"B2B" used to mean "Business to business" but I think it will come to mean "Bot to bot."
https://x.ai/news/grok-voice-3 - the latest voice calling / answering service
There was already the joke where person A tells their AI agent to draft an email explaining topic X in more detail; and then person B gets the email and has their bot summarize it, throwing away the detail.
Now, your agentic voice calling service can call the agentic voice answering service at your cable provider (for example). The two bots can negotiate a new deal on your internet plan.
I assume the next step is that the calling bot will emit a tone that says, "I'm a bot, can we talk binary please?" and old school modem tones will fly back and forth carrying JSON packets.
Google has had a voice service to make restaurant reservations for some time; I never tried it. I think the only agentic financial transaction I've done was to get Muse to buy some night lights for me from Walmart. An eight dollar transaction and that was scary enough.
My agentic usage went from zero - meaning I only did copy/paste into chatbots - to having about 10 projects ongoing in parallel. I'm not to the point where I send a bot off on a multi-hour task. My agents complete subtasks while I'm watching and those tasks are programming or computer management related. When I say "in parallel" I mean that they are "in-progress" and available to do more work when I think of the next task I want them to do. They have context they remember. A few of them have worked together - more about that later. It's infrequent that I have my agents doing multiple tasks at the same time, but that has started for me. Computer maintenance tasks, while potentially disastrous if they go wrong, especially in terms of security, are also the easiest to track what the bot is doing. And there is a lot of training on how to administer computers these days so they don't mess up too often.
I don't use AI for writing, except to make reports that I comment on in my own voice. I recently had a bot spend a week of calendar time, yes, a week! looking at all 1300+ posts here for link-rot and other errors. Probably the actual execution time was a few hours spread over that week. It's something I could never do - I've thought about exporting the blog, writing scripts to check the links, but even that would be problematic, as a lot of link-rot is links that get redirected to a homepage, rather than erroring. I'd have to look at each and every link that 'works' to see if it really works. My agent did that for me. I told it to go slow so as not to trigger any rate limiting - that's one reason it took a long time. But eventually over 1,100 blog posts were touched in some way - sometimes trivial updates (removing old footers that didn't make sense anymore) and others where features from Picasa, long dead, had to be recoded from scratch (e.g., slide shows).
So much work I would never have done on my own.
I am nowhere near the state of the art in agentic usage; I know people with multiple $200/month AI subscriptions who run big programming tasks overnight. And some businesses are spending $5000 a month on agent credits per employee.
I think I sound here like an advocate for AI agent usage; but I think my real point is how fast I went from no usage to 10 personal projects - about six weeks. All that stuff about worrying about hallucinations is much reduced because the better bots check their work.
They can still screw up - check out this typo.
So it's prudent to start slowly and backup everything in advance. And ask the bot to check their work! In fact, ask them how they are going to check their work! Ask them what their plan is before they do it! Ask them to test every assumption! The best bots do this automatically but on the free plan you're probably not getting the best bots, so ask them!
The speed that AI adoption is occurring for certain technical tasks is stunning. AI adoption has grown faster than the internet!
You personally might be anti-AI (because of IP theft, datacenter electrical and water usage, or you're just anti-change) but the odds are high you're already using a bots without knowing it. I wouldn't say I am pro-AI because that would endorse way too much! But I'm definitely pro-learn-about-AI.
