Summer release: AI, To-do's, Conversations, Workflow, and more

Summary: A ton of ideas may sound familiar to you. Likewise, we as a team have a lot of ideas we want to implement. We have been busy the last few months creating new features in RUMvision! Read about the features we've included in our summer release!

  • by Jordy Scholing, Karlijn Löwik, Roderik Derksen & Erwin Hofman
  • Published
  • Reading time ± 5 minutes
  • Update
Summer release: AI, To-do's, Conversations, Workflow, and more

Our goal with the summer release ☀️

The features included in the summer release contain mainly useful features in the area of collaboration and being able to work even better with RUMvision. A brief summary of what you can expect in this article:

  • OpenAI dashboard assistent: In the field of AI, we obviously cannot be left behind and truth to be told, it is super convenient.
  • Workflow: To collaborate with your external or internal team members, we have introduced a to-do module.
  • Conversations: To keep your entire pagespeed workflow centralized in one place, we created a module where you can chat with team members or possibly RUMvision support.

Noteable technical changes 🆕

With new features also come many technical adjustments, which in turn result in other new ideas:

  • Identify and anonymize PII data: Other (marketing or third) parties could still add PII to specific (often utm) parameters. We advice to add additional expressions to replace potential PII data that applies to your activities and audience. A first identifier for e-mail addresses is automatically added.

  • Tech Stack: A platform, CDN, hosting or even plugin provider might expose additional (often) TTFB or even server location related information. RUMvision is then able to collect them for either dashboarding or dimension purposes.

Experimental features 🧑‍🔬

A group of RUMvision users are already participating in the experimental features, if this is something you are interested in please contact info op rumvision punt com. Currently the following features are experimental:

  • Long Animation Frames (LoAF)
  • Back_forward cache filtering: Find out which elements could not be restored from the BF_cache and the reason why.

We will cover the above points more in-depth below, so be sure to keep reading if you are curious about how to use it all!

OpenAI dashboard assistent

With our AI assistant, you easily navigate through the dashboards and can specify prompts to see data from a particular filter. These are a few examples of prompts that can be used:

  • Show me INP of last month
  • Did my CLS regress on Google Chrome in the 28 days?
  • Compare Google Chrome TTFB of last month with the current month
  • Show the 75th percentile LCP on high-end devices by template type

But you can also make up your own prompts as long as it falls within the scope of RUMvision. Anyway, we are curious to see how creatively you will use it and if you like it!

The screenshots above show how it works and what the results are. For example, in the last screenshot (technical dashboard) you can see that that of high-end devices itself already knows that this is 8192 MiB device memory, pretty clever right?

Your performance workflow in one place

Optimizing pagespeed and Core Web Vitals requires collaboration between a development team, marketing, product owner, or even an independent consultant. Work together on to-dos, watch how they perform live, and discuss issues, ideas, and solutions.

To-do's and annotations

For the workflow, we created our own module for to-do's. This allows anyone with access to the domain to create tasks and tag people. You can also tag important things with a to-do:

  • Description (via Markdown)
  • Affected domains
  • Status
  • Affected metrics
  • Impact level
  • Affected templates
  • Labour intensity
  • Expected completion date

This way you can create tasks for your development team or have them created by, for example, a page speed consultant working for your company.

Above is an indication of how we see clients using to-do's. Are you adding them too?😉

And extra convenient, you can add them as an annotations when they're done. This makes it easy to compare whether this release has had the expected impact on your Core Web Vitals and pagespeed.

Once you have sent a to-do to your annotations they will show up in a calendar view where you can easily select it to then compare it to before and after the annotation. This way you know exactly if it has had an impact on the expected metrics and templates.


You may need to consult with several people from the same or even a different organization. No doubt this already happens via Slack groups as well, but should this not be the case you can easily chat, share files, discuss certain to-do's, annotations and metrics in RUMvision conversations. This keeps everything in one place and saves you an unnecessarily long email chain. A powerful addition to annotations and to-do's.

More insights for you tech stack

In addition to our existing support for platform and CMS data gathering, we are excited to announce that we are now also collecting information from a broader range of Content Delivery Networks (CDNs). Previously, this was something we only did for Cloudflare, but now we are expanding our reach.

Should another CDN begin to automatically implement server timing, we can easily integrate all the domains we already know are running on a certain CDN. This enhancement simplifies our approach and allows for even more insights.

If something is linked to the selected CMS, CDN, or the like, you will immediately receive notifications for the chosen values (see screenshot above).

Technical: Experimental features

As already mentioned, some domains make use of our experimental features. If this is something you would also like to do please contact us. For the experimental features we are dependent on Google Chrome origin trials and our own roadmap. Thus some features will not be permanent. Currently we are running experiments for the following features.

Join experiments 🧪

Long Animation Frames (LoAF)

A brand-new performance entry type called LoAF (long animation frame) has been proposed and is meant to be a development of the long task idea.

It is the amount of time between when the main thread began working and the time it takes to be either ready to paint or idle (having nothing to do). There could be more than one task, but typically there are only two. It includes all of the rendering observer callbacks (requestAnimationFrame, ResizeObserver, etc.) because it ends at the paint-mark time, but presentation time (also known as "pixels on screen" time), which is an implementation-specific term, may or may not be included.

Getting valuable insights via filters and sub-filters about LoAFs is quick and easy with RUMvision. Although it is still a very technical metric at this moment. But if you're willing to try it out than request access to the experimental features!

Back/forward cache NotRestoredReason

Back/forward cache (bfcache), a feature that improves history navigation, is available in modern browsers. This makes it possible for users to instantly load previously visited pages.

Today, pages may be prevented from entering bfcache or removed while in bfcache for a variety of reasons, including those mandated by specification and those unique to the browser implementation. The complete list of causes that can be reported is as follows: spreadsheet.

Information provided by Google

Filters and data for showing problems of the BF/cache looks like this in RUMvision:


Goals of Back/forward cache experiment

  • Provide a way to gather data as to why a page is not served from bfcache on a history navigation.
  • Provide an easy way to debug a website and make it bfcache compatible.

Share blog post