Three years of launches. Is any of it actually working better?
New models every few months. Benchmark records broken so often they stopped being news. Every launch billed as the leap. So we sorted three years of first-hand reads by the year somebody wrote them. Praise and complaints barely moved. In the aisles a small business actually shops in, the two have sat near even the whole time. None of which means the tools are bad. The pool is net positive by a comfortable margin and features draw far more praise than complaints. It means the thing getting better is the model, not your Tuesday. We recheck on every pool refresh.
· the Unbought desk · zero affiliate dollars in this opinion · checked against first-hand reads, not vendor demosthe claim, quoted
“It's early days. Every new model makes these tools better, so the stuff that annoys you now gets fixed on the next release.”
Spun out of the Shrug Economy take, Aug 31, 2026 · three-year sentiment-by-year cut run Aug 31, 2026 against the unbought.ai read pool, refreshed Aug 29
real today
The tools do the job they advertise. Features draw far more praise than complaints and it's the thing owners talk about most.
not yet
The improvement curve. Three years of reads and the split barely moves. Same story in 2024 as now.
the evidence · what real users say
“Salesforce is robbery at this point. How is there no API access at $10k/yr?!”
“we called Salesforce, the phone was down”
“ive been able to learn salesforce with no prior experience and build a foundation of knowledge”
data as of oct 13, 2026Salesforce, the full ranking →“Someone renamed a property in the HubSpot settings UI and our writes went into a field that no longer meant what it used to. No PR, no git history”
data as of oct 13, 2026Hubspot, the full ranking →the move
Public Enemy said it in 1988. Don't believe the hype. Stop waiting for the release that fixes it. The next one will be billed as the leap. So was the last one. Then ask yourself one question about the tool that's been bugging you: is the middle fine for now? Does it do the job, badly enough to annoy you but well enough that the work still gets done? If yes, that's your answer. Embrace the Shrug Economy. Stop shopping, stop reading the launch coverage, put a reminder six months out and get back to work. Switching costs you real time, and the upgrade is not arriving on the schedule they're implying. If no, it isn't going to fix itself either. Pull your last statement and write down every subscription on it, then next to each one write the last time it did something you'd have happily paid for on its own. Can't remember, cancel one this week and see if anything breaks. And when something new does land, test it free on one real job from your own shop. Judge it on your output, not their chart.