Insights & how-to

MGS 2026 Recap: 7 AI Marketing Insights from the Sessions

By the daglo team · November 3, 2026

New AI tools ship every month, so why does your team's actual way of working look almost the same as it did a year ago? When each new tool only makes the first draft of a report a little faster, "adopting AI" starts to feel like extra work rather than real progress.

MGS 2026 (Modern Growth Stack Week 2026) is a growth marketing conference hosted by AB180, held July 21 to 24, 2026 at the Westin Seoul Parnas under the theme "Stack AI, Rewrite Everything." The daglo marketing team was there and sat in on many of the sessions. Practitioners from OpenAI, Amplitude, Braze, MUSINSA, KRAFTON, and others shared how they are rebuilding marketing and growth with AI. This post collects the messages that came up again and again, plus links to our recap of every session.

What is MGS 2026? The event at a glance

A speaker presenting on stage to an audience

MGS 2026 is a conference where marketing, growth, and data practitioners share growth strategies for the AI era. This year's theme was Stack AI, Rewrite Everything, and sessions ran in parallel on two stages, the MAIN HALL and the PLAY HALL.

Item	Details
Event	Modern Growth Stack (MGS) Week 2026
Date and venue	July 21-24, 2026, Westin Seoul Parnas
Theme	Stack AI, Rewrite Everything
Tracks	MAIN HALL, PLAY HALL
Topics	AI marketing, user acquisition (UA), CRM, analytics, ad tech, global growth, gaming

Event details are on the official MGS website.

Seven messages that ran through the sessions

A team gathered around a whiteboard sharing ideas

Speakers from very different companies kept arriving at the same conclusions on their own. The biggest one: don't bolt AI onto existing work. Redesign the work itself.

Redesign, don't just improve: AB180 compared it to factories that adopted electricity but kept their old line shafts, and saw only small productivity gains. OpenAI described a model where AI does the work first and people step into the loop when judgment is needed. KRAFTON aimed for workflows where people could step out entirely.
Measure AI by impact, not usage: Adoption rates and token spend are input metrics. MyRealTrip evaluates AI use not by "how much do you use it" but by "what revenue and profit did it create."
Write down done criteria and guardrails: OpenAI advised giving agents a goal, context, constraints, and a clear "done when." Kolon FnC wrote "AI never sends messages" into every instruction document.
A single AI reinforces its own hypothesis: W Concept runs the same data through two agents, one quantitative and one strategic, and re-checks wherever they disagree. KRAFTON ends prompts with a devil's advocate step to block hasty conclusions.
Start from the assumption that most bets fail: Actionfit says fewer than 0.5% of new games hit their early KPIs, so it builds 100+ a year and drops losers fast. Appier puts the hit rate for finding a winning ad creative at 5-8%.
Without documented context, AI only gives generic answers: Kolon FnC trained its AI on a 25-page CRM guide. When MUSINSA let AI reference its verified dashboards, the AI started recommending those dashboards on its own.
When AI mediates discovery, direct customer relationships are what's left: Moloco noted that 80% of Google searches with AI summaries end without a click. NNT added that to be found at all, you need to be a source AI cites, and that GEO is an extension of SEO.

All MGS 2026 session recaps by topic

People standing around a table reviewing materials

Here are the session recaps the daglo team wrote, grouped by topic. Start with the area you care about most; each recap covers the key points and what you can apply right away.

AI agents and new ways of working

Data and analytics

CRM and customer relationships

Ads and user acquisition (UA)

AI search optimization (GEO)

AI Brand Citation Patterns from 200+ SEO/AEO/GEO Projects (NNT)

Global, subscription, and game growth

What marketers can try this week

A presenter standing in front of a large screen in a dark venue

We picked only the advice you can start this week without a big budget. Even one of these moves your AI use a step from "assistant" toward "delegation."

Ask AI where your bottleneck is: Try "Based on our conversations so far, diagnose the bottlenecks in my work," then decide what to automate first (from the OpenAI session).
Add done criteria to your prompts: For every recurring task, write one line each for the goal, constraints, the finish line, and what the AI must never do.
Add a counterargument step: When you hand off analysis, end with "Find evidence against this conclusion."
Move the rules in your head into a document: Metric definitions, brand voice, send criteria. Write down what people know implicitly so AI can use it.
Change how you measure success: Track hours saved, more experiments run, and revenue impact instead of how often AI was used.

How the daglo team captured MGS 2026

With 20+ sessions a day running on two stages, notes alone can't keep up with every number and example. The daglo team transcribed the sessions with daglo and wrote this series by rereading the transcripts after the event.

Many sessions had two or three speakers taking turns. daglo's Speaker ID remembers voices you've named once and automatically fills in real names instead of "Speaker 1" in your next recordings. It's the first feature of its kind in Korea offered to every user, so it's worth a try if you attend a lot of conferences and seminars. Learn more about Speaker ID

We'll keep sharing what we learn from the marketing and AI scene on the daglo blog. Pick a session recap above and start with the topic you care about most.

#MGS2026 #ModernGrowthStack #GrowthMarketing #AIMarketing #MarketingConference #ConferenceRecap #AB180 #daglo

👉 Transcribe your next session

← Back to blog