Howdy 👋🏾.
Last week, I gave a virtual presentation for Startup Grind Maryland on AI. Instead of pulling out my standard deck, I decided to test-drive something new working draft of my keynote for the upcoming Black Is Tech conference.

As I worked on the slides, one theme kept pulling me in. Not just a hot topic, but the problem I keep hearing from nearly every client I talk to: Their data, and the systems wrapped around it, are holding them back.

It’s becoming more obvious with each conversation. We’re in an age of stove-piped apps, SaaS lock-in, and fragmented platforms that don’t want to play nice. For companies trying to bring AI into their workflows, that’s not just frustrating, it’s a dealbreaker.

I mentioned in a previous newsletter that it’s still far more common to hear about failed AI projects than wins. That’s not because the tech isn’t ready. The potential is massive. But the reality is: if your chatbot can’t access your data, or increasingly worse, it can’t act on it, it’s never going to deliver what your team actually needs.

A few months ago, I would’ve pointed straight to RAG (Retrieval-Augmented Generation) as the fix. Pull in live data from your systems, augment your AI agent’s responses, and you’re good. But we’re entering a new phase. Agentic AI is shifting expectations; it’s not enough to read your data. These agents need to take action. To move data. To orchestrate across systems. And that requires a different kind of infrastructure.

The data layer you build now has to do more than inform; it has to empower. That could mean yes to tools like Snowflake and Databricks, but also yes to lighter-weight API orchestration platforms like n8n and Zapier. Especially for small and mid-sized businesses, that combination of visibility and action is where the real opportunity lives.

So this week’s big deep dive is all about that problem: The Problem with Data

How we got here. What needs to change? How to build systems that let AI do its job. Enjoy!

-jason

👉🏾 At PerryLabs, we advise, implement, and support AI from idea to impact.
Advise. Implement. Support. Real AI, real results with PerryLabs.

---

The Problem with Data

Everyone wants AI that works across their business, but few have the data infrastructure to support it. This week’s deep dive breaks down why most companies are stuck between clunky ERPs and scattered best-of-breed tools, and why data lakes alone aren’t enough anymore. From API orchestration to real-world examples, here’s what needs to change to make AI usable. Read more.

---

📼 Talking Tech: Watch & Learn

A few weeks ago, I had a great time joining Midday on WYPR with guest host Farai Chideya and UMBC’s Anupam Joshi to talk about how AI is transforming the workplace. We dug into what these changes mean across generations, how companies can move beyond the hype, and why digital fluency is becoming critical, not just for tech folks, but for everyone. Grateful for the thoughtful conversation and the chance to connect big shifts in tech to what’s happening on the ground.
🎧 Listen to the full conversation here.

---

🔗 The Best in Tech This Week

💥 Windsurf Whiplash – Talk about a neck-snapping move. Windsurf’s CEO jumped ship to Google, and the OpenAI acquisition collapsed overnight. I’ve got friends who swear by Windsurf, and it looks like Claude might be back.

🛡️ AI Gets Its Security Clearance – OpenAI, Anthropic, and Grok just locked down massive U.S. government contracts. AI is officially moving into the federal stack, powering workflows and replacing manual processes across agencies.

🧨 The Grok Problem – Researchers are raising alarms about Grok’s safety practices and for good reason. This is the model that recently referred to itself as “MechaHitler”. Now it might be embedded in sensitive government workflows? What could go wrong?

---

🎤 The AI Roadshow: Workshops, Talks & Beyond

July 26, 2025 – BLK Tech Connect

August 11, 2025 – Black Is Tech Conference

September 16, 2025 – AI Tech Summit w/ Central Maryland Chamber

September 25, 2025 – BannerX: Where Cybersecurity and AI Shape Business

---

P.S. Before you go…

Thanks again to John Gruber and Daring Fireball for introducing me to my new favorite Mastodon feed: @classicweb@indieweb.social.

I’m starting to realize that when I give talks and reference the AOL days of AI or CD-ROMs, I’m officially dating myself. A whole generation has no idea what the early web even looked like.

In a time when much of the internet feels templated and polished, it’s genuinely fun to scroll through the weird, wonderful design ideas that shaped what we use today. Enjoy! I definitely did.