Very soon, you’ll be relieved to call a company number and get an AI. But honestly, that experience is few and far between right now. At speed-to-lead.ai, we are working towards a future where AI allows small businesses to treat every customer like a VIP. But… We shouldn’t forget the power of a human interaction, either. While Uncle Bob may be in the small minority that’s ready to talk to a voice AI, 33% of your customers aren’t. You can’t afford to risk your reputation like that. Speed-to-Lead AI is the balanced approach. Phone call? That goes to you. If you can’t answer, an AI texts them back to salvage the lead. Web forms, emails, and DMs go to the AI, but it can be paused at any time if a human is available to handle the convo. That’s the model we believe in. Human first, AI to fill the gap. One day, the average person will be ready for voice AI in small businesses like Uncle Bob is. And when they are, we’ll have voice support ready for our clients. But small business runs off of personal connection. So we’ll keep building the AI tool that our clients trust to preserve their hard-won reputation.

Really enjoying Claude Projects v2 right now (some minor hiccups but impressed with the self-learning) Completely gave up on Dots today in favor of GPT 6 Astra on Ultra. I don’t get what OpenAI thought was so special about it compared to its competitors.

If you own an AI agent solution running against the Anthropic API, the MAX plan now pays for itself. Super cool!

v sick of OpenAI, the “metered intelligence” company, being the exact opposite of it.

These demos are cool (really cool, not trying to hate) but in my experience with logistics the hard part isn’t in a missing visual layer, although it makes for a super fun demo. The problems are in things these UIs can’t help with: - Inventory isn’t where the WMS says it should be - Customer has a dispute on our damaged inventory claim and we need to provide proof. - Surprises happen and no one on the floor has time or cares to put it in the system - We don’t integrate with a customer’s portal so we have to manage their orders manually. Again, shout out to the guys playing around with game-like UIs, I do think they are the future. But very few operators have the data to really make something like this happen, and even if they do, it’s not going to capture the non-normal flows that make up the hardest parts of the job.

The data on how many people switched to Opus 5.5 must be beyond the pale

I can’t believe what I just found… This book, which ranks #1 for several Amazon categories, is complete AI slop. I think the reviews are all fake too. Who knows a journalist who’d be interested in this?

The Tasks: Task 1 A nonprofit runs four programs, and each program’s December 2024 spending was compared to its budget, program by program, in dollars and as a percentage. Some costs, like payroll taxes, sat in a shared pool rather than under any one program. Those had to be spread across all four programs and the two support functions first. Any line where a program ran $14,000 or more over budget was flagged with a note explaining what drove the variance. Task 2 A hotel’s booking system and its accounting books disagreed about December room revenue, which also threw off the commissions owed to its online travel agencies for the month. Both gaps had to be tracked down item by item, with the amount and the account or platform for each, then fixed with correcting journal entries. Finally, the net effect of both corrections on December income was calculated. Task 3 A venue’s events in December brought in more revenue than budgeted, but the average revenue per event came in under plan. Actual and budgeted figures, along with the gap between them, were needed for: Number of events Average revenue per event Food and beverage minimums Mix of event types by count Finally, an explanation of the overall difference was provided, backed by the revenue and event count for each type of event. Task 4 Three companies in one group each rent a property. The holding company has an office, the trucking division has a terminal yard, and the logistics division has a warehouse. Under U.S. lease accounting rules, each lease sits on the books in two ways: there is a lease liability for what is still owed and a right-of-use asset for the value of using the space. Both balances decrease over time. For December, each lease schedule was rolled forward by one month, calculating how much of the asset was used up and the reduction in the lease liability. Those amounts were then included in the December journal entries for each company’s QuickBooks file.

Definitely a little frustrated with the launch today. Dots sounds cool, but not (yet?) available on my apparently peasant-tier $100/month plan. Spaces is something I'd like to explore, but doesn't seem to launch at all. GPT 6.1 Sol is here, which is super weird when GPT 6 Sol came out less than a fortnight ago. Sign in with ChatGPT is a very cool idea with potentially huge implications, but it's limited to design partners (and also overloaded anyway). Ultrafast is irrelevant unless your company gives you unlimited tokens. So... if nothing had launched today, nothing would be different for me than it is right now.

You’re meant to say: “We’ve been talking about how SI (Super Intelligence) would lead to SI (Super Intelligence) for a decade.” See the issue now?

jesus fucking christ

oh geez, rick

I hate that I have to say this... Superintelligence is already a coined term, and a separate one from Artificial Intelligence. Compelling the government (and of course attempting to influence the private sector as well) to use one term for another makes it difficult to talk about SI and confuses the important discussion about where AI has not reached SI. This is not a "Gulf of America" scenario where you're just grandstanding. This is an actual linguistic problem that Trump is insisting on for no reason except that he wants to name things and he thinks "Artificial" means fake. reddit.com/r/ArtificialIntel…