Introducing Multi-Agent Answers in Ask AI

Picture the question a frontline employee actually asks at the start of a shift: "What shifts do I have this week, and how much leave do I have left?" That's two questions wearing one sentence — one for scheduling, one for leave. Until now, Ask AI answered the strongest intent and let the other slip. Multi-Agent Answers changes that: Ask AI now recognizes when a question spans several areas, sends each part to the specialist agent that owns it, runs them at the same time, and hands back one combined reply — with a trace of which agents contributed.

Real questions don't respect app boundaries

People don't think in apps. They think in moments — the start of a shift, the end of a week, the morning after a call-out. A single sentence routinely touches scheduling and time off, or open tasks and tomorrow's roster, or an IT ticket and remaining PTO.

The old way of handling that was to make the person do the routing themselves: ask one thing, read the answer, switch context, ask the next thing, stitch the two together in their head. Every one of those switches is friction, and friction is exactly what the frontline can't afford mid-shift. It's the same fragmentation that has managers bouncing between a scheduling tool, a messaging app, and an HR portal to assemble a picture that should have arrived in one place.

Multi-Agent Answers meets the question the way it was actually asked. Instead of forcing the employee to think like a software architect — "which app owns this part?" — Ask AI does the decomposition, fans the work out to the right specialists, and brings back a single answer. The interface finally matches how people actually phrase what they need.

What's new

The right specialists, picked automatically. When a question looks like it spans more than one area, Ask AI breaks it into scoped sub-questions and routes each to the agent that owns that domain — scheduling, leave, tasks, service desk, reporting, attendance, and the rest of the 80-agent network. There's no app-switching and no "ask that one in the other tab." The employee asks once, in plain language, and the routing happens underneath.

Answered in parallel, not one-by-one. The specialist agents run simultaneously, each inside its own securely restored tenant context. You get a cross-domain answer in roughly the time a single agent would take — not the sum of three sequential round-trips. Coordinating several agents shouldn't mean waiting three times as long, so it doesn't.

Combined into one coherent reply. A final synthesis pass weaves the specialists' answers into a single response that reads in one voice, preserving every link, ID, and number exactly as the agents returned them. You don't get a copy-paste collage of three separate answers with three different tones — you get the one answer the question deserved, with the relevant facts woven together.

A "show your work" trace. Every multi-agent reply ends with a collapsible "🤝 N agents contributed to this answer" panel. Expand it and you see exactly what each specialist said before synthesis. A combined answer is only trustworthy if you can see where each piece came from — so we made that auditable by default, on every reply, both live and when you scroll back through history.

Permission-aware and safe by design. Each agent only runs if the employee is allowed to use it, and each one only ever sees its own scoped part of the question. Anything that would change data — filing a request, creating a task — is never merged or run twice in parallel; if an action needs your confirmation, that surfaces on its own, exactly the way it does today. Coordinating more agents doesn't loosen a single guardrail.

What it looks like

These are real cross-area questions Multi-Agent Answers is built to handle — and they show up as ready-to-tap prompts in the new "Ask Across Areas" suggestions inside Ask AI:

• "What shifts do I have this week, and how much leave do I have left?" — scheduling + leave
• "What are my open tasks, and what shifts do I have tomorrow?" — tasks + scheduling
• "Do I have any open IT tickets, and what's my remaining PTO?" — service desk + leave
• "Show headcount by department and this month's attendance summary." — reporting + attendance

Each one returns a single answer, drawn from up to three specialists working together — followed by the contributor trace so the employee (or their manager) can see precisely how it was assembled.

Availability

Multi-Agent Answers is on by default for every business with Ask AI enabled — there's nothing to install and nothing to configure to start using it. Employees simply ask cross-area questions in the Ask AI sidebar, or tap an "Ask Across Areas" suggestion to begin. Each question coordinates up to three specialist agents.

Admins who want to manage the behavior can toggle it from the AI Features settings for Ask AI. And as always, every agent stays bound to the permissions each employee already has — multi-agent coordination expands what a single question can answer, never what a given person is allowed to see.

What's next

Multi-Agent Answers is the coordination layer for the entire Ask AI agent network. That has a compounding effect worth calling out: the more specialist agents your business turns on, the more cross-area questions a single ask can resolve. An organization running scheduling, leave, tasks, and service desk agents can already answer questions that touch all four; as you enable more of the network, the range of "ask it however you think about it" questions widens with it.

We'll keep pushing on that range — making the decomposition sharper, the synthesis cleaner, and the set of questions a single ask can resolve broader over time.

See it in your workspace — open Ask AI and ask something that spans two areas at once. Explore Ask AI →