Incredible thoughts of the week #55
It's been a long time, I shouldn't have left you, without a strong rhyme to step to..
To be honest, I’ve had maybe two or three more sizeable topics I’ve had tee’d up to write about for a hot minute, but writing is just so 2024. Plus (and maybe the real reason), I’ve just not had the time to sit down and write. It’s taken a three hour train journey to force me to sit still and actually put pen to paper (digitally, obvs). I’ve split the content into two newsletters, entirely because I had no faith that anyone would sit through the prospect of all of it in one go.
Undoubtedly, the biggest theme in my writing of 2024 was gen-AI, but it seems strangely fitting that I’ve been brought back here primarily to write about Amazon commercial models again (a topic that was surely amongst the least popular with readers, but still one of my favourite topics I wrote about last year).
🤖 🛍️ Alexa hits the Prime time
🕵️ 💻 Archer rules the web
Alexa Plus
Amazon have, finally, announced an update to their Alexa devices with modern LLM capabilities. Since ChatGPT first exploded into our lives, an incredibly obvious application of the technology has been smart assistances.
I really like my Alexa devices and use them on a daily basis. Sure, it’s for incredibly menial tasks: 1) playing music1 2) setting timers whilst cooking 3) checking the day’s date; but they get daily usage and would definitely replace them if they were to die. If I could then add to the mix the chance to have random conversations, to be able to ask it actual tip-of-my-tongue questions “what’s that song that goes like this?”, “whats the name of that film with that guy who’s also in this other thing?” and actually conversationally probe it to get answers, it’d get a lot more use. The frequency with which I reach for my phone whilst watching something on TV to google the cast to work out where I recognise them from, or to ID something on the soundtrack is quite astonishing.
So, great, Amazon are finally rolling that out. A fun distraction, but.. so what?
If you read my previous Amazon ramblings, you will probably know where this is going..
Free with Prime.
One thing that has been oft observed by analysts is that Amazon have never really found the purpose of Alexa. How do they monetise it? What does it bring to the business? Yes, if you go along with my vision for the Amazon Smart Home of the Future, then it makes some sense in that context, but that’s a pretty long game and outside of that, it feels like a product struggling to find commercial fit. It has a long upgrade cycle (compared to the ~2 year average cycle of smart phones), is low cost/margin and doesn’t currently have any real means to generate Amazon any significant or recurring revenue.
With this LLM upgrade, and with broad consumer acceptance that LLM platforms charge for premium plans or to use them without limits, it has opened the door for Amazon to charge a subscription price. Alexa+ will be being charged at $19.99 a month on launch (UK prices TBC), whilst being free for Prime customers.
I don’t think I need to get into this, again, but this is the beautiful thing about the Amazon model. They aren’t a regular subscription platform whereby they have to rely on the recurring subscription revenue for digital products (live sports, music, movies and now LLM usage), because they have their underlying e-commerce business that can supplement the revenue as a primary source, and those subscription products are really just hooks to drive customers to the Amazon shopping platform.
When it launches, in America you will have two options:
Pay the $19.99/month Alexa+ subscription
Sign up for Amazon prime at $14.99 and get Alexa+, movies, music, sport and free next day delivery
There will be a decent chunk of consumers who will be willing to pay the Alexa+ subscription price for that service (much like there are large numbers of people paying for personal Pro subscriptions to ChatGPT etc), and when offered the chance to get the Alexa+ subscription plus a whole lot more, for cheaper, you’d have to assume most people will just sign up for Prime instead, and that’s exactly what Amazon want.
They understand the Life Time Value of Prime customers at scale. They know that if they convert another big segment of the market to Prime customers just to access Alexa+, they won’t just get the subscription cost, they’ll get an increased shopping revenue from those folks too. It seems, at last, Alexa might just be finding a commercial fit.
Agents
The future of the web and UI design, is something that I find super interesting.
I’ve talked specifically about LLM interfaces and what it might mean to potentially transition to a purely voice interface as primary interaction models, but right now I’m thinking quite a bit about the rise of AI agents. Agentic AI definitely appears, right now at least, to be the direction of travel for gen-AI tooling, which has some somewhat subtle implications for my previous view of whether all products become voice based.
That previous version of the future I described was one where LLM technology was pervasive and the de facto interface and interaction model for the majority of websites, applications and products (e.g. each app or product had LLM capabilities embedded in it, allowing end users to navigate and describe their intent with natural language). This definitely seemed a plausible scenario at the time, given the rush with which so many platforms and apps were going out to add LLM tech to their products in any way they could.
But the shift towards a focus on agentic AI changes that, as it re-frames the whole problem. Instead of every app or product needing to have built-in support for natural language and LLM technology, products could remain largely unchanged and the LLM technology simply abstracted to a layer above in the form of an intelligent agent.
I recently read somewhere the suggestion that things like accessibility concerns could be a thing of the past, as it would be an agent interacting with the websites/products and they could then present whatever we needed it in any way we needed (I would link to the article, if I could vaguely remember where it was).
On this topic, I also recently read an interesting LinkedIn post about ads and agents which was really up my street. The gist of it was basically: if AI agents are soon going to be doing our bidding, what does this mean for the web? what does this mean for ads? if we passing purchasing power and decision making to agents, does it mean the end of ads as we know them?
Love them or hate them, the advertising ecosystem props up and monetises large swathes of the internet, if we are talking about a future where these sites are being actively visited by AI agents instead of humans, how can we optimise for that? Marketing campaigns, nudge psychology, FoMo - these things are based on psychological, emotive responses from humans, which just aren’t going to work the same on AI.
This general theme is quite broadly applicable, I think - what happens to the web if we eventually transition to personalised AI agents? How should we be building tools and websites for a future that is dominated by AI agents? Is there a viable commercial model that supports the internet as we know it that fits with AI agents?
Lots to think about (and more to come soon)..
Sidenote: I think an aspect of the Alexa et al revolution that isn’t much talked about is the impact they have had on Sonos. For a while, if you wanted connected home audio systems, Sonos was the go-to, and there really wasn’t much else competing in that space. They looked like they were on their way to becoming an Apple-esque status symbol for the home. The reality was, though, they weren’t just multi-room speaker systems, they also included fairly high end audio equipment, that should be sitting along side professional grade turntables and speaker setups with eye watering price tags. The likes of which would normally only be of interest to serious audiophiles with cash to spare. They seemed to successfully get a foot into the regular consumer market by being the only real player in multi-room connected speaker systems game. But the arrival of smart devices like Alexa changed this landscape, because, of course, the majority of people don’t actually care about high end audio (despite what they might tell you) and faced with a device that costs less than a round of drinks and can immediately start streaming any song with no real setup, most people chose cheap convenience. I was looking at multi-room Sonos set ups for a while, but then I bought a couple Alexa devices (one that cost me just £1.99 in a Black Friday sale) and haven’t looked back.