Skip to main content
Most chat assistants answer the question in front of them and then forget you existed. Abbie is built to behave more like a teammate.

She keeps a backlog

Ask her to do something involved and she doesn’t need to finish it in one reply. She can keep it as an open item, pick it back up in a later message, and tell you when it’s parked waiting on a decision from you. See Abbie’s backlog.

She remembers

Facts about you and your organisation persist between conversations: your role, your preferences, standing context about the account. You don’t have to re-explain yourself every time. See Memory.

She adapts to you

Abbie has an account-wide voice, and a personal layer on top that calibrates to how you individually like to work: shorter answers, a more formal tone, whatever fits. See Rapport.

She checks her own work

Before she answers, an independent pass reviews the response against your preferences and the facts available, and can send a weak answer back for another attempt. See Confidence and verification.

She reaches out first

If you let her, Abbie will check in after a quiet stretch, follow up on something she’s still waiting on you for, or suggest an idea, never more than the cadence you’ve set. See Proactive Abbie.

She's honest about what she can reach

Abbie only uses the tools and integrations you or an admin have actually authorised. If a job needs something she doesn’t have access to, she’ll ask rather than fail silently. See Connecting Abbie.

She thinks before she answers

Every reply from Abbie runs through a short pipeline: understanding what you’re asking, pulling in the right context, drafting an answer, and then reviewing it, rather than a single model just generating text. You can watch this happen live as a strip of stages above her reply. See Watching her think.
Abbie is a QuivaWorks assistant. Where this documentation says “she”, it means the assistant behaviour described here, not a person, and not a general-purpose model you’re prompting directly.