Answers that show the rows they came from.
Ask about any campaign, number or customer in plain words. Obelisk reads your connected data and cites what it used.
Ask why leads dropped, and see where the answer came from.
The agent reads Google Analytics and Search Console, answers with the rows it used and asks before it changes anything.
Why did leads from the website drop last week?
CMO agentRead Google Analytics and Search Console
| Channel | Week before | Last week | Change |
|---|---|---|---|
| Search | 96 | 41 | −55 |
| Paid | 82 | 79 | −3 |
| Direct | 36 | 28 | −8 |
| All leads | 214 | 148 | −66 |
| Page | Week before | Last week | Change |
|---|---|---|---|
| /guide | 1,180 | 212 | −968 |
| /guides/implants | 0 | 64 | +64 |
Ask in plain words, or give it a job.
Questions and instructions go in the same box. The answer always lists what it read.
- Show last week’s leads by channel
- Compare Meta and Google spend this month
- Write the Monday report
- Find the pages that lost traffic
- Draft three headlines for the spring offer
- List the leads nobody has called
What the answers are made of
Where does an answer come from?
From the accounts you connect and your own documents. Each answer lists the sources and rows it used, so you can check it.
Can it read our briefs and past reports?
Yes. Briefs, past reports and your documents sit in a knowledge base the agent searches before it answers, and it cites them by name.
Can chat change a campaign?
It can ask to. Anything that changes a live account becomes a request a person approves, and the request says exactly what will change.
Which AI model answers?
Each task uses the model that suits it. When a better one arrives, Obelisk switches without you changing how you work.
Bring the question your marketing data should answer.
On the call we show you Obelisk on a business like yours.