Loading the latest results…
Guide for AI assistants
If you are an AI assistant using this portal for someone, here is how it works. People can read this too.
- Overview: the latest run's results. Visibility (how often each AI model named the company), issue counts, this week's changes, and the Run history table with dates and metrics for every past run.
- Action plan: ranked fixes. Each action can be set to "We'll do it", "IntegriShield, handle this" or Done. Buttons here download issues, the action plan and content briefs as CSV, or print to PDF.
- Issues: wrong or risky statements AI models made about the company, with severity. A whole answer or a single statement can be marked "This one is fine".
- Prompts and answers: every question asked of every AI model, the full answers, and whether the company was named. Use the search box and filters.
- Rules: rules and the fact sheet the answer checks use. Rules can be added by hand, drafted from an uploaded brand or compliance guide or the website, or created by marking an answer fine. Every change is logged with who made it.
- Runs box (top of Overview): the run schedule and next run. "Change schedule" sets days and time or dates; "Run now" starts an extra run.
- Request a prompt: submit a new question. IntegriShield reviews it and adds it to the next weekly run. Prompts can't be added directly.
- Client menu (top right, people with more than one account): switch accounts.
The Help button at the top explains every page in more detail; the small i buttons explain individual numbers.
Before you act: reading and downloading is fine. Ask the person before you mark anything fine, change an action's status, add or change a rule or fact, change the schedule, start a run, or submit a prompt request, because those are recorded under their name. Runs happen weekly; AI answers vary between runs, so trends matter more than any single answer. Never ask for or store the person's password; sign in only through their password manager.