If half your buyers now start in ChatGPT, your SEO playbook is the minority strategy and your stack has some catching up to do.

This week's edition covers:

  • Kevin's Take: your next email reader isn't a human — here's how to write for the agent that screens it first
  • The Signal: Signal: more than half of buyers now start in ChatGPT, and your branded traffic split is hiding the real story about your SEO program
  • Tools & Tactics: ship custom prospecting agents in hours with Claygent, plus a repeatable brief for marketing to mid-market RevOps
  • Quick Links: 197 Claude skills replacing RevOps, why fresh content out-cites authority in AI search, and a podcast that grew 397% in 28 days

Kevin's Take

If an AI agent is going to be the first reader of your email, you need to be writing for the agent

I read this Tomasz Tunguz piece on the disappearance of email and it really got me thinking, because his core claim is that within five years AI agents are going to be triaging inboxes down to just the six or so messages that actually deserve a human's attention. The inbox becomes a filtered queue, not a conveyor belt. If you're a B2B marketer running cold email or nurture sequences, that's a pretty wild thing to sit with for a minute, because it means the first reader of your email isn't a human anymore, it's a model deciding whether the human ever sees it.

What I find interesting is that we're already running this experiment on ourselves at Mighty and True. I have a Claude project that scans my inbox and pulls out what actually matters for me to look at on any given day, and I've watched it get really good at this really fast. It tosses anything that smells like a sequence, anything that's pattern-matching to other vendor pitches it's seen, anything where the sender clearly doesn't know what we actually do. What survives is stuff that's specific, stuff that references something real, stuff that reads like a person who did the work to understand the context. That's the bar. If my janky personal triage system can already do that, the production version Tomasz is describing is going to be ruthless.

What I think this means for marketers is that the whole game of volume sequences and A/B subject lines and just keep hitting send is going to get crushed at the gate, because none of that survives an LLM triage layer that has seen ten thousand variations of the same playbook. The pivot is to write emails that an agent would actually score as worth surfacing, which means real specificity about the recipient, real signal that you've done the work, real reason this person needs to see this today. If you're running nurture programs right now, you should be pressure-testing your own sequences by running them through a triage model and seeing what gets killed. I'd bet most nurture programs would not survive their own test. It's a wonderful world these days, and also a slightly terrifying one if you're still running 2019 playbooks.

That's it for this week. Talk soon.

— Kevin Kerner, CEO, Mighty & True

The Signal

The Answer Economy: G2's 2026 AI Search Insight Report (1 min read)

G2's 2026 AI Search Insight Report finds 51% of B2B buyers now begin software research with AI tools rather than Google. The shift reframes vendor visibility, trust signals, and how buyers shortlist before ever hitting your site.

Why it matters: If more than half of your buyers are starting in ChatGPT instead of Google, your SEO playbook is already a minority strategy and your AEO work needs to be the priority this quarter.

What Your Branded Traffic Split Is Really Telling You (1 min read)

Analysis of 90 days of GSC data across 8 client sites benchmarks branded vs. non-branded search traffic splits by industry. The piece argues the ratio is a diagnostic for brand health and demand capture balance.

Why it matters: If you can't explain why your branded/non-branded split looks the way it does, you don't actually know whether your SEO program is building demand or just harvesting it.

Tools & Tactics

Claygent Builder: Build, Test, and Deploy GTM Agents | Clay - Clay (1 min read)

Clay launched Claygent Builder (Sculptor), letting GTM teams build, test, version, and deploy custom AI research agents in natural language across workflows. Agents can be tested free on real data before deployment.

Why it matters: If your GTM stack already runs through Clay, you can now spin up reusable research agents without burning credits on trial-and-error — meaning your ops team ships custom prospecting logic in hours instead of weeks.

Audience Research Brief: How We’d Market to Mid-Market RevOps Leaders (1 min read)

SparkToro's Amanda Natividad walks through an applied audience research brief targeting mid-market RevOps leaders, showing how to translate ICP signals into concrete positioning, channel, and content choices. The piece doubles as a template for running this exercise on your own buyer.

Why it matters: If your team starts with personas instead of real audience signals, you're guessing at messaging — this gives you a repeatable brief to fix that.

Resources & Tools

Strategy

Score your B2B marketing — free

The Blueprint Score benchmarks your brand's competitive velocity (content activity, market presence, search momentum) against 50 mid-market B2B tech companies. Enter your domain, get your score in 5 minutes.

▶ Run your Blueprint Score

Pipeline feel stuck?

Every week I talk to CMOs who are spending six figures on channels that stopped working two quarters ago. If your pipeline feels stuck and you are not sure which lever to pull next, let us talk. 30 minutes, no strings. Reply to this email.

▶ Book a strategy call →