Help
Frequently asked questions
Short, plain-English answers about how Saidly works. If your question isn't here, email support@saidly.ai.
What is Saidly?
When someone asks ChatGPT, Claude, Gemini, or another AI assistant about your company, product, or name, the answer they get shapes what they think of you, and you usually never see it. Saidly asks those models the same questions on a schedule, scores how positive or negative each answer is, tracks how it changes over time, and tells you what to fix. In short, Saidly is AI sentiment monitoring: it measures how positive or negative the AI assistants are about you and tracks it over time. Think of it as search-result monitoring, but for the AI assistants people now ask instead of typing into a search box.
The three search modes
Every report is built one of three ways. The mode decides whether the AI models look at the live web before they answer, or rely only on what they already learned. You set the mode per name, and you can change it any time in Settings.
Web search
What people actually see today.
The models answer with live web search turned on, the same way a person using ChatGPT or Claude with web access would. They read current pages, then answer, and we keep the sources they cited. This is the closest match to what a customer, reporter, or investor would find right now.
On every plan
Model-only
What the model believes on its own.
We turn web search off, so the model answers only from what it learned during training. No live pages, no citations. This is what the AI will say when someone asks a quick question without it looking anything up. It can lag reality by months, because training data has a cutoff date.
Pro and Business
Compare
Both answers, side by side.
We run the question twice, once with web search and once without, and show both answers, both scores, and the gap between them. A wide gap is the useful part: if the live answer is warmer than the model-only one, the internet has caught up but the model's built-in view has not, so it pays to wait or to seed fresh content.
Pro and Business
| Web search | Model-only | Compare | |
|---|---|---|---|
| Looks at the live web | Yes | No | Both |
| Includes cited sources | Yes | No | Yes (web side) |
| Reflects today | Yes | Lags by months | Shows the difference |
| Best for | What people find now | The AI's baseline view | Spotting the gap |
| Plans | All plans | Pro, Business | Pro, Business |
Which mode should I pick?
If you only want one, choose Web search. It mirrors what people actually see and it is on every plan. Reach for Model-only when you want the AI's baseline impression of you, the answer it gives off the top of its head. Use Compare when you are running a PR push or a launch and you want to see whether the live web and the models' built-in view agree yet.
Which AI models do you check?
Saidly covers the four major AI assistants people use to ask about companies and people: Claude, ChatGPT, Gemini, and Grok. Each report shows exactly which models answered so you always know what the score is based on.
How often do reports run?
Reports run on a schedule you choose: monthly, weekly, or daily, depending on your plan. You can also run an on-demand report at any time from your dashboard when you want an answer right now, for example right after a launch or a news story. Scheduled reports arrive by email and stay in your dashboard.
What counts as a "name"?
A name is anything you want the models to be asked about: a company, a product, a person, an organization, a campaign, or a topic. You pick the type when you add it, which helps the report frame the question the right way. Each name is tracked and scored on its own.
How is the score calculated?
For each model we read its full answer and rate the overall sentiment from negative to positive, then roll those up into a single 0 to 100 score for the name. The report also pulls out the recurring themes, the sources behind them when web search is on, and the week-over-week change, so you can see both the number and what is moving it.