How to Use Agentic AI in Marketing Strategies

AI is BS. Not the technology. The talk track. I attended the NRF Foundation Big Show this week, and everything was “AI-something.” AI for inventory. AI for pricing. AI for customer service. AI for world peace (okay, maybe not yet). With all that noise, it’s easy to feel overwhelmed—and a bit cynical. The possibilities are incredible, but slapping “AI” on everything doesn’t make it useful. Understanding how these tools work to solve actual business problems is critical. I’ve found it’s helpful to kind of simplify it into the two categories that really matter: ✍️ Generative AI is like an extremely knowledgeable friend who can produce new things—written content, images, & beyond—if asked in just the right way. A chatbot interface makes that generative AI friend more accessible: you give it a prompt (for example, “Write a short product description for a new running shoe”), and it instantly creates a response from all the information it has internalized. 🕵️♀️ Agentic AI goes further. It's more like a proactive personal assistant with the same deep knowledge. Instead of waiting on precise prompts, it can infer tasks and even carry them out automatically. For example, it can figure out when stock is running low & reorder items without being explicitly told every step to take. How retailers might use each: 1️⃣ Generative AI: Product Descriptions: Automatically create rich, engaging product descriptions for online catalogs that match the brand’s voice. Marketing Content: Draft email campaigns, social media copy, & blog posts. Store Layouts & Visuals: Suggest store display ideas or mockups, using AI-generated images to spark new merchandising concepts. 2️⃣ Agentic AI: Inventory Management: Monitor incoming sales data & reorder items proactively before inventory runs out. Customer Service Automation: Act on customer requests (like returns or shipping updates) without a staff member walking it through each step. Dynamic Pricing: Continuously check market trends, competitor prices, and demand patterns, then adjust product prices accordingly—without needing a person to oversee it all. I think Agentic AI will provide the biggest benefits and the biggest disruptions because consumers love convenience and businesses love efficiency – and it delivers both. AI is evolving faster than Moore’s Law—doubling every 3 months instead of 18. Do the math—it’s mind-blowing. Moore’s Law gets you 10X improvement in 5 years. At this pace, AI could be 1,000,000X in 5 years! (h/t Kasey Lobaugh) In just a few years, we could see retail transformed by super-powered sales associates, hyper-personalized shopping journeys, and supply chains optimized to unimaginable levels. But first we have to cut through the noise to make sure we’re making the right choices. Are you experimenting with any tools successfully—or are you overwhelmed by the hype (or both!)? #AI #agenticAI #agents #retail #NRF

After building 58 AI Agents, here are 12 essential tips (steal these and get ahead): 1. Delegate. - Stop thinking, “What manual processes can I automate?” - Instead, ask, “If I had a marketing agency, what would I want them to handle?” - Think bigger—AI isn’t just a time-saver, it’s a workforce multiplier. 2. Automation vs. AI Automation vs. AI Agents. - Automation: A series of steps executed automatically. - AI Automation: The same, but with an AI step. - AI Agents: Decide how to act, what to do, and what data to use. 3. AI Agents go beyond chat. 3 ways to trigger an Agent: - On demand (chat or button click). - On a webpage (via Chrome extension). - Via an event (just like an automation). 4. Use ChatGPT (or similar) to build. - Writing clear instructions (“prompts”) is harder than it looks. - Determining an Agent’s decision-making process is even harder. - ChatGPT is an essential tool for thinking through both. 5. There’s a fine line between useful and over-engineered. - Simple Agents get used. Complex ones get abandoned. - Start small—iterate later. - Traditional automation is no different. 6. Stronger use cases I’ve found: - Prioritizing feature requests based on product strategy - Pulling insights from a Zapier Table of consolidated data (cost savings, top-performing areas, etc.). - Researching a company, person, or product—then structuring the data and determining when to notify someone. 7. Use decision-making frameworks. - AI Agents, like humans, need structured decision-making. - MoSCoW, Eisenhower Matrix, SWOT—pick one and embed it. - You’ll understand why your Agent made a decision, not just what it did. 8. Data sources are the most powerful component. - Agents process large data sets instantly—that’s their edge. - The better your data, the better your Agent. - Build robust databases, and your Agents will thrive. 9. Agents need systems (just like you). - The future isn’t just Agents—it’s Agents + Tables + Workflows + Interfaces. - You’re not just automating—you’re designing an AI-powered organization. - Systems > Standalone Agents. 10. Two essential skills for building. - Delegating future work (that you've already done before). - Pushing the Agent to tackle tasks that haven’t been done before. - Both require serious brainpower and take time to master. 11. Set guardrails while also allowing for mistakes. - Restrict access in integrated apps to avoid risk. - Be okay with the Agent making some mistakes. - Master the balancing act to become an expert Agent builder. 12. The biggest bottleneck is you. - Are you clear on priorities? Goals? Expectations? - An Agent can only be as clear as you are. - Get your own systems right, and your AI will follow. One of the best skills you can learn in 2025 is Agent building. Models are getting better every. single. day. They'll do more and be smarter. Best way to learn: start building. Let's all learn together 💪 Consider subscribing to my newsletter: https://lnkd.in/gtxpSwap

Prediction: We will see a massive evolution in restaurant marketing this year. Agencies ➡️ Agents Recently, Scott Lawton and I discussed how AI-powered agents, like OpenAI’s Operator or Salesforce’s agentic workforce concept (aka “digital labor”), are shaking up the traditional model of relying on agencies for mass media campaigns and redefining how we approach marketing in the modern era. AI agents can analyze your data and understand context using machine learning and natural language processing. Within the restaurant marketing department, it can segment your guests and personalize interactions down to a segment-of-one level, delivering the right message, at the right time, to the right guest through their preferred channels. Tailored to each individual, this approach can help drive desired behaviors: increased visit frequency, higher spend per visit, and broader menu exploration/greater mix—ultimately boosting their guest lifetime value to the brand. For restaurant brands managing ad budgets that account for billions annually, this will be revolutionary. Those dollars can now fuel precise, measurable guest connections instead of broad, one-size-fits-all campaigns. For the small and mighty brands—this could level the playing field and offload the burden of manual work. At Olo, our focus on guest data empowers brands to seamlessly step into this AI/agentic marketing era. With the right tools, the industry can turn data into deeper loyalty and greater lifetime value, all while crafting experiences guests genuinely enjoy. Is agentic AI the future of restaurant marketing? Let’s talk about it.

😎 How I designed a complex multi-agent system to run a marketing department for a US Telco. Here’s how I approached this complex multi-agent system design — shifting beyond the traditional thinking 'ai agent == human'. ✅ Flow: We need to think about agents and their purpose more in terms of information design and flow of decisions, concepts and ideas. Just because a marketing analyst has a specific role, it doesn't mean there's a direct 1:1 mapping to an AI system. We need to break out of this thinking; a common trap of Conway's law that seasoned engineers are falling into. ✅ People: Designing large-scale AI agentic systems goes beyond just tech. It’s equally about people and process. When should the human-in-the-loop step in? And what should they do? These decisions are pivotal in ensuring seamless AI-human collaboration ✅ Modularise: Breaking complex systems into manageable layers and creating hierarchies of information are crucial to diagnosing large-scale agentic systems. A modular approach lets AI agent ‘teams’ effectively ‘hand over’ tasks to one another, anchored by a live artifact like a dynamic marketing strategy document that is native to humans. ✅ Feedback Loops: Keeping track of decisions made by AI agents through a decision register or log is essential. By comparing these decisions to outcomes, agents can learn from mistakes and capitalise on successful strategies, allowing for faster scaling and continuous improvement ✅ Cold Start: In environments with no pre-existing assets or data, how do we give AI agents a meaningful starting point? The human role becomes crucial here — to provide that initial seed data and context to kickstart the AI’s decision-making. In this scenario I decided to use the gap analysis from competitors as the initial starting point. ✅ External Data: How and when can agents become more grounded through use of live data. In marketing sense we can pull in competitor campaigns and census data in this example to ground the choices on real-world outcomes. P.S. This is part of an ongoing process and not a final design, but wanted to share my thoughts with my followers!

After my last post, someone asked, “So what use cases are you actually seeing value from AI?” So here’s a peek into how we’ve been using AI across our Marketing team. Below is our actual team/org chart with AI Agents as part of the team 𝗢𝗻 𝘁𝗵𝗲 𝗢𝘂𝘁𝗯𝗼𝘂𝗻𝗱 𝘀𝗶𝗱𝗲 – 𝗔𝗴𝗲𝗻𝘁𝘀: 𝗚𝗲𝗺-𝗘𝘀 We’ve been fortunate to have Gem-E, our AI outbound agent, early on so this side of the team is more built out & measured Impact: Outbound capacity increased by 74% after 9 months Jobs to be done: - Research & Capture All Buying Signals & CRM Data - Account & Contact Scoring & Prioritization - Building Lists of Contacts - Writing Outbound Emails (& sometimes Sending on Autopilot) We apply Gem-E to specific Campaigns like: Past Champions, ABM Prioritization & Outreach, Revive Closed Lost, Event Outreach That makes it super easy to measure the pipeline impact. 𝗢𝗻 𝘁𝗵𝗲 𝗠𝗮𝗿𝗸𝗲𝘁𝗶𝗻𝗴 𝘀𝗶𝗱𝗲 – 𝗔𝗴𝗲𝗻𝘁𝘀: 𝗚𝗮𝗶𝗮, 𝗚𝘂𝘀, 𝗚𝗶𝗻𝗻𝘆, 𝗚𝗶𝗻𝗮, … We’ve started creating internal custom GPTs for specific “jobs to be done.” Impact: The business value here is harder to measure besides time saved (I’m still working on quantifying this). Personally, these Agents help with easier context switching, faster output, and fewer repetitive tasks. A new competitive comparison used to take 45-60 minutes, now takes 15. Jobs to be done: - Market research - Video-to-Playbook - Battle Card - Paid Ads Performance Reporting - Email Nurture 𝗪𝗵𝘆 𝗜'𝗺 𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗶𝗻𝗴 𝘁𝗵𝗲 𝗢𝗿𝗴 𝗰𝗵𝗮𝗿𝘁 𝗹𝗶𝗸𝗲 𝘁𝗵𝗶𝘀? I noticed that all our team members already use AI to do their work — but their GPTs and training are "siloed" in their personal instances. The rest of the team cannot access it to get the output, like "a battle card for X" 👉 So we got everyone on the Corporate Team Plan (maybe Enterprise soon tbh) Many AI experiments tend to fizzle out because they don't align with clear business initiatives or tangible impacts. Combining with the silo-nature above, they tend not to last. 👉 So we have each team lead treat their Agents as direct reports — the lead is responsible for creating (~hiring) & improving (~managing & coaching), and let the company work with the Agents (~ visibility) My hypothesis is that if there are clear owners & visibility, the Managers will make sure the Agents add value to the team & the company. That’s the shift I think & hope will matter most long-term: Not just using AI, but operationalizing it. Treating Agents not as novelty tools, but as teammates that need direction, coaching, and integration into the way we work. Curious how others are doing this. How are you making AI a durable part of your workflows & teams, and not just an experiment that fizzles out?

Here’s a tip for using AI to solve complex tasks: Add models to multiple value.    How? In my latest Code to Care video, I venture into the world of agentic AI and explain why linking models leads to better outcomes.    Imagine using a large language model to create a marketing presentation.    You feed the model a request and background information, and it crafts a response by predicting one word at a time — The… best… marketing… presentation… — until the task is complete.    Efficient? Yes. But LLMs need to do this work without a back button. But you probably don’t work that way – none of us do. You’d probably take a step-by-step approach that leverages revisions: Write an outline, draft the presentation, edit, ask for feedback, rewrite, and so on.    Agentic AI lets you mimic this approach by linking multiple LLMs back to back. (In this arena, we refer to LLMs as “agents.”)    For example, you might break the task into three steps: asking one agent to draft a presentation, a second agent to critique the presentation, and a third agent to refine the first draft with the critique.    Presto. You have a better presentation.    Watch the video to learn more about agentic AI and how you can even use agents to plan and manage the step-by-step process for you.