How to Stop Reinventing the Wheel & Let AI Do the Heavy Lifting
Takeaways from Grant Hushek's talk at the Foxwell Meet-Up Lisbon, March 2026
Why it Matters
Most people use AI like a more advanced search engine — they type in questions and hope for useful answers. But Grant’s approach is a game-changer: it's about creating systems that harness your unique expertise, rather than relying on generic prompts.
Here’s why this matters:
Your Best Ideas Keep Slipping Away: Think about all the insights you share during meetings or the frameworks you’ve developed over time. Too often, those brilliant explanations vanish into thin air, stored only in your memory or buried in an old document. When someone needs that information later, you find yourself starting from zero. Grant believes this can be fixed, using AI as the key.
Context Matters: Generic AI tools are like a well-read stranger, knowledgeable but lacking your specific insights. When you share your context, whether it's your CRM data or actual conversation transcripts, the AI becomes more like a trusted team member who understands your processes. The difference between a clueless AI and one that’s in tune with your needs is massive, and many people miss the opportunity to bridge that gap.
Building Systems is a Competitive Advantage: The teams that create and refine these systems now will have an edge later on. Every improvement made benefits the whole group, building a strong foundation over time. The gap between someone effectively using these systems and someone just trying to get by is widening rapidly. This isn’t just a theory; it’s already happening in agencies and firms that embraced this approach early on.
By adopting this mindset, you can ensure your valuable knowledge doesn’t get lost and your team stays ahead of the game.
Here's a thought experiment: write down one task you do regularly at work. Just one. Now ask yourself, what are the actual steps? What's the output? How long does it take you? This is where Grant Hushek started his talk at the Foxwell Founders annual European meetup in Lisbon, Portugal in March 2026, and it sets up everything that followed. Grant runs an AI education and implementation agency, and his whole purpose is operationalizing unique talent. Basically, turning smart people's know-how into repeatable, scalable processes. His tool of choice: Claude.
Here's the framework he walked through.
1. Projects = Your Frame
Think of a Claude project as the lens via which the AI sees your world. With it, you're talking to someone who actually gets your context. Without it, you're just talking to a very well-read stranger.
Grant's example: he read The Power of Value-Based Selling, dropped the PDF (converted to a Markdown file [more on that in a sec]) into a project as a knowledge base, and now gets coached through that specific framework whenever he's prepping for a sales call. Same deal for his CRM. He wrote instructions for navigating HubSpot, explaining what pipeline stages mean and which fields are required. Now he can hand in a call transcript, and it automatically updates his deals.
Tip: Don't feed Claude PDFs. Convert them to Markdown or plain text beforehand. PDFs are basically asking the AI to squint at a blurry photocopy.
2. Connectors = Your Context
"Prompts are dead. Context is king."
The more real-world information you add to Claude, like your emails, Slack messages, call transcripts, and task lists, the less generic and more useful it becomes.
He also uses Fathom to record calls, then stores all the transcripts in Supabase. From there, Claude can search across hundreds of calls to answer questions like "what did every Google client say about pricing?" He literally answered client questions from a taxi because all his call history was at his fingertips through Claude.
Tip: If you don't have a call recorder yet, get one. You won’t regret it, and it’s worth every penny.
Foxwell Founders is a high-level, exclusive membership for the best in digital marketing, and we’d love to have you!
3. Skills = Your Process, Bottled
This is where things get genuinely powerful.
A "skill" in Claude is basically a saved workflow that is a Markdown file that captures all the steps of a process so you don't have to rebuild it from scratch every time. The key thing: don't ask Claude to create a skill upfront. Do the work first, then say "turn this into a skill." That way it learns from what you actually did, not what you think you'll do.
Every time the output isn't quite right, feed it back and say "this is better than my last version, improve the skill." It reads what it made, compares it to what you gave it, and gets smarter.
Skills are also shareable. On Claude Teams, admins can push skills to the whole company. Suddenly, everyone's working from your process, not a watered-down version of it.
Grant's pre-meeting playbook? One skill. It pulls information from Clickup, Slack, call transcripts, client brand docs, and spits out a prep brief. That's months of refinement, available to anyone on the team.
4. Co-work = Automation That Actually Thinks
The last layer is Co-work (Anthropic's desktop agent tool), which lets you schedule Claude to run tasks automatically. Grant has a "vacuum skill" that runs every Thursday at 4pm: it sweeps all his emails, calls, Slack threads, and task updates, consolidates them, and drops a summary agenda into a Slack canvas for his team. No human involved.
That's intelligent automation, and it understands what it's reading and synthesizes it.
The One Thing to Build This Week
If you only have to take one thing from Grant's talk, it’s this: build a branded document skill.
Here's why: the most valuable thing you do at work is think. The ideas you share on calls, the frameworks you explain to clients, the way you break down a problem, that's your actual product. Most of it disappears the moment the call ends.
Using a call recorder + a branded document skill, you can turn any interesting moment from a conversation into a polished, shareable asset. Grant demoed one he'd made the day before. A client asked him something smart; he thought, "that was a good answer," fed the transcript to Claude with his branded doc skill, and had a finished document without writing a single word.
Lead magnet. Client deliverable. LinkedIn post. Whatever you want.
The Big Picture
Work is a collection of tasks. Tasks are steps with deliverables. If you start treating them that way and building tools around them, then you can do significantly more, at a higher quality, in less time.
The stack looks like this:
Projects set your frame
Connectors bring in your context
Skills crystallize your process
Co-work multiplies your output
Start with the task you wrote down at the beginning. That's your entry point.
Tired of inconsistent results? Start here
