypted vault. Token charges hit your provider's invoice, not ours. We never touch your tokens and never mark them up.

Are worker containers autonomous?

No. You spawn them explicitly — a slash command or button in chat. The deck never decides on its own to spin up a billed container. If a worker gets stuck, it asks you on whatever platform you're on and waits for your answer.

Why not just self-host?

You can — both Moltis and OpenHands are open source. We're for people who'd rather not deal with Docker, TLS, networking, security updates, and the orchestration glue between the workspace and the coding agent layer.

Is my data safe?

Dedicated sandboxed container per customer — deck and workers both. Keys encrypted at rest, zeroed on drop, redacted from every log. Commands run inside the sandbox with restricted syscalls. We never read your conversations or access your provider keys.

Your AI workspace.
Always on, one bill.

Get started, $7 / mo