How it works
Four steps. You can be through all of them in an afternoon, and nothing acts on its own until you say it can.
Step 1
Create the workspace
You get an organization of your own. Invite your team, put people into teams, and decide which roles can approve spend, outbound email or customer-facing changes. Data isolation and permissions are in place before any AI touches a record.
- Roles and teams
- Approval rules per action
- Isolation enforced in the database
Step 2
Load the Business Brain
This is the difference between generic AI output and output that sounds like your company. You describe the company, products and pricing, ideal customers, brand voice, competitors, policies and goals — once. Every AI employee reads from it.
- Company, products, customers
- Brand voice and AI rules
- Documents and knowledge sources
Step 3
Brief the AI Manager
Ask in plain language: find 50 fitting companies in Ontario, draft a follow-up sequence, tell me which customers are at risk. The manager breaks it into tasks and assigns them to the right specialists with owners and deadlines.
- Plain-language briefs
- Automatic delegation
- Tasks with owners and due dates
Step 4
Review, approve, ship
Finished work comes back for inspection. Anything sensitive — sending email, spending money, changing customer-facing things — waits in Approvals for a named human. Everything that happened is in Activity, permanently.
- Approvals gate
- Full activity history
- Outcomes recorded, not assumed
What a realistic first week looks like
- Day 1 — workspace, team, permissions, first pass at the Business Brain.
- Day 2 — connect your mail service and run a first lead search against your ideal customer profile.
- Day 3 — review the AI-drafted outreach, approve it, and let the sequence send.
- Week 1 — pipeline, revenue and customer risk views start filling from real activity.