Devin Venable

Skyline Technology Solutions is honored to support TxDOT's mission to keep Texans safe. Our Sitch platform enables seamless distribution of live feeds from 3,500+ traffic cameras to DriveTexas.org and TxDOT districts, delivering real-time visibility into road conditions, congestion, incidents, and severe weather. Technology partnerships like this showcase how innovation directly impacts public safety. #TxDOT #Innovation #SkylineTechologySolutions #Sitch #TrafficManagement #PublicSafety 

42

👨‍💻 We don’t micromanage at Optavyn. Every dev here knows they’re not “just writing code”, they’re building products that matter. Every feature we ship solves a real-world problem. Every API we build connects with a user’s workflow. Every screen we design has intent behind it. Our team works remote, but we’re aligned, focused, and committed to the end goal: Delivering results not just tasks. If you're a founder or agency who values clear ownership, clean code, and reliable delivery, we might be the team you’ve been looking for. #TeamCulture #RemoteTeam #SoftwareProduct #OptavynCulture #ExecutionMatters #TechThatWorks

4

I had 24 hours to prepare a Microsoft Copilot training I didn't see coming. Monday started normal. Then Ashley Boyd from Collin College sent an urgent request: "We need someone to teach Copilot training at a utility company tomorrow. Are you available?" The company was CoServ - a North Texas energy cooperative with 600 employees beginning their AI journey. My answer: Why not? THE CHALLENGE Microsoft Copilot isn't one product. It's several. The free version, individual paid license, and enterprise version all have different interfaces and very different capabilities. Most large companies have an inner circle with full paid licenses. Everyone else gets a version that produces noticeably weaker outputs than ChatGPT or other AI models. How do you show meaningful use cases with those limitations? On top of that - I couldn't use my own computer. Slides only. No stakeholder interviews. No prep time. Except one call - with Michelle Davis-Perkins, CoServ's Director of Training. That call changed everything. WHAT WE PLANNED We came up a great idea: open the session by asking every person one question. "What's one task you spend too much time on that you'd love to hand off?" Then use THEIR answers to drive the entire hands-on portion. No generic demos. Real work, from real people, tested live in their actual Copilot environment. WHAT HAPPENED Volunteers shared use cases right there in the room. Together we tested Copilot Agents (their version of custom GPTs) and focused on two things that work regardless of version: → Better data inputs → Better prompting methodology Because the quality of the output always depends on the quality of the input - no matter the tool. WHY IT MATTERS When someone asks you to serve in a pinch, the reward goes both ways. They get value on 24 hours' notice. You build trust the way it actually works - by showing up when it matters. I also appreciated that IT was in the room. We had an honest conversation most companies avoid: Employees need personal AI accounts too. Company Copilot handles company work. Personal AI — ChatGPT, Claude, Gemini — builds skills that make you more valuable everywhere. Both can coexist. Smart companies respect the policy AND encourage the growth. Walking out, I could see the opportunity clearly. Deep Research. Better prompting. Data analysis. Workflow automation. These employees are using maybe 10% of what's available. That's not a Copilot problem. That's a training problem. The tool is only as powerful as the person using it. Have you been asked to do something on short notice that turned into something bigger? I'd love to hear your story. P.S. Not sure where AI fits into YOUR business? Take the free AI Readiness Assessment from the AI Academy 4 All - link below. David Gold

33

13 Comments

This one might hurt a little. 👀 I’m being honest. 🫡 I’ve worked in teams that said, “We’ll add tests later.” Later never came. —————- ❌No TDD Feature starts fast. PR looks big and impressive. Everyone claps. Then… One small change breaks three modules. QA finds regression. Hotfix goes out. Another thing breaks. Now the same feature costs 3x the effort. Speed at the start. Stress at the end. —————— ✔️With TDD Start with a failing test. Write just enough code to pass. Refactor safely. Repeat. It feels slower. It feels uncomfortable. It forces you to think. But here’s what I’ve seen: Fewer surprises. Cleaner design. Refactoring without fear. Confidence during releases. —————— TDD isn’t about loving tests. It’s about respecting the future version of yourself who will maintain this code. From my QA lens, the difference is clear: Without TDD → QA becomes a damage controller. With TDD → QA becomes a quality partner. And that shift changes everything. Agreed? Learn With Psudo #TestDrivenDevelopment #TDD #SoftwareDevelopment #QualityEngineering #ShiftLeft #AutomationTesting #Agile #DevOps #CleanCode #ContinuousIntegration #SoftwareTesting #TechLeadership

3

I suspect it's because many PMs have existed in a world where they're perceived as being in charge, hierarchically, and they think that means they get authority over what the developers work on, when and *how*. Instead, they are generally only qualified to say what order product capabilities are done in, and why... not when and how. The teams get to say how and when it's possible based on that info, taking account of the strategic ability to keep evolving the product. All is open to negotiation, based on the "why". If a PM can show the business criticality of a deadline, then the team can agree to doing things differently if shortcuts or deferring maintenance can be commercially justified with evidenced arguments (not just arbitrary demands on dates). But, they may also push back when the consequences of shortcuts or deferred maintenance are worse for the product and the company than the benefits of any "debt". A good Engineering Manager sets this boundary, and defends the teams' autonomy on the when & how. This can feel like a reduction in status to the PM, from "above" the team to a peer in the team. This can promote defensive conflict, resulting in a cycle of mistrust. Setting and enforcing these boundaries is where I've experienced conflict as an EM, and observed as a coach. Others' experiences may vary.

9

Talked to a founder on Monday that swore codex was better than claude. They had 11 MCPs configured in claude that chewed up over 50% of the context window. Their codex setup had 0 MCPs. Their claude config was like telling someone about railroads, Japan, and Apple then asking them a calculus problem.

7

2 Comments

Is open source a "silver bullet"? Fred Brooks claimed we wouldn't see anymore of those, but luckily that was one of the few things he turned out to be wrong about. Check out JUXTCast podcast where Henry Garner and I talk about this, the mission of @FINOS and the reason why open source is hard for financial services firms to take maximum advantage of. Henry's LinkedIn Post: https://lnkd.in/en3vi8NU Check out JUXTCast here: https://lnkd.in/e9CjMJrD

4

1 Comment

It seems crazy to me that people are just letting AI program entire software ecosystems in one shot with virtually no guard rails. Very little planning, if any. No task list. No standards. No testing criteria. No DoD. No wireframing. Using .md files like openclawd does has been a failure in my testing. It creates two sources of truth and the AI gets lazy and doesn’t maintain the md. Good old standard project management and project planning with MCP as the control layer has yielded the best results. With shorter testible iterations. MCP as the gatekeeper to force the AI to not skip the paperwork. Then pile on quality checks. Every check you can think of and really good logs and git to troubleshoot and roll back if it goes crazy.

11

9 Comments