How to use Gems in Neuro for ad operations

In Neuro, a Gem is an operating mode

Media buyers tend to ask Neuro the same handful of things: “Summarize last week’s performance,” “Find budget-cut candidates,” “Draft report copy for the client.” But even when the question is identical, the standard for a good answer changes every time. A response written for a client report and a response written for budget-change review shouldn’t come back in the same tone or format.

A Neuro Gem captures that difference. Rather than just saving a long prompt, it creates an AI role built for one specific ad task and applies it to the chat. Pick the weekly report Gem and answers follow your client reporting standards; pick the budget review Gem and answers zero in on before/after values and approval risks.

1. In the chat input box, open the Gems menu and click Create new Gem.
2. Name it after a single task. Names like Weekly performance report, Budget change review, and Creative fatigue review work well because anyone gets them at a glance.
3. For save location, choose Workspace shared if it’s a standard the whole team will use, or My Gem if it’s a personal workflow.
4. Under Guidelines, spell out the role, the data to check, the order of analysis, the actions it must never execute, and the output format.
5. Select a Gem in chat, then ask away as usual. The same question gets read through the lens of whichever Gem you picked.
6. If answers keep missing the mark, don’t pile more onto your question—update the Gem’s guidelines instead. For recurring work, the standard belongs in the Gem, not in every prompt.

Five Gems worth starting with

You don’t need to turn every task into a Gem on day one. Start with the work that comes up most often and where inconsistent standards quickly drive up your operating costs.

All five play to what Neuro does best. Ad platform data gets pulled through MCP tools, brand context comes from the workspace, and any recommendation that would change a live account stops in a format a human can review first.

Write Gem guidelines like an ops doc

A Gem whose guidelines stop at “You are a marketing expert” is close to useless. An ad operations Gem should read more like an ops manual than a character sheet. Spell out the order in which the AI reads data, where it has to stop, and the format its answer should take.

Once you pick a Gem, your prompts get shorter

When your Gems are good, your prompts actually get shorter. You stop repeating conditions like “don’t execute,” “back it with evidence,” or “write it as client-facing copy” every single time. Those live in the Gem’s guidelines, so in chat you only have to name the target, the date range, and the goal.

• With the Weekly performance report Gem selected: “Compare the last 7 days with the prior 7 and draft a report outline.”
• With the Budget review Gem selected: “Find budget adjustment candidates for next week. Don’t execute.”
• With the Creative fatigue review Gem selected: “Segment the last 14 days of creatives by fatigue signal and priority.”
• With the Client playbook Gem selected: “Polish this month’s report copy in a conservative tone for this client.”
• With the New account owner onboarding Gem selected: “Tell me which campaigns and recurring issues to check first in this account.”

If your prompts keep getting longer, the Gem isn’t pulling its weight yet. Don’t leave recurring conditions in the prompt—move them into the Gem’s guidelines. That way, even as people rotate on and off the account, no one has to re-explain the standard for the same task.

How Google Gems and Neuro Gems differ

Google Gemini’s Gems let you build your own custom AI experts, saving guidelines and knowledge files to reuse. Google pitches Gems for repetitive tasks, writing help, and roles that need a specific context and style.

In Neuro, the real difference comes down to data and execution boundaries. A Neuro Gem isn’t a general productivity tool—it lives inside an ad operations workspace. It ships with your connected ad platforms, your workspace brand context, team sharing, and a clear scope for which changes need approval.

Why personas matter

A persona isn’t about dressing the AI up as a character. In Neuro, a persona is the mechanism that locks in your work standards. Given the same data, a report writer, a budget reviewer, a creative reviewer, and a new-owner coach each need to think in a different sequence.

In ad operations, the wrong answer isn’t the only dangerous one. A plausible answer with no execution boundaries is just as risky. “Cut budget on underperforming campaigns” sounds simple enough, but if it overlooks the learning phase, conversion lag, remaining monthly budget, or the client’s promo calendar, it can turn into a live incident on the account.

Good personas cut down on those incidents. They define which data the AI looks at first, which causes to rule out, which actions it can never take without approval, and which claims it can’t state as fact to a client. Once that logic lives in a Gem, Neuro chat stops being simple Q&A and becomes a place that repeats your team’s way of running ads.

A checklist you can apply in Neuro today

• One Gem, one task. Don’t cram reports, budget reviews, and creative reviews into a single Gem.
• Name Gems after the task, in words a teammate gets at a glance.
• In the guidelines, spell out the decision flow and execution boundaries in more detail than the role itself.
• Use workspace-shared Gems for team standards and personal Gems for individual workflows.
• When answers keep missing the mark, don’t write longer prompts—update the Gem’s guidelines.
• For anything with real-world impact—ad spend, bids, campaign status, client-facing copy—make it stop as an approval card.

The teams that get the most out of Gems aren’t the ones writing ever-longer prompts. They’re the ones baking the standards for repetitive work into the product, so people only review a narrower set of decisions. That’s exactly why Gems exist in Neuro.