The Three Core Pulses Explained

Every new organisation in The GAiGE is set up with three built-in pulses. Between them they answer the three questions that matter when you are measuring AI adoption: is the tool being used and is it saving time, is it worth keeping, and are your people equipped to get more out of it.

All three arrive paused. You activate them from the Pulses page once your tools are added and your team is ready, so nothing fires before you want it to.

This article covers why there are three rather than one, exactly what each one asks, when each fires, and what you can change.

Why three, not one

Different questions need different intervals. "Did this save you time today?" only works if you ask it while today is still fresh. "Is this tool worth keeping?" needs a longer horizon, and asking it three times a week would produce the same answer three times a week.

So the questions are split across three pulses, each with its own job, its own cadence, and its own answer format. Individually each is short enough that people answer it. Together they give you the full picture.

They also stay separate in the reporting, on purpose. The AI Pulse owns the performance score, The AI Review owns NPS, and The AI Capability Check feeds neither, so two different ideas never get averaged into one muddy number.

Pulse 1 · The AI Pulse

Default cadence: 3 times per week, per person, per tool. Time to answer: around 30 seconds.

The frequent, behavioural pulse. Its job is narrow: was the tool used today, and did it save any time. This is where the active-user, hours-saved and ROI figures come from.

It branches on the first answer, so nobody sees all five questions.

Answer yes and you get the time-saved question, then performance, then the optional comment. Answer no and you get the reason question, then the same optional comment.

The time-saved options are None, Up to 30 mins today, 30-60 mins today, 1-2 hours today, More than 2 hours, and I lost time because of it. That last option carries a negative value, so a tool that costs people time shows up as costing people time.

The reasons offered after a "no" are I didn't need it today, I forgot it was available, It was too slow or clunky, I didn't trust its output, I couldn't access it, and I prefer a different tool.

Pulse 2 · The AI Review

Default cadence: once a fortnight, per person, per tool. Time to answer: about a minute.

The deeper, tool-focused pulse. Its job is the keep-or-drop judgement a daily check cannot make: the difference the tool makes, how much its output has to be fixed, whether the person would miss it, and whether something better exists.

1. Value. Recently, how much of a difference has [tool] made to the work you produce? (A big difference through to It's made things worse)
2. Trust. How often do you have to fix or redo [tool]'s output before you can use it? (Rarely or never, Sometimes, Often, Almost always)
3. Recommendation. How likely are you to recommend [tool] to a colleague doing similar work? (0 to 10, this is the NPS question.)
4. Would miss it. How would you feel if you could no longer use [tool]? (Very disappointed, Somewhat disappointed, Not disappointed)
5. Something better. Do you know of another tool that would do this job better than [tool]? Answering yes opens a free-text follow-up asking which tool, and what it does better.
6. Improvement. What would make [tool] more useful to you? (Better quality output, Faster responses, Easier to use, Better fit for my tasks, Nothing, it works well, Other)
7. Anything else. Free text, optional.

This is where renewal risk shows up early. A tool with solid AI Pulse numbers but a falling NPS, or a rising number of people who know of something better, is telling you something worth acting on before the invoice arrives.

Pulse 3 · The AI Capability Check

Default cadence: once a month, per person, per tool. Time to answer: about a minute.

The people-focused pulse. Same tool-by-tool scope as the other two, but the questions are about the person rather than the software: how confident they feel, what is limiting them, and what would help.

1. Confidence. How confident do you feel using [tool] for your work? (Very confident through to Not at all confident)
2. Limits. What, if anything, most limits what you get out of [tool]? (I haven't had time to learn it, I'm not sure what it can do, I don't trust the results, It doesn't fit my tasks, I don't have full access, Nothing, I'm getting what I need)
3. Unmet need. Is there something you'd like to do with AI at work that you can't today? Answering yes opens a free-text follow-up asking what that would be.
4. Support. What would help you get more value from [tool]? (A short demo or walkthrough, Examples relevant to my role, Tips and best practices, Time set aside to experiment, Nothing right now)
5. Direction. Compared with a month ago, how much are you using [tool]? (More than before, About the same, Less than before, I'm new to it)
6. Anything else. Free text, optional.

This is the pulse that tells you where learning would pay off, and it surfaces the tools people want that you have not bought yet.

Who gets which pulse

The AI Pulse is on every plan. The AI Review and The AI Capability Check need Standard or above. On Lite they show on the Pulses page as locked, and they do not fire.

When a pulse appears

Delivery runs on cadence and nothing else. There is no page-visit trigger and no dwell-time condition.

The Chrome extension holds each person's schedule locally and works out the firing moments for the current period, spread across it and stable across browser restarts. When a moment passes, the pulse is offered on whatever tab is active at the time. Often that is the tool's own website, because that is where people are, but it is not a condition.

A few things shape the timing:

• The organisation's working window. Firing moments are snapped into it. The default is Monday to Friday, 9:30am to 4:00pm, and you can change it in Settings.
• The AI Pulse also keeps to weekday afternoons, because it asks about "today" and the day needs some substance to report on. The other two can fire any time inside the working window.
• Chrome's own pages cannot show a pulse, and nor can any page if the person has not granted the extension page access. In both cases the pulse waits in the extension's toolbar inbox rather than being lost.
• A pulse that goes unshown for 24 hours expires rather than piling up.
• Skipping puts a pulse in the toolbar inbox for later, and the snooze control quietens everything for a while.

Nobody is emailed or notified about a pulse. The extension is the only place a pulse can be answered, and people can open one themselves at any time with Take a pulse now in the extension popup.

What you can change

Core pulse questions and names are locked on every plan, Pro included. The three are the baseline every GAiGE organisation shares, and that shared baseline is what makes a number from your organisation comparable with anyone else's. An attempt to edit the questions returns a clear refusal rather than saving quietly.

Cadence is yours to set. The pulse builder shows a Reset to default link beside the schedule whenever your setting differs from the recommended one, so you can always get back.

If you want a core pulse with an extra question, or with softer wording, Pro plans can clone it and edit the copy freely. The original keeps running untouched. See Building a custom pulse.

Expected response rates

The extension is the only surface where a pulse can be answered, so your response rate can never be higher than your install rate. Getting the extension onto people's browsers is the biggest lever you have.

For what counts as a healthy rate once the extension is out, and the three things that most often hold it down, see Why is response rate low?.

Related: Building a custom pulse · The pulse library · Why is response rate low? · Rolling out the extension to your team