Charles Gabriel already named the trap: "I have 10,000 agents" is a marketing number
Today's data says AI output has exploded while the hours haven't moved. On Brew Bytes Ep 13: 95% of enterprise AI projects fail. Charles Gabriel described the same confusion arriving in his own industry, and called it the next phase of the problem before the numbers landed.
What it actually means
In his words: "now what I'm starting to see is companies in our industry are creating agents. And the marketing, if you're a holdco or a big agency, might be I have 10,000 agents. Therefore we can do, you know, the work better. We have better outcomes. We have better output for our customers… That to me is going to be the next phase of the issues." He'd seen the film before. The ad network boasting 10,000 publishers when the average consumer visits ten. His counter is unglamorous and it holds up against Linear's chart: pipes, then people. Know the problem you're solving, get the data actually ready, and go after short-term ROI. You can point at two to five hours a week per person is real, where "transform the whole business" is a moonshot that shows up as 95% failure. The distinction he draws is the practical one for anyone deploying agents this quarter: volume of agents versus interoperable agent workflows. Fifty agents that can't see each other's work is fifty new silos, and you'll feel it as exactly what Linear measured → more activity, same week.
Ten thousand agents is a number you can market. It isn't a number that ships anything.
Source: Brew Bytes · Friday 21 August 2026