The request is always the same: which AI tool should we use?
It's the wrong question. The right question is how you move a team from scattered experiments to fluency. From personal tinkering on weekends to workflows that actually reduce friction during the workday.
Eku Malcolm knows the difference because he's built it. Before joining commonsku as Director of AI Operations, he spent four years at OfficeSpace Software turning AI from buzzword into operating principle. The company scaled from $18 million to $42 million ARR while cutting manual work by 30%. Not through tool selection. Through enablement.
His first week at commonsku? He built a custom GPT in roughly 15 minutes that teaches people how to write better prompts. Team members describe their problem. The tool generates an optimized prompt they can use immediately. No expertise required. Just better output for everyone.
That's what 2026 demands. Less tinkering. More systems.
We dive into:
Stop asking which AI tool you should use. The real question is simpler: how do we start without creating chaos?
Eku's first move at commonsku was running a survey. The results showed that 94% of the team already uses ChatGPT. The barrier isn't adoption or awareness. It's integration. Teams know what AI can do. They're overwhelmed by how it fits into workflows that already work, not confused about whether AI matters.
The shift from experimentation to fluency requires guardrails. Not the kind that kill innovation, but frameworks that build confidence so people actually use the tools. Leaders want to introduce AI without triggering fear, shiny object syndrome, or the paralysis that comes from 47 different options.
The mandate is straightforward: help teams make fewer decisions while accomplishing more work.
Remember when teachers insisted you wouldn't always have a calculator in your pocket? That warning feels quaint now that everyone carries a supercomputer everywhere they go.
Generative AI sits at that same inflection point. What seemed impossible is suddenly at your fingertips, and the only way forward is experimentation. Use it. Try it. Break things if you need to.
Eku breaks down the critical distinction between generative and agentic AI. Generative AI creates content. Drafts, analysis, ideas. It requires human triggers at the start and human review at the end. Agentic AI runs autonomously in the background, executing tasks without constant human intervention.
Understanding this difference shapes your entire strategy. Fifty personalized emails you need drafted? That's generative AI with a human trigger. A system that automatically sends follow-up emails based on client behavior? That's agentic AI.
Both have their place. The mistake is treating them the same way.
Teams are split. Some are Google houses running Gemini. Others are Microsoft houses with Copilot. Meanwhile, ChatGPT, Claude, and a dozen other tools compete for attention daily.
Eku's advice: start with one model and let it build for a while.
Most teams already use ChatGPT in their personal lives. That familiarity matters. It'll get you 80% of the way there for most tasks. Whether it's the absolute best tool for every specific use case doesn't matter initially—what matters is adoption and fluency.
Start with one platform. Build enablement around it. Identify where it falls short. Then strategically add specialized tools for specific use cases—Claude for coding, specific models for financial analysis, whatever your outliers require.
The goal isn't perfection. It's progress with consistency.
Customer data is the obvious line. Don't put customer information into AI tools where you can't control how that data educates the backend models.
The solution: paid licenses. Freemium tools sacrifice something—usually your data. When you pay for Gemini Pro, ChatGPT Plus, or Claude Pro, your workspace data stays private. It doesn't train the masses.
Eku's rule of thumb: if you'd be comfortable seeing it on the front page of a newspaper, it's probably fine for open-source AI. If not, lock it down in paid, private environments.
Note-taking tools that jump on video calls perfectly illustrate this balance. Incredibly valuable for capturing transcripts and summaries. But where does that information go? If it's a paid license, you're likely protected. If it's free, you need to know exactly what's happening with your data.
Establish these guardrails early. They create confidence to experiment aggressively within safe boundaries.
Overnight transformations define this space. Gemini's Imagen release changed image generation completely. What every LLM struggled with suddenly became easy for one tool—and it wasn't close.
Agentic AI continues climbing. HubSpot, Salesforce, massive players investing millions in R&D. They're getting the most use cases because they're pouring resources into solving specific problems at scale.
But here's what Eku sees coming: after AI proliferation everywhere, we're entering a tactical phase. The initial excitement is recalibrating. People are asking smarter questions. Putting up intelligent guardrails. Protecting themselves from themselves.
If your workflow is repetitive, manual, or causing pain—AI can probably help. The doors of efficiency it opens let teams focus on work they want to do instead of work they have to do.
Human intervention will always bookend these processes. Humans start. Humans review. But the middle? That's where AI eliminates friction and reclaims time.
AI implementation isn't a tool problem. It's a people problem disguised as a technology question.
The companies that scale AI successfully in 2026 will do three things well. First, understand where teams actually stand with AI comfort and usage. Second, identify specific pain points that AI can address without creating new problems. Third, establish clear boundaries about what they won't use AI for.
The promotional products industry prides itself on relationships. AI doesn't replace those connections. It eliminates the friction that prevents you from deepening them.
commonsku is betting big on AI and automation in 2026. Both strategic initiatives aim to automate backend workflows so distributors can focus on progressive sales, creative problem-solving, and the client relationships that actually matter.
Ready or not, 2026 is when you stop tinkering and start scaling. The question isn't whether AI belongs in your business. It's whether you'll lead the transformation or watch from the sidelines.
[00:01:51] How to start without creating chaos
[00:07:50] Generative AI as this generation's calculator
[00:08:38] Building the custom GPT prompt architect
[00:11:09] Generative AI vs. agentic AI
[00:14:28] Starting with one platform
[00:16:58] Two pillars: operational rigor and enablement
[00:19:15] Balancing AI risk with guardrails
[00:22:45] From proliferation to tactical implementation
[00:25:38] Three tips for leaders heading into 2026