The Auxano Weekly.
What changed this week in GTM, RevOps, and marketing, and what to do about it. Every claim carries a source link, and every number a vendor publishes about itself is labeled as a claim.
-
AI answers still lean on earned media, and new agents keep a person in the approval seat
Muck Rack says earned and third-party sources supply over 84% of AI citations. Sprinklr and Oracle shipped agents with review steps, and Optimove keeps customer data in the warehouse.
Read the issue -
This week's agent launches came with a spend meter and an approval default
Peec, Kochava, ThoughtSpot, and Demandbase shipped agents with budgets, approval steps, or audit trails attached. We also cover Salesforce buying Braze and one AI search case study.
Read the issue -
Agents now arrive with the approval step, the log, and the spend cap attached
This week's agent launches came with approval steps, audit logs, and spend caps attached. We cover Salesforce Hunter, Seismic, Lightspeed, and Webflow, plus two data points to read with care.
Read the issue -
The agent prepares the work, and a person still presses launch
Vendors answered last week's question about agent limits by building the approval step into the product. Clay went the other way. Pricing is shifting too.
Read the issue -
Who sets the limits on an agent became the question this week
OpenAI held back a model over scope. Braze and Amazon shipped agents with different answers. Three vendors published lift numbers with no baseline.
Read the issue -
Last week the platforms carried themselves into Claude. This week they handed over the keys.
Amazon, Adobe, Zig.ai, and Typeface all made their products operable from inside Claude or ChatGPT in the same week. None of them published what those agents are allowed to do.
Read the issue -
Two conferences said the same thing, and neither one said what it costs
Dreamforce and HubSpot UNBOUND gave the same answer: AI replaces the interface, and written context feeds the agent. Neither said what it costs to run.
Read the issue -
The screen stopped being where the work happens
Klaviyo, Box, Databox, and Amazon moved work outside their own screens in one week, while agent ownership and AI ad measurement stayed unsolved.
Read the issue -
Everyone uses it, nobody owns the record it reads
Every revenue leader in one survey uses AI, yet only one in five calls it production-ready. Meanwhile AI answers about your pricing mostly come from sources you do not control.
Read the issue -
Write access arrived, and the data underneath it is not ready
Salesforce put its CRM inside Claude and X gave agents write access to live ad accounts, while most marketers say their CRM data is not ready for either.
Read the issue -
The shortlist is getting built before you know the deal exists
Four research firms landed on the same finding in one week: buyers let AI build the shortlist, and most of them buy from it, before you know the deal exists.
Read the issue -
The assistant became the interface, and it is starting to pick winners
Product marketers now reach for a chatbot before their competitive intelligence tool, and 6sense made its intent data callable from inside Claude and ChatGPT.
Read the issue
Want help acting on this in your own stack?
Auxano works inside your business, not on your account. One conversation, and we will tell you honestly what to do first, or whether you need us at all.