Skip to main content
Every task you hand to a coworker runs in one of two modes: Everyday or Deep Work. It’s not a setting you configure once — it’s a per-task choice about how much depth a specific piece of work deserves.

What it actually is

Everyday and Deep Work aren’t tiers of a subscription, and they’re not a toggle sitting in your console. They’re a real difference in how a coworker approaches the task in front of it right now — how much it plans before starting, how many angles it considers, how thoroughly it checks its own work, and, for research-heavy tasks, how many separate outputs you get back.

Why you’d care about this

Two situations bring you here: you’re not sure whether a quick answer or a thorough one is what a task actually needs, or you already got a fast answer and want more depth on top of it. This page is about making that call deliberately instead of guessing.

How it actually works

Everyday

The default. Fast and efficient, and it handles the large majority of requests well — from quick lookups to solid overviews. You get the one deliverable that answers your question.

Deep Work

For complex, strategic questions where you want more breadth and depth. The coworker plans more thoroughly up front, considers more angles and sources, and cross-validates what it finds before it hands anything back.
For research-heavy work in particular, Deep Work doesn’t just mean “a longer version of the same file” — you get a full set of outputs across formats instead of one condensed one: a Word report, an Excel workbook with the underlying data, and a PowerPoint for presenting it, each covering a different angle of the same question.
Under the hood, Everyday runs on a faster model and Deep Work runs on a more capable one — worth knowing if you’re curious, but not something a coworker will bring up unprompted. What changes for you is depth and output, not a model name.

Getting into Deep Work is always a conversation, never silent

1

The coworker notices, and asks

If your request has the kind of scope that clearly benefits from more depth — multi-source research, a full deliverable package, a complex multi-chart dashboard, a full brand system — the coworker recommends Deep Work before starting, and waits for your answer. It never switches modes on you mid-task.
2

Or you just ask for it yourself

You don’t have to wait to be asked. Something like “Hannah, really go into detail on this one — use Deep Work mode” works anytime.
3

Simple requests just proceed

If a request is clearly straightforward, the coworker handles it in Everyday mode without asking — no unnecessary back-and-forth for a quick question.
When in doubt, start with Everyday. You can always ask for more depth on a specific part afterward — you’re not locked into one mode for the whole task.

What pushes a task toward Deep Work, per coworker

Multi-source competitive analysis, full audience profiling, a complete research package rather than a quick lookup. Ongoing political monitoring always runs in Deep Work, given what it’s for.
A multi-chart interactive dashboard with filtering, a large or messy dataset pulled from multiple sources, a scrollytelling piece, or turning Hannah’s research into a polished presentation. One difference from the others: Deep Work doesn’t multiply his output — you still get one dashboard, just built with more validation and design care.
Usually it’s the coordination itself — pulling several coworkers together on one project — that pushes a task into Deep Work territory, more than the strategic advice on its own.
A single document or concept is Everyday. A full brand system — guidelines, templates, tone of voice, examples across formats — is Deep Work territory.
A single-channel plan is Everyday. A complete customer lifecycle strategy with a full media plan is Deep Work territory.

The trade-off

EverydayDeep Work
SpeedStandardRoughly 2–3x longer
CostStandard creditsRoughly 3–5x more credits
DeliverablesOne deliverable that answers the questionA full set — report, data, presentation, and more as needed
Best forMost day-to-day requestsComplex, strategic, or high-stakes questions
Those multipliers are rough planning estimates, not a guarantee — there’s no mechanism to cap spend mid-task, so actual cost is always billed from real usage. See Credits & Billing for how credits work generally.

Limits

Deep Work means more thoroughness, not an unlimited budget — a coworker won’t promise you a fixed cost ceiling upfront, because the real cost depends on what the work actually turns up. And as noted above, Alex’s Deep Work doesn’t add extra files the way Hannah’s does; the depth shows up in the build quality instead.

How this connects

Every coworker handles this the same way — see Hannah, Alex, Elena, Maya, and Jamal for what their work looks like day to day. For the mechanics of handing off a task at all, see How delegation works. For what any of this actually costs, see Credits & Billing.
Try it: next time you send a research-heavy or strategic request, add a line like “this one’s worth going deep on” — or just wait and see if your coworker offers.