If Salesforce is treating the UI as a loss leader, your entire stack thesis and AI GTM playbook are up for renegotiation.

This week's edition covers:

  • Kevin's Take: Salesforce just made the UI a loss leader, and what that shift means for your stack
  • The Signal: Signal: how AI engines actually pick YouTube videos to cite, plus one marketer's real numbers on AI's GTM impact
  • Tools & Tactics: inside the build of an AI agent running paid media, and the five-model check you should run on your own brand
  • Quick Links: what LinkedIn leads actually cost in 2026, 38% of B2B buyers already buying via agentic AI, content marketing's 12-year low

Kevin's Take

Salesforce is telling us the UI is the loss leader now

Ben Thompson had a read on Salesforce's Agentforce play that I keep coming back to (here), and the part I can't get past is that Salesforce is basically admitting the dashboard, the thing they spent twenty-five years teaching us to log into every morning, is not where the value sits anymore. They're pointing at the data and the agent orchestration layer and saying that's the moat, and they're willing to let the UI fade into the background to defend it. If your martech stack is duct-taped to Salesforce, and most of ours are, that's a pretty big signal about where you should be spending your attention.

I think the easy read on this is that it's a Salesforce story, but it's actually a stack-wide story. Every SaaS vendor we buy from is running some version of the same calculation right now, and the ones being honest with themselves are landing in the same place, which is that the interface is becoming a commodity and the data model underneath it is the thing worth owning.

We're seeing this in web too. The whole move to headless CMS is the same shape of bet, decouple the presentation from the substance because the presentation is going to get rebuilt by an agent for every visitor anyway. Same physics playing out in two different corners of the stack at the same time.

What I think we should do with this as marketing leaders is stop evaluating our tools on the quality of the dashboard and start evaluating them on the quality of the data they hold and how cleanly an agent can get at it. If your CRM's data model is a mess, no amount of pretty Salesforce UI was ever really saving you, and now that the UI is on its way out, there's nothing left to hide behind.

The next vendor conversation I'm having is not about screens, it's about schemas and APIs and whether an agent I build can actually read and write against their system without me having to beg for a partnership tier. That's the real question sitting inside Thompson's piece, and I don't think most of us are ready to answer it yet.

That's it for this week. Talk soon.

— Kevin Kerner, CEO, Mighty & True

The Signal

AI engines cite YouTube videos for structure, not popularity (1 min read)

OtterlyAI analyzed 100+ million AI citations and found YouTube videos get cited by AI engines based on structural clarity like long-form format and timestamps, not popularity metrics such as views or subscribers. 94% of citations came from long-form videos rather than Shorts.

Why it matters: If you're investing in YouTube for AI search visibility, this tells you to optimize video structure and timestamps rather than chasing view counts.

One marketer's monthly data tracker on AI's GTM impact (1 min read)

Adam Schoenfeld shares his monthly first-person data roundup tracking how AI is reshaping go-to-market metrics, from buyer behavior to pipeline signals. The report compiles his own measurements rather than syndicated survey data.

Why it matters: You get a practitioner's real numbers on AI's GTM impact instead of another vendor survey, so you can benchmark your own pipeline shifts against someone actually tracking the data.

Tools & Tactics

Inside the build of an AI agent that runs paid media (1 min read)

LangChain's team details the architecture and workflow behind an internal AI agent that manages their paid media campaigns end-to-end. The post walks through how the agent handles targeting, bid optimization, and creative testing using LangChain's own agent framework.

Why it matters: If you're evaluating whether AI agents can actually run paid media instead of just assisting with it, this gives you a real build blueprint to steal or benchmark against.

Why you should ask five AI models what they know about your brand (1 min read)

The essay proposes a 'mirror test'—asking five AI models what they know about your brand—citing G2 data showing 51% of B2B buyers now start research in a chatbot (up from 29% a year ago) and 71% use AI search for vendor research. It outlines a fix-the-source playbook: correct wrong facts at the origin, then work outward to improve how AI models represent your brand.

Why it matters: If half your buyers are forming first impressions of you inside a chatbot instead of a search engine, you need to know exactly what those models are saying about your brand right now.

Listen: Tech Marketing Rewired

Tech Marketing Rewired is the podcast for CMOs rebuilding their orgs for AI. Each episode is an unscripted conversation with a working operator on what they're actually shipping, what broke, and what they would do differently.

▶ Listen to Tech Marketing Rewired

Help us with a research report

We are putting together a research report on how mid-market tech companies are actually allocating budget in 2026. Not the aspirational version, the real one. If you are willing to share your numbers anonymously, reply and I will send the survey. You will get the full report before anyone else.

▶ Share your numbers →