MAICON 2026

Thirteen themes from three days of MAICON

By Asma Stewart, Clarity & Motion Collective · Published

I just attended my third MAICON and as always, came away with a lot of food for thought. I decided to work with Claude to dig deep into this year's themes and compare them to what was discussed last year. You can also explore each theme and how to apply it based on your team's AI level.

Want the full sessions? MAICON 2026 replays are available as a paid on-demand bundle on the Marketing AI Institute site.

Register for the replays

How to use this

  1. 1
    Click a themeBigger words came up in more of the 32 sessions.
  2. 2
    Pick your team's AI levelExploring: using AI on your own. Operationalizing: shared workflows and rules. Building: agents and automations in real work.
  3. 3
    Get your next movesPractical steps for that theme at your level, plus the sessions behind it.

Click any theme in the cloud.

How Clarity & Motion works

Steps to consider when you want to agentify a workflow

Each theme above leans on one of these steps. The highlighted step is the one the current theme depends on most.

  1. 1Formal discoveryTalk to the people doing the work and set a baseline.
  2. 2Document the workflowWrite down how it actually runs today, workarounds included.
  3. 3Reinvent or improveDecide whether the work is ripe for reinvention or just needs fixing.
  4. 4Formal PRDRequirements, human checkpoints, guardrails and the success metric.
  5. 5Architect the contextStructure the information so AI can actually use it.
  6. 6BuildOnly now does anything get built.
Read all 13 themes and the 2025 vs 2026 comparison on one page

Agentic workflows: what MAICON 2026 said and what to do about it

Discussed in 14 of the 32 MAICON 2026 sessions.

What came up

The conversation has moved past prompting and AI basics. Speakers showed chains of AI steps where each output feeds the next, with people supplying the raw experience at the start and the judgment at the end. Almost everyone was candid that agents worked best when they were narrow, well briefed and checked.

Why it matters for marketing teams

Most teams are stuck between using AI chat and running agentic systems. Crossing that gap has less to do with the tool than with whether you can describe the work precisely enough for a system to successfully act on it.

What to do about agentic workflows at your team's AI level

Exploring: People use AI on their own, mostly through chat and prompting.
  1. Pick one multi-step task you do and break it into its steps. Mark each step as human (your judgment, experience or final call) or AI.
  2. Codify each AI step as its own skill. Give each skill one job, spell out its inputs and what a good output looks like, include an example of past work you liked, and test it on a real task before you rely on it.
  3. Run the whole task through your skills and note which step you correct most. That's where a human checkpoint belongs as you scale it.
Operationalizing: The team shares prompts, skills and workflows, and has some rules.
  1. As a team, pick one recurring workflow, break it into steps and agree which are human and which are AI.
  2. Codify each AI step as a skill, then package the skills, review steps and connectors into one shared plugin: a working system everyone installs instead of running their own version.
  3. Test the plugin together on real work, with a named reviewer at each human checkpoint and a critique step, since AI tends to agree with whatever it's handed.
  4. Keep a shared log of where it broke and why, and update the plugin before you roll it out more widely.
Building: You're running agents, automations or custom tools in real work.
  1. Give each agent one job, clear inputs and the narrowest permissions it needs.
  2. Put approval gates in front of anything that reaches a customer or goes public.
  3. Plan to rebuild. More than one speaker built the same agent several times before it held up.

MAICON 2026 sessions on this theme

Context is the moat: what MAICON 2026 said and what to do about it

Discussed in 9 of the 32 MAICON 2026 sessions.

What came up

Which model you use mattered less than what you feed it. Written procedures, clean data, your industry's vocabulary and clear success criteria separated useful output from generic output. Several teams turned their SOPs into reusable skills so the context travels with the work.

Why it matters for marketing teams

Every competitor has the same models. Your documented knowledge (how you decide, what good looks like, what customers actually say) is the part they can't copy.

What to do about context is the moat at your team's AI level

Exploring: People use AI on their own, mostly through chat and prompting.
  1. Write a one-page brief on your role, audience and standards, and start every serious AI session with it.
  2. Save your best prompt and its context as a reusable template instead of retyping it.
  3. When output is generic, add context before you switch tools.
Operationalizing: The team shares prompts, skills and workflows, and has some rules.
  1. Turn your team's top three SOPs into shared AI skills or projects everyone uses.
  2. Build a shared glossary of industry and brand terms for AI to reference.
  3. Give the shared context an owner, the same way someone owns the brand guide.
Building: You're running agents, automations or custom tools in real work.
  1. Connect AI to the systems where your knowledge already lives, through integrations or MCP, instead of copying files around.
  2. Structure your knowledge base so both people and agents can query it.
  3. Version your skills and context files, and test changes before they reach the team.

MAICON 2026 sessions on this theme

Guardrails, not gates: what MAICON 2026 said and what to do about it

Discussed in 9 of the 32 MAICON 2026 sessions.

What came up

The organizations moving fastest replaced permission-based approvals with clear rules: which tools are approved, what data can go where, and which outputs need a human sign-off. Experimentation stayed open, customer-facing work kept its review gates, and a named person stayed accountable for every AI output.

Why it matters for marketing teams

Without guardrails, teams either freeze or improvise in ways that create risk. Clear rules are what let people experiment without asking permission every time.

What to do about guardrails, not gates at your team's AI level

Exploring: People use AI on their own, mostly through chat and prompting.
  1. Find out which AI tools and data your company has approved, and stay inside that list.
  2. Check every AI output before you share it as if your name is on it, because it is.
  3. Keep sensitive customer data out of tools that aren't approved for it.
Operationalizing: The team shares prompts, skills and workflows, and has some rules.
  1. Write a one-page usage policy in plain language: approved tools, data rules and what needs review.
  2. Keep your existing review gates for external content and route AI output through them.
  3. Set up a small AI council that updates the rules as the tools change.
Building: You're running agents, automations or custom tools in real work.
  1. Give agents the minimum permissions and data access they need, and log what they do.
  2. Bring security in early, as a collaborator, before a tool goes beyond you.
  3. Decide in writing how much autonomy each use case gets, based on what a mistake would cost.

MAICON 2026 sessions on this theme

Every marketer a builder: what MAICON 2026 said and what to do about it

Discussed in 9 of the 32 MAICON 2026 sessions.

What came up

Non-engineers built working tools on stage, from apps to self-running project plans, with vibe coding and no-code tools. Speakers were just as clear that a tool other people depend on needs engineering discipline, and that one AI leader can't also be the team's only builder.

Why it matters for marketing teams

In most marketing teams the bottleneck isn't ideas for AI. It's people with the time and skills to build them, and spreading building skills across the team adds more capacity than any single hire.

What to do about every marketer a builder at your team's AI level

Exploring: People use AI on their own, mostly through chat and prompting.
  1. Build one small tool for yourself (a calculator, a checklist, a formatter) with a no-code or vibe coding tool.
  2. Before you start, write down what it should do and what done looks like.
  3. Keep it personal until it works reliably.
Operationalizing: The team shares prompts, skills and workflows, and has some rules.
  1. Run a team build day where everyone ships one small tool for their own work, paired with stronger builders.
  2. Make a shared space to show what people built and what broke.
  3. Agree on when a personal tool becomes a team tool, and what it needs first.
Building: You're running agents, automations or custom tools in real work.
  1. Write a spec before you prompt: users, inputs, edge cases and how you'll test it.
  2. Use version control and basic tests for anything other people rely on.
  3. Make the case for dedicated builders embedded in marketing rather than queued behind a central team.

MAICON 2026 sessions on this theme

Measure what matters: what MAICON 2026 said and what to do about it

Discussed in 8 of the 32 MAICON 2026 sessions.

What came up

Usage counts and hours saved weren't enough for anyone. Speakers pushed toward outcomes: pipeline, the quality of traffic rather than its volume, how often your brand shows up in AI answers, and results compared against the market instead of only last year. Reporting cadence should follow the question being asked.

Why it matters for marketing teams

If you can't show what AI changed for the business, the program gets treated as a cost. Baselines set before you start are what make the results believable later.

What to do about measure what matters at your team's AI level

Exploring: People use AI on their own, mostly through chat and prompting.
  1. Before you use AI on a task, note how long it takes today and how good the result usually is.
  2. Track one before-and-after comparison for a month.
  3. Ask a few AI assistants about your category and see whether your brand comes up.
Operationalizing: The team shares prompts, skills and workflows, and has some rules.
  1. Baseline the hours your core workflows take before you change them.
  2. Add traffic quality and AI referral sources to reporting, broken out by platform.
  3. Review quarterly whether your metrics still reflect how buyers behave.
Building: You're running agents, automations or custom tools in real work.
  1. Build alerts that flag performance changes as they happen instead of waiting for the monthly report.
  2. Tie AI workflows to pipeline or revenue wherever you can.
  3. Instrument every build with the one metric its requirements say defines success.

MAICON 2026 sessions on this theme

Culture leads adoption: what MAICON 2026 said and what to do about it

Discussed in 7 of the 32 MAICON 2026 sessions.

What came up

Tool proficiency was treated as table stakes. What separated successful rollouts was the culture around them: leaders learning first and experimenting where their teams could see, enablement with protected time, and an ongoing program instead of a time-boxed pilot. Rollouts driven by fear stalled.

Why it matters for marketing teams

How a team feels about AI depends on what its leaders model. If the message lands as doing more with fewer people, adoption goes underground or stops.

What to do about culture leads adoption at your team's AI level

Exploring: People use AI on their own, mostly through chat and prompting.
  1. Share one thing you tried with AI this week with your team, including what didn't work.
  2. Swap one technique with a colleague who uses AI differently than you.
  3. Block a recurring hour each week just for learning.
Operationalizing: The team shares prompts, skills and workflows, and has some rules.
  1. Benchmark where the team actually is before you design training.
  2. Run role-specific enablement with protected learning time and visible progress.
  3. Name AI guides on the team who help others without becoming the only people who can do it.
Building: You're running agents, automations or custom tools in real work.
  1. Pair your strongest builders with people at earlier levels so the skills spread.
  2. Turn successful builds into show-and-tells so people see what's possible in their own work.
  3. Track adoption by outcomes in each role, not by logins.

MAICON 2026 sessions on this theme

Seen, believed, chosen: what MAICON 2026 said and what to do about it

Discussed in 7 of the 32 MAICON 2026 sessions.

What came up

Buyers are asking AI assistants instead of searching, and clicks are falling. Visibility now depends on whether AI systems recognize your brand and your experts as credible, and that comes from consistent mentions, real commentary and third-party proof as much as from your own pages.

Why it matters for marketing teams

If AI answers don't mention you, a growing share of buyers never reach your website. Credibility is being decided before anyone clicks.

What to do about seen, believed, chosen at your team's AI level

Exploring: People use AI on their own, mostly through chat and prompting.
  1. Ask three AI assistants the questions your buyers ask, and note whether and how your brand appears.
  2. Check that your company and experts are described the same way across your site and profiles.
  3. Look at which sources the AI cites.
Operationalizing: The team shares prompts, skills and workflows, and has some rules.
  1. Make an AI visibility check a recurring report alongside SEO.
  2. Get your subject-matter experts quoted in credible third-party places, not only on your blog.
  3. Align PR, content and SEO around the same topics and the same experts.
Building: You're running agents, automations or custom tools in real work.
  1. Automate tracking of how your brand shows up in AI answers across platforms.
  2. Build workflows that turn expert commentary into timely PR pitches.
  3. Report AI referral traffic separately from search, by quality as well as volume.

MAICON 2026 sessions on this theme

Redesign before you automate: what MAICON 2026 said and what to do about it

Discussed in 6 of the 32 MAICON 2026 sessions.

What came up

Session after session came back to the same point: wiring AI into a process nobody has examined just makes the broken parts run faster. The teams getting real results wrote down how the work actually happens, including the workarounds and the expertise that lives in one person's head, decided which steps belong to people and which to AI, and only then automated anything.

Why it matters for marketing teams

Most stalled AI rollouts aren't a tooling problem. When the starting process is vague, AI runs anyway, and you find out what the vagueness cost later, in rework and in output nobody trusts.

What to do about redesign before you automate at your team's AI level

Exploring: People use AI on their own, mostly through chat and prompting.
  1. Pick one task you do every week and write down how you actually do it, including the shortcut you'd never put in a process doc.
  2. Paste that into your AI tool and ask it to find the redundant steps and the decisions you make without noticing.
  3. Fix the process on paper before you ask AI to do any of it.
Operationalizing: The team shares prompts, skills and workflows, and has some rules.
  1. Map one team workflow end to end with the people who run it, not just the people who manage it.
  2. Label each step human only, AI-assisted or AI-led, and name an owner for each human checkpoint.
  3. Turn the redesigned version into the team's single shared SOP.
Building: You're running agents, automations or custom tools in real work.
  1. Before any build, write a short spec: inputs, decision points, human approval gates and the one metric that says it worked.
  2. Automate the redesigned process, not the original.
  3. Measure cycle time and rework before and after, so you know whether the build helped.

MAICON 2026 sessions on this theme

The apprentice problem: what MAICON 2026 said and what to do about it

Discussed in 6 of the 32 MAICON 2026 sessions.

What came up

AI is absorbing the entry-level work people used to learn from, and several speakers asked how future experts get made. New roles are forming around designing systems, directing agents and judging output. Middle managers without domain expertise or AI skills were named as the most exposed, and apprenticeship models came up again and again.

Why it matters for marketing teams

Cutting junior roles saves money this year and leaves you short of senior talent in five. Teams need a deliberate way to pass judgment down once the grunt work disappears.

What to do about the apprentice problem at your team's AI level

Exploring: People use AI on their own, mostly through chat and prompting.
  1. Learn the reasoning behind the tasks you hand to AI, not just the output.
  2. Ask a senior colleague to walk you through how they'd judge an AI draft.
  3. Aim to be the person in your field who uses AI best and most responsibly.
Operationalizing: The team shares prompts, skills and workflows, and has some rules.
  1. Pair junior and senior people on AI-assisted work so juniors see how judgment is applied.
  2. Rewrite job descriptions to include agent and system work.
  3. Decide where human leadership is non-negotiable, and say so.
Building: You're running agents, automations or custom tools in real work.
  1. Design workflows where juniors own a review step and seniors check it, so the system teaches.
  2. Capture expert judgment in skills and SOPs so it outlasts any one person.
  3. Reinvest the capacity AI frees into growth roles instead of only cutting.

MAICON 2026 sessions on this theme

Human judgment is the job: what MAICON 2026 said and what to do about it

Discussed in 6 of the 32 MAICON 2026 sessions.

What came up

As output gets cheap, the human contribution moves to framing the problem, choosing the context, asking better questions and making the final call. Speakers described AI as a sparring partner for creativity as well as an efficiency tool, and warned that treating it as a straight substitute for people is a race to the bottom.

Why it matters for marketing teams

When everyone has the same tools, taste and judgment are where the difference shows. Teams that hand those off lose what made their work theirs.

What to do about human judgment is the job at your team's AI level

Exploring: People use AI on their own, mostly through chat and prompting.
  1. Use AI to argue with your ideas, not just to finish them.
  2. Before accepting a draft, write one sentence on why it's right for this audience.
  3. Practice asking a sharper question before you prompt.
Operationalizing: The team shares prompts, skills and workflows, and has some rules.
  1. Match each AI use case to how much variation it can tolerate, and set human oversight to match.
  2. Make final review an explicit role with a named owner.
  3. Talk as a team about what you reward when output gets cheap.
Building: You're running agents, automations or custom tools in real work.
  1. Use generative tools for exploration and critique, and predictable code for processes that must be consistent.
  2. Build critique and verification steps into every pipeline.
  3. Design systems that free people for the creative and relationship work only they can do.

MAICON 2026 sessions on this theme

Create what AI can't: what MAICON 2026 said and what to do about it

Discussed in 5 of the 32 MAICON 2026 sessions.

What came up

If AI can create a piece of content without your input, the argument went, stop creating it. The value is in original inputs AI doesn't have, like expert interviews, original data, customer conversations and lived experience. Those become anchor assets that AI helps scale across formats and teams, and eventually a knowledge base people and AI can query.

Why it matters for marketing teams

Content supply is exploding while attention shrinks, and buyers increasingly get their answers from AI. Generic content now competes with endless free copies of itself.

What to do about create what ai can't at your team's AI level

Exploring: People use AI on their own, mostly through chat and prompting.
  1. Before you write, ask what you know that a model doesn't.
  2. Have AI interview you and draw out a real story instead of inventing one.
  3. Record customer and expert conversations so they can be used later.
Operationalizing: The team shares prompts, skills and workflows, and has some rules.
  1. Plan content around one original input each quarter, such as a survey, expert interviews or proprietary data.
  2. Turn each anchor asset into a set of formats for channels and for internal teams like sales.
  3. Keep a shared library of original inputs the whole team can draw from.
Building: You're running agents, automations or custom tools in real work.
  1. Put your content archive somewhere people and AI can both query it.
  2. Build skills that turn one anchor asset into a portfolio of drafts for human review.
  3. Set up agents that monitor and refresh content from that knowledge base.

MAICON 2026 sessions on this theme

Always-on intelligence: what MAICON 2026 said and what to do about it

Discussed in 4 of the 32 MAICON 2026 sessions.

What came up

Market research doesn't have to be a quarterly project. Speakers showed AI pulling customer reviews, competitor moves and sales data into one view on a schedule, verifying sources and ending every insight with a recommended action. Short, visual outputs got read and long reports didn't.

Why it matters for marketing teams

Leaders decide with whatever information is in front of them that week. If insight arrives monthly in a long document, it loses to opinion.

What to do about always-on intelligence at your team's AI level

Exploring: People use AI on their own, mostly through chat and prompting.
  1. Use AI deep research on one competitive question, and check every source it cites.
  2. Ask AI to summarize recent customer reviews of you and one competitor.
  3. End every summary you share with one recommended action.
Operationalizing: The team shares prompts, skills and workflows, and has some rules.
  1. Set up a recurring research brief on the questions leadership asks most.
  2. Combine customer voice, competitor activity and sales notes into one shared view.
  3. Agree on a verification step before research gets shared.
Building: You're running agents, automations or custom tools in real work.
  1. Schedule automated listening across reviews, social channels and competitor sites.
  2. Deliver insight as a short, phone-friendly page with recommended moves.
  3. Feed findings back into your knowledge base so they build on each other.

MAICON 2026 sessions on this theme

Growth, not just efficiency: what MAICON 2026 said and what to do about it

Discussed in 4 of the 32 MAICON 2026 sessions.

What came up

Speed gains were treated as the baseline. The bigger opportunity was work that used to be impossible because no team had the scale for it, like reviewing every account or testing every message on every persona. Several speakers argued that capacity freed by AI should go into growing the business, not only into cutting cost.

Why it matters for marketing teams

Efficiency gains get competed away fast. Teams that only use AI to do the same work for less end up in a race to the bottom.

What to do about growth, not just efficiency at your team's AI level

Exploring: People use AI on their own, mostly through chat and prompting.
  1. List one task you skip because there's never time, and try it with AI.
  2. Track the time AI saves you, and decide on purpose where it goes.
  3. Write down a question you've never had the bandwidth to answer.
Operationalizing: The team shares prompts, skills and workflows, and has some rules.
  1. Name one previously impossible task each quarter and pilot it.
  2. Decide as a team how reclaimed hours get reinvested.
  3. Report growth outcomes alongside time saved.
Building: You're running agents, automations or custom tools in real work.
  1. Build reusable workflows for work that only failed to scale because of headcount.
  2. Run personalization and testing at a scale people couldn't manage by hand.
  3. Redesign roles around growth work as automation takes over routine tasks.

MAICON 2026 sessions on this theme

How the MAICON conversation changed from 2025 to 2026

We compared 33 MAICON 2025 session summaries with 32 from MAICON 2026. Last year was about getting permission and getting started. This year was about building the systems.

Culture and governance were the 2025 conversation

Culture: 13.5% of the conversation in 2025, 7.4% in 2026. Last year the biggest topics were permission topics: getting people on board, deciding who owns AI, setting the rules and proving it's worth it. They're still here in 2026, but they've moved from being the topic to being the starting assumption.

Agents went from a future slide to a live demo

Agentic workflows: 9.0% of the conversation in 2025, 14.9% in 2026. In 2025, agents mostly showed up as the top rung of a maturity ladder. This year speakers built them on stage, showed where they broke and rebuilt them.

AI search moved to the main stage

AI visibility: 3.0% of the conversation in 2025, 7.4% in 2026. Being found and cited by AI assistants came up in a few 2025 sessions, mostly about how search was changing. In 2026 the question was how to earn credibility inside the answers, and its share more than doubled.

Custom GPTs and prompting grew up

Custom GPTs: a main topic in 8 sessions in 2025, and none built around them in 2026. They were two of the most common topics in 2025. A year later, OpenAI is retiring custom GPTs altogether: they stop running at the end of this year, and creators are being moved to plugins, where each GPT's instructions become a skill. Agentic tools have taken center stage. Prompts turned into skills and context, and GPTs turned into agents and plugins. Source: OpenAI: Custom GPT retirement and migration FAQ.

Every theme's share of the conversation, 2025 vs 2026

Theme20252026Change (points)
Agentic workflows9.0%14.9%+5.9
AI visibility3.0%7.4%+4.4
Growth1.5%4.3%+2.8
Intelligence2.3%4.3%+2.0
Context8.3%9.6%+1.3
Original inputs4.5%5.3%+0.8
Redesign first6.0%6.4%+0.4
Builders9.8%9.6%-0.2
Measurement10.5%8.5%-2.0
Human judgment9.0%6.4%-2.6
Guardrails12.8%9.6%-3.2
Future of work9.8%6.4%-3.4
Culture13.5%7.4%-6.1

Same speakers, one year later

  • Paul Roetzer: “The Move 37 Moment for Knowledge Workers” (2025) → “The Architect, The Orchestrator, and The Apprentice” (2026). From warning that AI will outperform us to redesigning how work and careers are built around it.
  • Mike Kaput: “30 AI Tools Shaping the Future of Marketing” (2025) → “The AI-Native Content Team” (2026). From a tour of the tools to a playbook for how a content team runs.
  • Jessica Hreha: “The Missing Role in Marketing AI: Why You Need a Transformation Leader” (2025) → “The Bottleneck of One: Why Your Marketing AI Leader Needs Builders” (2026). From making the case for an AI leader to making the case for the builders that leader needs.
  • Andy Crestodina: “Better than Prompts: How to Build Custom GPTs for Marketers” (2025) → “From Process to Prompt to Automation: How a Marketing Job Becomes a Marketing Tool” (2026). From custom GPTs to turning a whole job into a tool.
  • Liza Adams: “How to Build AI Marketing Teammates: Custom GPTs in Action That Transform Teams” (2025) → “How to Reimagine and Build Workflows That Weren't Possible Before” (2026). From GPT teammates to workflows nobody could run before.
  • Brian Piper: “Transform Your Content with AI and Data” (2025) → “Building a Content Marketing Agentic Workflow” (2026). From personal assistants to a connected content pipeline.
  • Keith Moehring: “Make Your Own AI-Powered Analytics Assistant” (2025) → “AI-Native Playbook: From Solo to Scalable” (2026). From one assistant to an operating model for the team.
  • Brian Brinkman: “Optimize Your Organization with AI + Automation” (2025) → “Build Your Own Marketing App in 40 Minutes, No Code Required” (2026). From wiring automations together to building the app yourself.
  • Wil Reynolds: “The Future of Search: From SEO to Generative Discovery” (2025) → “Seen. Believed. Chosen: AI Optimization That Builds Visibility and Credibility” (2026). From explaining how search is changing to earning credibility inside AI answers.
  • Pam Boiros: “The Problem-First Approach to AI Adoption” (2025) → “AI Adaptation: The People Side of Scale” (2026). From adopting AI to adapting how people work at scale.
  • Jim Sterne: “Integrating Generative AI Into Your Organization: From Vision to Implementation” (2025) → “Trust, Verify, Let It Fly: Surfing the Certainty-to-Serendipity Spectrum” (2026). From rolling AI out to deciding how much autonomy each use case gets.
  • Dale Bertrand: “How AI Can Elevate Human Marketing Creativity” (2025) → “AI-Driven Content Performance Monitoring” (2026). From creative help to always-on monitoring.
  • Taylor Radey: “Expertise, Squared: The AI Toolkit for B2B Marketing Teams” (2025) → “AI Workflows to Research Anything, Better” (2026). From a toolkit to repeatable research workflows.
  • Christopher S. Penn: “From Text to Video in Seconds (with Katie Robbert)” (2025) → “Build an Analytics Powerhouse with AI” (2026). From generating media to building your own analytics engine.

How we measured this: each session was tagged with the themes it covered in substance. A theme's share is its tags as a percentage of all theme tags that year. Sponsored sessions are excluded.

We help teams take AI workflows from discovery to a running system.

Talk to Clarity & Motion

The actions are our own guidance, not the speakers'.