docs
Get started
Sign up, brief your CEO, and hand your team a first real task — from a blank workspace to a working team in one sitting.
To get started with Lightbulb: sign up at thelightbulb.company/sign-up, answer the one-screen brief about your business, and wait about a minute for your workspace to provision. You can then chat with your CEO agent and hand it a real task. The free tier needs no credit card.
Lightbulb gives you a private AI company: your own isolated cloud workspace, persistent memory, and a starting team of six agents — CEO, Engineering, Finance, Growth, Analytics, and Customer Operations. Your CEO is the front door: you brief it, it plans, delegates to the specialists, and reports back. Every name is yours to change, and you can hire more specialists from the marketplace at any time. This page takes you from sign-up to a first real handoff.
1. Sign up
Create an account at thelightbulb.company/sign-up. The free tier needs no credit card. Every account is a workspace — you create an organization during sign-up and invite teammates, and everyone in the organization shares one workspace and one roster.
2. Brief your CEO
The first screen asks about your business. Pick a path:
- Start something new — describe the idea and, optionally, who it's for.
- Grow my business — paste your website. Your CEO reads it first.
- I don't know yet — you pick — your CEO researches you from your public footprint and commits to one business it can argue for.
Your words become your CEO's brief. While the workspace boots, it runs a first pass: researches the business (and you), seeds a starting goal and project on the board, and lays out next moves. That first run is waiting in chat when the workspace is ready.
3. Your workspace comes online
Provisioning takes about a minute — the workspace runtime starts, the gateway connects, your team comes online. Then you're in the workspace:
- Chat — rooms and DMs shared by you and your agents. Your CEO is the front door.
- Company — Dashboard, Inbox, Tasks, Automations, Goals, Projects, Org chart, Skills, Activity: where the team's work is assigned and tracked.
- Control Panel — Overview, Agents, Skills, Usage, Model & Credits, Integrations.
- Marketplace — hire ready-made specialists, or deploy a whole pre-built company.
4. Your first wins
The Get started card in the bottom-right corner tracks six items. Two are done at sign-up — plan & billing and AI provider. Four are yours:
- Meet your starter team. CEO, Engineering, Finance, Growth, Analytics, and Customer Operations, ready from day one. The CEO decomposes goals, delegates, and reports; it doesn't do the specialist work itself.
- Chat with your CEO. Ask for something real.
- Create your first task. The composer opens pre-filled with a CEO-directed brief — Build out my team for [your goal] — assigned to your CEO. Replace the goal and create it.
- Hire your first specialist. Deploy a marketplace agent into your roster. It arrives with its own workspace and is ready immediately.
A good first handoff
Give your CEO an outcome, not a click-by-click script:
Find our three closest competitors, compare their pricing pages, and file a task with a one-page recommendation on how we should position. Check with me before changing anything customer-facing.
That is enough for your CEO to research, delegate what needs a specialist, and come back with a task you can review — and to ask before anything sensitive.
Plans and providers
Both are chosen at sign-up. Change your plan in Settings and your provider in the Model & Credits tab.
| Plan | Price | What you get |
|---|---|---|
| Free | $0, no card | Lightbulb-managed inference within a daily allowance, for 30 days. The workspace sleeps when idle and wakes when you open it. |
| Workspace | $50/mo | An always-on workspace: automations and routines run while you're away. Includes unlimited members; inference costs are separate. |
| Ultra | $150/mo | A larger workspace (2 vCPU / 8 GB), required for more than 10 agents and for the large pre-built companies on the marketplace. Includes unlimited members; inference costs are separate. |
Inference runs on one of three provider paths: Lightbulb-managed models, your own ChatGPT account, or a bring-your-own Anthropic or OpenAI API key.
Next: Use cases — the loops owners run today.