Quarterly issue
Q2 2026
Issue 2: the vertical-agency model, the AI marketing stack, and twelve teams' tooling choices.
Editor's letter
Issue 2 lands at the point in the year where the agentic stack market has visibly consolidated and the live arguments have moved up a layer. A year ago we would have spent an issue arguing about which orchestration framework to pick. This quarter we are barely talking about frameworks. We are talking about the architectures teams are building on top of them, the vertical-agency model that has emerged as a credible distribution channel for the new stack, and the auditability story that has gone from “nice to have” to “show me before we sign a contract.”
The lead is Ginger’s piece on the AI marketing stack that is actually shipping at the kinds of teams whose work you would recognize. Reza’s longer essay covers what the vertical-agency model actually looks like inside a small operator-led shop.
Two pieces in this issue cover tools rather than patterns. Reza’s “eight open-source tools” piece is the kind of practitioner-side recommendation we get asked for at conferences; we wrote down the canonical version. The Editorial Team’s twelve-teams tooling survey is the recurring landscape piece we promised in Issue 1’s editor’s letter.
Q3 lands the first week of July. Between now and then, a small number of rolling pieces on topics we did not have room for in this issue.
Thanks for reading.
— The Editor
Contents
- 01AI Marketing Stacks That Don't Suck
What the working marketing-engineering teams are actually shipping with — and which 'AI marketing stack' pitches are not credible.
- 02The Quiet Power of the Vertical Agentic Agency
Why a small operator-led shop with a tight beat outperforms the generic 'AI consultancy' on the kind of work clients keep coming back for.
- 03What 'AI Agency' Actually Means in 2026
The agency model has bifurcated. We name the two species, describe what each one ships, and explain which one is selling the future.
- 04Building a Marketing Agent: A Walkthrough
A working code walkthrough of a marketing agent that has to call three tools, hold memory across a session, and produce structured output.
- 05Eight Open-Source Tools Every Agentic Engineer Should Know
The eight libraries we reach for most often, with the use cases and the failure modes that the marketing pages do not mention.
- 06Inside the Tooling Choices of Twelve Frontier AI Teams
A survey piece on what tooling the better-known agentic teams are actually running with — drawn from public writeups, conversations, and our own audits.
The Q2 2026 issue of Stack Quarterly drops the first week of April. Six pieces this issue.