Personal sign-ins
A personal sign-in is one of your own accounts — a supplier portal, a carrier’s site, a web app your team uses — saved so an agent can sign in to it while it works on your tasks, without asking you each time. The password is stored encrypted and is never shown again: not to you, and not to the agent.
Find them in Settings → Personal sign-ins, or from the Configure & help page at the foot of the sidebar.
Adding a sign-in
Section titled “Adding a sign-in”Add a sign-in opens the form:
-
Name — what the agent sees, such as Amazon. Up to 80 characters.
-
Sites — the sites this sign-in may be typed into, one per line,
httpsonly, up to ten. The agent may only type it into these exact sites. -
User name — shown here so you can tell your accounts apart. The agent never sees it.
-
Password — when editing, leave it empty to keep the saved one.
Save sign-in stores it. The list shows each sign-in’s Name, Sites, Account and when it was Last used, with Edit and Delete. Deleting one means agents that were using it have to ask you to sign in instead.
How a sign-in is used
Section titled “How a sign-in is used”The panel How sign-ins are used on the page spells it out; in short:
| Stored | Encrypted, and never returned by AptiveAI Agents to anyone |
| What the agent sees | The sign-in’s name, its sites and an id — never the user name or the password |
| Where it can go | Only into a field on one of its exact https sites, and a password only into a password box. AptiveAI Agents checks both before typing |
| What the agent is told | That the field was filled — never the value. Text on the page that would show it is hidden from the agent. Screenshots are not |
| Submitting | Filling a sign-in never submits the form by itself — see Sending the sign-in |
| The browser | The agent’s browser does not save passwords |
| Changes | Adding, changing and deleting a sign-in is recorded |
If no saved sign-in matches the site in front of it, or more than one could, the agent asks you what to do. When exactly one matches, the agent uses it without asking the first time it signs in to that site in a task, and the transcript says so; a second attempt on the same site asks first. On an automation, a sign-in that did not work is not tried again, so the account is not locked.
Sending the sign-in
Section titled “Sending the sign-in”Filling a sign-in does not send it: the agent then clicks the form’s sign-in button, or presses Enter. At any approval level, that one step needs no approval card if the agent has just filled both the user name and the password of one stored sign-in into one sign-in form on the page. The transcript says Sent the stored sign-in it had just filled on this page, without asking — “‹button›”.
Unless the task is at Never ask, the send still waits for your approval when:
- The form looks like something other than a sign-in — a signature, a payment, an approval, a transfer, a release. An electronic-signature box that asks for your password keeps its card.
- It is a retry. After a sign-in that did not work, sending it again on the same sign-in page asks.
- The sign-in is split over two pages — the user name, Next, then the password on the next page, as on Microsoft’s and Google’s sign-in pages. The final click there still asks.
- The agent did something else in between. Anything other than reading the page, or typing somewhere else, between filling the sign-in and sending it brings the card back.
- It presses Enter on a form with other buttons, such as Cancel or Send code. Clicking the form’s sign-in button is not affected.
Nothing else changes: only the agent’s own fill of a stored sign-in opens this way, only for the one step that sends it, and Approve all and automations that run on their own work as before.
Yours and the agent’s
Section titled “Yours and the agent’s”An agent can hold sign-ins of its own, set in its builder. On a task, the agent can use its own sign-ins and those of the person who started the task. When both have a sign-in with the same name, yours is used and the agent’s is not.
Signing in with a sign-in puts the task’s browser in as that account, and the session carries into your later tasks on the same browser.