Introducing CoBrowse AI Beta on AppleVis and Inviting Community Participation and Feedback

I can’t see the form

Are you viewing this thread using a content blocker, or in 'Reader' view? That could be causing the form to not display for you.

Hi Abdulaziz,

We received your participation request (along with the others sent thus far) and will be in touch with next steps later this week.

Best,
Michael

Hi, I just signed up for the beta. What are my next steps?

Hi Zac,

First, thank you for signing up for the CoBrowse on AppleVis Beta!

Towards the latter part of this week, we will be reaching out to beta testers with some tips and tricks to help users get started with CoBrowse on AppleVis. Once that initial information has been sent out, we will activate CoBrowse on your AppleVis account.

Thanks,
Michael

Where is this feature exactly? Is it currently on the website? Before I saw this post, I actually got an email asking me to participate, but since I didn't find the task anywhere on the website to do this, I never paid much attention to it

Really interested to see how this AI integration with websites and voice control works
So I signed up!
I mean I have so far, never seen any websites offering controls with voice before.
This would be great if it takes off.

Hi Winter Roses,

Thank you for the questions.

For the time being, CoBrowse AI is enabled by us on a per-account basis; this was done so that we could test the feature internally and have a controlled rollout. Should we move forward with a pilot program for all AppleVis users, it will be enabled for everyone who is logged into the website. We did do an initial invite-only beta earlier this month, with the invitee list drawn heavily from those who participated in the recent discussion of ChatGPT Agent (as CoBrowse AI has some similarities with that technology). We are finalizing the getting started materials and hope to reach out to testers in the next couple days.

I've enjoyed playing around with the other Innosearch oferrings thus far, and the team behind it seems solid. I'd love a shot at participating in the beta!

From what I’m reading on this blood post, it seems like you’d have to have prior knowledge of a website layout in order to give this AI commands. I don’t see how that’s useful with somebody who is looking at a brand new website that they’ve never explored before to be able to tell the AI where they need to go. You still have to do initial keyboard commands to explore a website before you know what to give the commands to the AI I would think. I’m not sure how that’s going to be that but for somebody who is just learning a website. I mean, it might be OK for people with this website if they already know it, but I don’t know how that’s gonna make navigation easier for people if it’s a brand new website that they’ve never explored before. Maybe I’m not understanding it right, but that’s the way that it seems like it works so far.

Agree with singer girls observation. Let’s be very clear that cobrowse AI only works on already fully compliant accessible and usable content and controls on webpages. It is not designed not as capable I believe, to work with any portions of a page where either the content or any of the control elements are inaccessible or not usable with screen readers. this is unfortunate however indirectly alluded to by Michael Hansen in his original post. Because it only works on content and control that are already fully Accessibility and usability compliant to screen readers, I personally don’t care to test it because I don’t find it helpful. I am a semi decent jaws screen reader user and on formatted accessibility and usability compliant pages, just using jaws alone is pretty quick and efficient to navigate. However again as the original post says, this really is for those who are weak in jaws screen reader usage and not proficient enough and therefore browse AI is an intermediate two to interact with an accessible and usable page for them.
Initially I was hoping that this is in his current form or at least there is some sort of a roadmap for the development of Cole browse AI where with some of the state of theart AI technology it would start making inaccessible content and controls on webpages not just readable but also allow clicks on inaccessible controls however alas, this is not the case. I fully appreciate how this tool can help those who don’t have screen reader proficiency. So kudos to the development team for coming up with this product targeting that sub user base inside the screen reader using blind population.

Hi all,

For clarity, we only have information about the implementation of CoBrowse on AppleVis. The presence of CoBrowse AI on AppleVis should not be taken as a statement, either express or implied, about what types of websites the technology may or may not work on. If you have questions about CoBrowse AI outside of our platform, please contact InnoSearch directly.

On the point about requiring prior knowledge of the website, I don’t believe this is true.
You can simply start it up and ask “What is this website about?” or “What are the main sections on this page?”
This is a beta though so we will understand more about how useful it is or is not through testing and learning. We want to know how useful a tool it might be in real life.
Dave

Is there gonna be a separate thread for individuals in the program? What I mean is, I know that we can submit feedback through the email directly, but will there be a separate thread, like, on here directly, where members who our participating in the beta program can talk to each other about what is working, what's not working, and to also share tips and tricks? I'm sure that there will be plenty of feedback, but even so, what will most likely get prioritized is what the majority things or what is realistically viable to implement. I figured that since this is gonna be a team effort, shouldn't there be some public discussion, or, I don't know, like a private thread for us to talk about the experiences in real time, inside and outside of the official feedback emails for the participant. Not sure if this would work, but thought I would make a suggestion, just in case you haven't already thought about this angle. Are you still accepting submissions? I mean, members could talk about their experiences on this thread if you're gonna keep it open anyway

Hi Winter Roses,

Great question. We envision this post being the main place for public discussion of the CoBrowse AI on AppleVis beta, be that by users of the beta or anyone else who has an opinion they wish to share. That being said, please feel welcome to share your experiences with us or the community in whatever format you wish.

While I don't feel CoBrowse AI substantially helps my current web browsing, I'm still willing to give it a try, as this technology opens up endless possibilities for the future. However, I don't primarily use English, and I've seen in posts that this tool also supports other languages. So, what languages does it actually support?

Like the subject said, I believe those people who are very proficient in screen reader use can also benefit from it.
An example is looking for previous posts.
For example, I have found a use case for this during this beta testing period is looking for threads with no update for a long period of time.
I can just say.
Open the latest thread about discussion about iPhone 17 pro max”. Or “open the last thread about discussion about the Mac” I believe no matter how proficient your screen reader skills are, searching for those topics still consumes a lot of time.
Hence I believe this is an extremely powerful feature and anybody in the Vi community can benefit from no matter your skill level.
Imagine if this kind of features were implemented on government websites.
You can just say “submit a health declaration form” it bring directly to the link for the form instead of having to look through thousands of other items you can already do on the website with a screen reader only to fail due to session time out.

well, I t hink ti is a good start.
and it is very good features.
I tried it for 30 minutes and it is really cool.
I hope it can be released soon

Hi Carter,

According to the team at InnoSearch, CoBrowse AI supports 32 languages:

• ARA Arabic
• BUL Bulgarian
• CMN Chinese
• HRV Croatian
• CES Czech
• DAN Danish
• NLD Dutch
• ENG English
• FIL Filipino
• FIN Finnish
• FRA French
• DEU German
• ELL Greek
• HUN Hungarian
• HIN Hindi
• IND Indonesian
• ITA Italian
• JPN Japanese
• KOR Korean
• MSA Malay
• NOR Norwegian
• POL Polish
• POR Portuguese
• RON Romanian
• RUS Russian
• SLK Slovak
• SPA Spanish
• SWE Swedish
• TAM Tamil
• TUR Turkish
• UKR Ukrainian
• VIE Vietnamese

This sounds like an incredibly useful tool and I hope it goes very well--I know someone who actually couldn't make heads or tails of this website and finds it unusable, so this sort of development is very much needed. However, I don't know how I feel about AI being baked into a page that has a great deal of user generated content, including comments. Personally, I am furious about the amount of data that large language models have stolen and the ways in which it has repeatedly violated artists and creators in taking/using their work without consent. As a result, I do not want to contribute to that with my own writing. This is not to say that I am against it--accessibility is of vital importance--but I would like to know a bit more about how any data encountered will be collected and utilized, if that option is available. Anybody else in this boat?

I’ve been using CoBrowse for a few days now and am finding it quite useful. The typing mode works very well for me, and I’m getting great results with it. CoBrowse is an excellent way to quickly browse a site when looking for specific information.

However, I’m not sure if I’m using the voice mode correctly. When I activate voice mode on my iPhone 16 Pro Max, it doesn’t behave as I expect. After granting microphone permissions, I hear that CoBrowse is listening. But when I give it a command such as “summarize the page” or “tell me about the latest posts” I hear sounds, yet the text on the screen doesn’t change.

I have speech turned off, as the instructions suggest, but nothing happens after I issue a command. Is the voice mode supposed to speak the results aloud, or should I turn speech back on to have them read? Either way, I don’t see any change after giving a command.

As mentioned, I’m using an iPhone 16 Pro Max running the latest version of iOS. The typing mode continues to work perfectly, and I’m really finding it helpful. I’d just like to get a better handle on how voice mode is intended to function.

I plan to test CoBrowse on my computer to see if I get different results.

When you use CoBrowse with voice mode, you should be able to talk with CoBrowse to give instructions and then hear the CoBrowse voice tell you what it is doing and give you an answer to any query you gave. After you give your prompt or instructions, there should be a continuous beeping sound indicating that CoBrowse is "thinking". then you will hear the response.

I find it convenient to turn off any speech from my PC screen reder or Voiceover when using the CoBrowse voice mode since my screen reader or Voiceover may sometimes echo CoBrowse's responses that are displayed on the screen. I believe they are working to address this issue, but it isn't consistent.

You can also see how CoBrowse works on other web sites with different content. We have recently integrated CoBrowse into the Eyes On Success web site at:
www.EyesOnSuccess.net

there you can search for episodes, play them, get descriptions, etc. by using CoBrowse.

CoBrowse is a work in progress and can sometimes be a bit slow to respond and a bit clunky. But it is quickly getting better and feedback from users will help. Once you get used to using this AI agent it gets a bit easier and can be more efficient for some tasks.

Enjoy!

--Pete