Quickstart
From signup to your AI operator’s first brief in about five minutes. No implementation project, no onboarding call.
1. Sign up
Go to waveguide.adwave.com/signup, enter your email and a workspace name (business, project, or your own), and create a free account. You start with $5 in free AI credits — no card required. Paid plans (with an optional 14-day trial) are available from Pricing.
Within a minute you’ll receive a welcome email with a sign-in link. Waveguide uses email sign-in links instead of passwords: there’s nothing to remember and nothing to leak. The welcome link is valid for 24 hours; you can always request a fresh one from the sign-in page.
2. The guided setup
Your first sign-in drops you into a five-step setup:
- About you — name, website (optional), industry, what “growth” means for you (leads, sales, or bookings), and your timezone. This seeds your agent’s persona so its first actions make sense.
- Connect accounts — one OAuth click each for Gmail, Google Calendar, Meta Ads, or Stripe. Cards are ordered for your goal; you can start with just one. See Connect for what the agent does with each.
- Limits — how carefully the agent should check with you (Careful / Balanced / Hands-off) and, if Meta is connected, a daily ad-spend cap. See Needs OK & policies.
- Updates — where the agent should talk to you: Slack, Telegram, and/or email. Email is on by default so you never miss an approval. See Channels.
- Your first weekly plan — it looks at what you’ve connected and suggests playbooks to start with.
Every step except the first is skippable, and your progress is saved server-side — close the tab and pick up where you left off.
3. Daily life with Waveguide
After setup you land on Ask — “What do you want to accomplish?” Type a goal and your agent makes a plan (drafts only by default). Approve the plan, then it runs while you’re away.
The agent also wakes on its own:
- It wakes on real signals — a lead arrives, a campaign crosses a threshold, a scheduled digest fires. It doesn’t poll or burn budget when nothing is happening.
- It sends a morning brief with what happened, what it did, and what needs you.
- When it wants to do something risky — spend money, send an external message — it sends an approval card. You approve or deny with one click from Slack, Telegram, email, or the dashboard. If you do nothing, the request expires safely in 15 minutes and the agent moves on.
- Everything it does is in History, down to individual tool calls.
Where things live
| What | Where |
|---|---|
| Ask — Inbox + Chats (Needs OK, suggestions, updates) | Dashboard → Ask |
| A single job or Q&A chat | Dashboard → Chat (open from Chats or a Job) |
| Suggestions your agent spotted | Dashboard → Suggestions |
| Needs your OK | Dashboard → Needs OK |
| Jobs (goal-driven work) | Dashboard → Jobs |
| Connected accounts | Dashboard → Connect |
| Capabilities (MCP tools) | Dashboard → Capabilities |
| Full history | Dashboard → History |
| About my business | Dashboard → About my business |
| Plan, usage, invoices | Dashboard → Billing |
| Kill switch, export, delete | Dashboard → Settings |
Next steps
- How it works — the architecture in plain language
- Workspaces — personal vs business under one login
- Approvals & policies — teach the agent what it may do alone
- About my business — tune how your agent thinks and writes