Agent chat

Overview#

Agent chat lets you talk to an agent directly: ask about something it found, ask why it made a particular recommendation, or ask it to do a specific piece of work for you.

It is the interactive counterpart to scheduled runs. Those bring things to you on their own timetable; chat is how you go to them when you have a question that will not wait until the next one.

Interpreting results#

The agent can see your project's data, meaning your topics, visibility, citations, competitors and recent runs. It is not a general business assistant with a view of your company, and it answers from what Seerly measures rather than from anything broader.

That makes specific questions considerably more productive than general ones, because the agent can ground them in something. "Which topics did we lose ground on this month, and what is being cited instead?" is a question it can answer properly. "How do we grow?" is not, and the answer will be as vague as the question deserves.

Messages count against a monthly allowance that varies by plan, and work requested through chat still goes through approval like everything else. Asking for a draft produces a draft, not a published page.

A grounded exchange

You

Which topics did we lose ground on this month, and what is cited instead?

Seerly Advisor

Two topics dropped, onboarding from rank 4 to 6 and time tracking from 3 to 5. For onboarding, a review site and one competitor page are cited in every run. Want the pages?

Panels and controls#

New Chat starts a conversation, and Chat History keeps everything, described on screen as All your conversations with Seerly Advisor, so you can return to an answer later rather than asking the same question again.

Also available on Telegram. Connecting Telegram lets you ask from your phone without opening the dashboard, which in practice is the difference between asking a question when it occurs to you and asking it when you next happen to sit down at a laptop.

Where chat fits in the workflow#

Chat earns its place at two moments. The first is triage: several things moved, and asking "what changed this week, and what matters most?" beats opening six screens to reconstruct the answer yourself. The second is the follow-up a screen cannot answer: the dashboards tell you what happened, and chat is where you ask why, across screens, in one place.

What it should not replace is the screens themselves for anything you will act on. Verify the specifics on the screen that owns them before the work begins, and when a conversation ends in a decision, record it in Tracked actions so the timeline knows.

Best practices#

The questions worth bringing here are the ones that would otherwise mean clicking through several screens and holding the pieces in your head while you do it:

• "Which topics have we been absent from for more than three runs?"
• "What do the pages cited for this topic have in common?"
• "Which sources cite our competitor but not us?"
• "Draft a brief for the biggest gap in this topic."

Next steps#

1. Ask about a specific topic, engine or competitor rather than in general terms.
2. Ask for the reasoning when a recommendation surprises you. Understanding why is usually worth more than the recommendation itself, and it tends to transfer to the next decision.
3. Use it to triage when several things have moved at once and you need to decide what matters first.
4. Verify anything factual before acting on it. Treat the output as a well-informed colleague's view rather than a system of record.

Limitations#

The agent answers from your project's Seerly data. It has no access to your CRM, your analytics, your revenue, or anything else you have not connected.

It cannot publish, change settings or spend anything on your behalf. Everything actionable goes through approval.

Allowances are set per plan and reset each billing cycle. See seerly.app/pricing.