This year, one of the most useful things I built wasn’t a new help article or a shiny launch page. It is a fairly unglamorous workflow that connects our help center, our CRM, and AI Studio so customer feedback can actually make its way into our day-to-day work. It’s already saving me time and context-switching, and it’s something I’m planning to lean on – and build on – a lot more next year.
We run the Help Center. It’s a whole bunch of articles, video tutorials, and how-to guides in about a dozen languages. There’s a lot of content to manage and a lot of people who read it and rely on it.
In total, there are well over five thousand unique pages involved. “Wow, 5,000 opportunities to learn?”, I hear you say. Sounds optimistic. As more of a part-time optimist myself, I tend to see it as 5,000 opportunities for things to go wrong.
And that’s where we actually rely on you, the reader. Stuff will go wrong, sometimes. At the bottom of every help center article is a little box with a pre-loaded question: “Was this article helpful?” Any logged-in user can leave feedback and let us know what they think.
- Maybe they’ve spotted an error, a typo, or something that’s gone out of date.
- Maybe they were expecting more visuals or more detailed steps.
- Maybe they just want to say it was helpful (less common, but it does happen).
Not all of the feedback we receive is actually actionable for the help center team. Some of it isn’t relevant to the article at all. We get comments about open tickets, billing questions, product issues, and everything in between. That’s totally understandable, but isn’t part of our scope.
One of the things I’m using AI Studio for is to separate the feedback the help center team can actually use from everything else, and move the rest to the people who can act on it. For the purposes of this post, I’ll stay focused on the actionable help center feedback.
The end-to-end flow in plain language
Here’s the full loop, at a glance:
Customer leaves feedback on a help center article.
An Asana task is created automatically with the relevant details, thanks to an integration between our CRM and Asana.
AI Studio triages the task by:
- Filtering out irrelevant or non-article feedback
- Translating non-English content into English for consistent triage
- Capturing the helpful / not helpful vote for reporting
For actionable feedback, AI Studio:
- Creates a subtask that fits into our existing Help Center workflow
- Preps a first draft of any necessary copy changes
- Includes a link to the affected article
- Outlines recommended actions
That subtask then flows through our usual Help Center process, alongside all the other work.
The result is that feedback doesn’t just sit in a CRM or a spreadsheet somewhere. It enters the same system we already use to run our documentation work. Read on to take a look at the workflow in more detail.
From everything in the CRM, to everything in Asana
I work at Asana. I use Asana all day, every day. Most of my work lives inside Asana. That means I’m spoiled, and I’m easily irritated when I have to switch to another tool to see what’s going on.
The good news is that Asana integrates with a lot of other tools. So I set up an integration-based rule that pulls information from our CRM directly into Asana whenever feedback is submitted.
For each piece of feedback, a task is created in my project with all the context I need:
Article title- Link to article
Feedback text
/ 
Vote
Language
Creation date
At that point, I’ve solved one problem (everything is in Asana) and created another: now I have a large volume of feedback tasks flowing into a project, and I still need to triage them. That’s quite a lot of admin, and coupled with my regular workload, suddenly it feels like spinning plates if I want to stay on top of everything. This is where I need to rely on AI Studio.
What AI Studio does for me
AI Studio runs a smart workflow in my feedback project that:
Automatically reads each feedback task, checks if it’s relevant, translates if needed, summarizes it, and routes it so I only spend time on the work that’s actually mine.
There’s no code. Just a few AI Studio rules working in the background.
It’s tempting to build a single rule that does it all; a one-shot, multi-step rule. For me though, that sounds like a house of cards, particularly when trying to build this workflow from scratch. I prefer a couple of simple rules, chained together by employing one rule’s action as the next rule’s trigger, at least early in the process. But that’s just me, I encourage you to experiment with AI Studio as you see fit.
The intake project and custom fields
First, I have a dedicated project with custom fields that make triage easier: Vote, Language, and Relevance.
There are a couple of additional fields in the background that help with reporting and routing (for example, whether it looks like a billing/support issue instead of a content issue), but these are the main ones I look at day to day.
Triggering the smart workflow
Once the rule integration is set up to pull information from the CRM into an Asana task in your project, you can simply use a Task added to project trigger to begin your triage workflow.
You may need help from your CRM administrator to set up the rule, depending on your permissions in the CRM software.
Whenever new feedback comes in from the CRM and creates a task in that project, the workflow is triggered. No buttons, no manual steps. It just runs.
Using AI Studio to filter and translate
The workflow has a small sequence of AI-powered steps that run in the background:
Relevance check
- Is this actually about a help center article, or is it something like “please fix my invoice”?
- If it’s clearly not documentation-related, the workflow can route it away from the help center and towards the teams who can act on it.
- I chose to build the relevance check into the workflow as an initial step so that I don’t waste AI Studio credits translating text or creating draft updates for something that isn’t relevant to our team.
Translation
- If the feedback isn’t in English, AI translates it so we can triage more effectively
- My instructions specify that the original feedback is retained, in case of any discrepancy
It’s at this stage of the workflow that those messy, hard-to-read, contextless pieces of feedback - previously hidden away in another software system - start to feel like a jigsaw falling into place.
Creating an actionable subtask
For feedback the help center team can act on, AI Studio creates a subtask that plugs straight into our existing Help Center workflow.
That subtask includes:
- A short AI-written summary of what needs to change
- A direct link to the affected article
- Draft copy or notes to speed up the update
- A clear outline of the actions to be taken
Because this is a subtask, rather than a comment for example, it can move through our normal process by being multi-homed into our regular projects. We don’t need to invent a new process just for “AI-powered feedback.” It’s simply another work item, with a little more context automatically filled in.
How this actually changes things
The nice thing about having feedback in the right place, with the right structure, is that it seamlessly becomes part of our normal work, not a separate “when I have time” stream. We can see actionable customer feedback right next to our usual work items, apply a prioritization matrix just like any other work, and spot high-impact feedback at a glance instead of hunting for it.
At this point, all the building blocks are in place: relevance filters, translation, first drafts, and actionable subtasks. We’ve tested these AI-powered steps enough that we trust them for everyday use. At the end of the day, though, a human in the loop makes the decisions.
I hope I’ve shown how AI Studio makes our team’s customer feedback process more efficient. The volume of feedback hasn’t changed significantly since this workflow was implemented, but the time spent manually triaging definitely has.
