Kerry Morrison | AI for marketers

AI doesn't replace what you know. It multiplies it. 25 years in marketing. I share the AI agent workflows I actually use. #MarketerMultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

AI doesn't save you time on day one. It saves you time every week after.

Here's what that looked like when I built the agents behind my newsletter, and how I'd start if I did it again.

[Kerry Morrison | AI for marketers @AIGrowthCo]

@AnthonyGaenzle For all of us it still comes down to a strategic question. We're going to work the same amount of time. How do we want to spend those hours?

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@AnthonyGaenzle The larger picture is real: we can do things we couldn't do before, and we can make bets we couldn't have made when everything took longer and cost a lot more to produce.

[Kerry Morrison | AI for marketers @AIGrowthCo]

@fivosaresti Fivos's full breakdown: https://t.co/C36pVkEeHV

[Fivos Aresti @fivosaresti]

The way I operate my agency wouldn’t have been possible 1 year ago.

We uniquely blend software and human expertise.

And call it “AI-Native Services”.

Here’s how it works operationally:

1️⃣ Company OS in git

Our entire company dataset lives in one GitHub repo called Company OS.

What's inside:

• company/ - team, voice guide, design system, industry intel
• wiki/ - SOPs, playbooks, campaign guides
• clients/ - per-client context files
• raw/ - client calls, market research, competitor data
• plugin/ - 26 agents, 23 commands, hooks
• skills/ - 79 Claude skills

Data is constantly flowing in to keep it up-to-date.

2️⃣ Client repos

Every client gets their own private repo.

Same engineering pattern as the Company OS, just personalized to their account.

What's inside:

• Their ICP, voice guide, brand assets
• Historical campaigns and what worked
• Onboarding form data and deep research
• Slack threads, call transcripts, GDrive changes
• API/MCP connections to their revenue stack

Result: every team member has full client + company in every session.

3️⃣ Human interaction layer

We still log into some SaaS UIs, but Claude is slowly taking over.

Across 20 team members, the efficiency gain has been massive.

We try to automate as much of the admin as possible:

• Client onboarding
• Content research and ideation
• Skill tuning from team feedback
• Reply triage + sentiment routing
• Campaign launch pre-flight checks

So AI does the legwork, but humans ship.

So we can spend more time on strategy + creative GTM.

4️⃣ MCP + CLI engine

MCPs + CLIs let Claude act across our stack vs. just advise.

Some of our favorite MCPs/CLIs:

• GitHub - Company OS + client repos
• Findymail - email verification waterfall
• Google Workspace - client docs
• Airtable - automation backend
• InstantlyAI - email campaigns
• Slack - team + client comms
• Apollo - list + enrichment
• Notion - internal wiki + PJM
• HeyReach - DM sequences

Plus HubSpot, Browserbase, Supabase, Vercel, Figma, Stripe, Pinecone, Clay, Apify, Firecrawl, and more.

We're also migrating a ton of workflows to custom code.

5️⃣ Operationalize

As an AI-native services company, we're constantly optimizing how we work with AI and software.

Built into the system:

• Guardrails: safety hooks gate 94+ risky operations.
• PR-based governance: anyone on the team can propose a new skill, agent, or tweak as a branch.
• Workflows-engineering plugin: 26 agents, 79 skills, 23 commands auto-propagated. Agent swarms split tasks into 5-20 sub-agents.
• Self-improvement loop: n8n syncs tech stack data back into the Company OS. Pinecone stores past content + performance metrics for skills to query. Human corrections feed back in.

There isn't ever going to be a finish line, so we're building like it's a marathon.

[Kerry Morrison | AI for marketers @AIGrowthCo]

@fivosaresti If you do client work and want a small start, here's what I'd do:

1. One folder per client
2. Drop in call notes after every call
3. Add a short "what worked, what didn't" file
4. Point your AI at that folder before each task, so it starts with the full picture

[Kerry Morrison | AI for marketers @AIGrowthCo]

@fivosaresti Fivos, if someone could build only one folder from your setup this week, which would you pick first?

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@OGMayank Fair. The writing is where AI gets caught. Where it still helps is before writing: finding what to say. Feed it real customer calls and comments and ask what questions keep coming up. Better ideas than any "write me a post" prompt. You still write it.

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@Origin_AI_01 Good starter map. One tip from running my own: skip the team leader bot at first. Start with two or three narrow bots that each do one job well, and you be the leader for a few weeks. Write down what you keep telling them. That's the leader's instructions.

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@webjuice_ie @squad_so_ "Nothing needs you" is the dream screen. What I watch in my own bot fleet is the handoffs. Each bot is fine alone, but bad inputs slide through when one passes work on. A short "what I got, what I'm passing on" note at each handoff shows where it broke.

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@jamiequint You can get close today by building them. I run narrow bots, each with one job and a written rubric from someone who knows that job. Same frontier model. The expertise is the rubric and good and bad examples. A great checklist beats a smarter model.

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@Jessie_HealyDTC Best answer I've found is a boring daily check. A small AI job compares yesterday's spend, purchases, and pixel events to the same day last week and flags anything off by a set amount. When the group chat panics, you already know if it's you.

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@JacquesHBastien @BookingKoala That hour is where most home service leads die. The other half is the ones who don't book. I'd have AI sort them by what they asked for and send a follow up that answers their likely worry, like price or timing. Then cut the search terms that never book.

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@noahiglerSEO True, and the smaller shop has one more edge: it reads its data faster. Have AI go through your search terms report weekly and pull out the jobs you actually win. Use those exact services in your Google profile posts and service pages. Ads feed organic.

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@seo_bint_ejamil Most of it is old local SEO. What feels new is review wording. AI answers seem to lift customer words, so "fixed our furnace same day in Burnaby" beats "great service." Have AI find services and towns your reviews never mention, then ask those customers.

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@om_patel5 Love that you went for the tests and skipped the naming fight. Cheap version: a small bot asks each AI the same 20 buyer questions every week and logs who gets named and which pages get cited. A month of that beats the naming debate on what to build next.

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@alexgroberman The category gap is real. Don't guess prompts — feed Claude or Codex a pile of sales-call questions and test those. AI makes that volume usable fast. Buyers rarely name the category like the site does. Fix that wording first.

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@MichLieben Same-day for Tier 1 is right.

In outreach I'd have AI write the first line from what they did (the post they engaged with), then run a "would a human say this?" check before it goes out.

Tier 2 and 3 get a slower drip — with a real reason each time.

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@JamesonCamp The transcript one is the sleeper. I put a huge pile of sales calls through Claude — not for summaries. It pulled the exact buyer words, counted how many calls said them, and surfaced themes and pain points we'd missed. One call lies. Patterns don't.

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@andrew_jennings Good walkthrough. I'd add what the agent is not allowed to do. In my ads work it reads everything and drafts changes, but big budget moves and new audiences wait for a human yes. Have it write a short "why" next to each change to see which ideas paid off.

#marketingmultiplied

[Kerry Morrison | AI for marketers @AIGrowthCo]

@fivosaresti His line sums it up: "AI does the legwork, but humans ship."

The tools free up time. His team spends it on strategy and creative work. The judgment stays human, and he says human corrections feed back into the system.

[Kerry Morrison | AI for marketers @AIGrowthCo]

@fivosaresti The part I like most: every client gets their own private repo.

Their ideal customer. Their voice guide. Past campaigns and what worked. Call transcripts. Slack threads.

That's where the value lives.

[Kerry Morrison | AI for marketers @AIGrowthCo]

@fivosaresti You could copy his folder layout in an afternoon. You can't copy what's inside it.

Years of calls, what each client really cares about, which campaigns flopped and why. That only comes from doing the work with real people.