Nebula review, pricing and verdict

Best for small teams where engineers, sales and marketing want to hand work to agents in a thread everyone can read and approve. It replaces scattered one-person chats with agents. A 7-day trial lets you test it, though funding, founding year and benchmarks are unpublished, so treat results as vendor claims.

Nebula is a team workspace from Agent Labs, Inc. of San Francisco in which people and AI agents share channels. Five agent integrations appear on its site, Claude Code, Codex, Gemini CLI, Cursor CLI and its own agents, and apps ship for macOS, Windows, Linux, iPhone and Android. Agents run on a cloud computer, and teammates review the results.

Maker: Agent Labs, Inc. · Autonomy: semi autonomous · Maturity: GA

About Nebula

Nebula is an agent-native team workspace made by Agent Labs, Inc., a company that says it is built in San Francisco (Nebula website, October 2026). It looks like a chat app with channels, but the people in a channel can be joined by AI agents, and everyone sees what the agents are doing and can step in. Compared with a personal assistant like Dots or Muse, where one person talks to one agent, the whole team sees the same thread.

You post a goal and Nebula divides the work between several agents. Coding agents you already run, including Claude Code, Codex and Gemini CLI, join on the plan you already pay for (from Anthropic, OpenAI and Google), and Nebula's own agents handle research and email through apps you connect with your own accounts, such as Gmail, Slack, GitHub, Notion, Linear and Stripe. The workspace also memorizes what the team writes, says and shows, and the site says it can sit in on calls and start follow-ups. Work runs on a cloud computer included with each seat, and the vendor's llms.txt describes agents that run on a schedule or on a trigger.

It suits small teams where engineers, sales and marketing share the same channels: a developer hands off a bug fix, a salesperson asks for a research brief or a customer email, and the team reviews and approves the result in the thread. Teams that want a single always-on business agent with no per-seat fee may prefer Tasklet, and people who want a general work agent inside an existing subscription may look at ChatGPT Work or Gemini Spark. For the one-person case, read our profiles of Poke and Cue. Two guides help when you set review rules and budgets: what a personal AI agent costs and approval gates for agents. If you are still choosing, describe your job to HokAI's matcher.

Pricing is one seat per person, billed monthly after a trial, and the owner needs a seat too. Messaging, calls, research and the cloud computer are included. Tasks given to Nebula's own agents use separate Nebula Credits, while any outside agent you connect bills through its own plan. Apps are listed for macOS, Windows, Linux, iPhone and Android. The pages we read list no country restrictions and quote prices in US dollars.

The product is new and moving: the site shows a team-calls feature, voice, miniapps and a connection library the vendor puts at 1,000+ OAuth toolkits. Funding, founding year and benchmark results were not published on the pages we read, so those fields stay empty. Teams that would rather host their own open-source agent can compare Hermes Agent and OpenClaw, and a single general-purpose agent is covered under Manus.

Pricing

Read from nebula.gg/pricing on 22 October 2026. The plan is a flat $25 per person monthly, started with a 7-day trial; no seat is free for the owner. A seat includes unlimited messaging, calls, research and AI orchestration plus a cloud computer. External agents such as Claude or ChatGPT run on your own subscription at no extra Nebula charge. Nebula agents use Nebula Credits for tasks, tools and computer hours, with a top-up prompt shown before further spend. Credit prices per model are listed in the vendor docs, which we could not read. Prices are in US dollars. Related: [Tasklet](/hub/agents/tasklet) bills per organization instead of per seat.

Key Features

• Channels shared by people and agents: Agents work in the same channels as your team, so everyone can see the work and step in when something looks off (Nebula website).
• Goal splitting across agents: You state a goal in a channel and Nebula splits it between agents, for example one researching leads while Claude Code builds a client portal, then brings the results back for review (Nebula website).
• Bring your own coding agents: Claude Code, Codex and Gemini CLI can be added and keep running on your own machine and subscription (Nebula developers page).
• Shared team memory: Everything written, said or shown in the workspace is memorized and can proactively start work, which the vendor says stops teams explaining things twice.
• Cloud computer included: Each seat includes a cloud computer where agents run work such as building and previewing a project (Nebula pricing page).
• Connected apps under your own accounts: Nebula agents use your own Gmail, Slack, GitHub, Notion, Linear and Stripe connections and act only on your behalf (Nebula pricing page).
• Schedules, triggers and voice: The vendor's llms.txt says agents handle jobs on a schedule or trigger, with miniapps and voice, and the site says Nebula can sit in on team calls.

Strengths

• External agents cost nothing extra: Claude, ChatGPT or Codex run on the subscription you already pay for, so the seat covers the workspace only (Nebula pricing FAQ).
• Work happens in shared channels where teammates can review and approve each result, which a one-person chat assistant like Dots or Muse does not offer (Nebula website).
• Spend is capped by design: Nebula says it asks you to top up credits before it spends more, so an agent run cannot create a surprise bill (Nebula pricing FAQ).
• Apps cover macOS, Windows, Linux, iPhone and Android, so a team on mixed devices can all join (Nebula developers page).

Weaknesses

• Every person needs a paid seat, including the owner, so cost grows with headcount before any Nebula Credits, unlike per-organization tools such as Tasklet (Nebula pricing page).
• Nebula-agent cost is metered in credits whose per-model prices sit in docs we could not read, so monthly spend on its own agents is hard to forecast.
• Evidence is vendor-published: no independent benchmark, funding figure or founding year was found, so those fields are empty.
• No free plan: there is a 7-day trial, then a seat is required even to use only your own external agents (Nebula pricing FAQ).

Visit Nebula Official Site