I saw the page. A couple of the things that I remember seeing were:
If I remember anything else I’ll post a message.
No, there is DNN in the chip, but it is fixed after training What is a Deep Neural Network?
Would be nice if developers read this Forum. What I see is users screaming for sturdy (sweat-proof) HAs with LE Bluetooth, telecoil, and rechargeable batteries that comfortably last a full day no matter what you do, and that can easily be replaced after a couple of years. AI will demand more powerful microprocessors and also swallow R&D costs (making HAs even more expensive), when we still need to see if it contributes anything.
Thank you @Neville for those snippets of information, I am as usual very much obliged 
Cheers Kev 
https://ow.ly/kC4250SKYex
Found through LinkedIn
Use this, it allows you to see website snapshot before it was taken down…
The number thing users want is speech in noise.
Lumity is pretty good with speech in noise. The real challenge is understanding speech in noisy environments where there are many other conversations going on nearby. This is basically the noisy restaurant scenario and Lumity often locks onto the loudest conversation vs the person you are talking to. Should be interesting to see if the new AI chip is able to offer some improvement without chewing battery life. Lumity definitely eats the battery faster when in speech in very loud noise mode. I’m wondering if the new platform is physically larger to accommodate a larger battery.
Jordan
I will be happy for improvement in speech in noise as most others. I was at a wedding reception Sat and it was dreadful. I was last to be seated at the table and as a result no choice in seat. Facing the speakers which were blaring music most of the time. So loud! And it masked all the voices around me. I even had trouble with my wife’s voice, next to me. She interpreted for me with the wait staff. I went to the car to get Roger and it helped some. The women’s voices were so shrill and the men’s voices muddy. I’m coming up for exam/testing next month. Maybe we can talk about this at that time. Roger program might need tweaking. (No complaints about hearing the wedding itself. I even heard the groom’s mumbling!) Others commented that he has a quiet voice.
WH
Yes absolutely. Also it is rechargeable to support increased power demands.
Time will tell if they will eventually be able reverse engineer that down to a smaller traditional battery, but I expect not before people here cry conspiracy.
If they manage to extend the battery life to 36 hours and provide a noticeable improvement to speech in noise, I’d be tempted to upgrade my L90’s.
Jordan
Anyone that really needs hearing aids to understand what is said wants aids that give them speech understanding. The ones that are brainwashed with connectivity just realt want earbuds.
I NEED to understand speech, connectivity is at the bottom of my list of importance just above wanting to swim with my aids on.
I agree. I found that using the standard fittings (not moulds) I could use Apple Ibud type earbuds over the top of my aids. Not ideal for audiophobes but pretty much OK.
Euan
Why do people want LE Bluetooth? I have Signias with those and LE Bluetooth sucks, unequivocally. My Lumities connect to the app via LE but the streaming sound and connection stability via standard BT is so much better. To me it would be a strike against upgrading. I use iPhone, so not dealing with ASHA. I gladly sacrifice battery life if that’s the issue.
LE Audio has enhanced capabilities over what traditional classic BT can do. It is a completely new standard and should be a big improvement. I expect when they do it, they will add LE Audio and not take away the classic because so many devices don’t support it yet.
WH
Improving connectivity and rechargeable battery capacity, and battery “swappability”, are things that are fairly easy to implement. But I remain extremely sceptical that AI can somehow help us with speech intelligibility or magically enhance the “cocktail party effect”. I wouldn’t spend my R&D money there (R&D is a major factor why HAs cost so much, not the components, and it us paying for this, whether it delivers or not.). Filtering out one voice from a set of similar speakers requires understanding of what is being said, context, and having a few good guesses of where your speaker of interest is going to. This interpretive activity is purely cerebral, happens between, not in our ears. Happy to be proven wrong, though!
Are you talking about Auracast or mfi/ASHA when referring to Bluetooth LE / LE audio It gets confusing!
Auracast falls under “technically achievable” and connectivity- yes! My point is I am very sceptical about AI and its return of investment from R&D.
Yeah but in marketing these days everything has to be AI. AI is one of those things I think in hearing aids that will be iterative through several generations until suddenly it seems magical.
They are bringing AI on chip de-noising as directional microphones have plateaued, so this is a game changer. Yes, other manufacturers have been using DNN tech but not in this way.
If it delivers and battery life isn’t terrible, you’re probably looking at a substantial lead for phonak in signal processing for noise for quite some time.
They have been developing this for a long time, so any response from competitors is years away: