The 5-question filter I run every agent launch through (so you can stop reading release notes)

“I am exhausted. Every week another agent launches. Everyone is telling me I should care. How do I know which ones actually matter?”

That is the most common thing I’m hearing from team leads and executives right now. Some version of that question, every week, from people who are good at their jobs and tired of pretending they have read every release note.

The agent conversation stopped being about models two quarters ago. It is about infrastructure now. What wins enterprise adoption this quarter is not which lab has the best frontier model. It is which products let you reach the data, run the workflows, and stack the agents your team already uses.

Teams still benchmarking model quality are losing the year to teams that picked the right data fabric. License spend is getting wasted on products that win demos and don’t survive real work. Buyers are asking “which agent is best” when the question is “which agent fits the work my team actually does.” The teams that figure this out before their competitors are pulling ahead in compounding ways, and the gap is widening, not closing.

There are four launches from the last six weeks that pass the filter I use, and most of the coverage you have read missed three of them. ChatGPT Workspace Agents, which I covered earlier this week, is one. The other three are getting less attention and matter at least as much. In some cases more.

Here’s what’s inside:

• The five-question filter. What I run every agent launch through before deciding whether to pay attention — reusable on the next four launches, and the four after those.

• Why Salesforce Headless 360 is the most important launch of the month. The cleanest infrastructure play any major enterprise vendor has made this year, and why most coverage missed it.

• A practical routing guide for Copilot, Perplexity, Claude direct, and Salesforce. Stop forcing one product to do every job — match the work to the tool.

• The “should I switch” question is framed wrong. Why this is a layering decision, not a switching one, and the three sub-questions that replace it.

• Three prompts you can use today. A reusable launch filter for every announcement that lands in your feed, a license spend audit for your current AI tools, and a layering map you can share with your team.

The filter comes first because it is reusable.