Derek Croft

Engineers prefer specialised tooling for tests, docs, and code reviews -> https://lnkd.in/drC-sgcn That preference creates a tension. Role specific tools help, but separate tools can break alignment. A “done” task means something different to an engineer than it does to a researcher, but both still need to land on the same common ground. So AI cannot only work on the last prompt. It must operate over shared state: intent, insights, and feedback signals, with role specific behaviour anchored to a unified truth underneath. This is why we are building Qi, a flow engine for human AI cooperation, to move upstream and anchor AI reasoning in authenticated shared state, provenance and trust signals, with humans able to challenge and refine outputs. -> Swipe to slide 5 for the specialisation versus alignment tension. Slide 6 lists the structural fix.

5

1 Comment

Staff Consultant Brian Hughes breaks down the myth that software tests slow down developers. If you're doing it right, software tests can actually speed you up.

7

1 Comment

Here's something that surprised us. We ran a build where every time the coding agent said it was done, we had a verification agent actually check the files against the spec. Every. Single. Time. — about 30–40% of the work wasn't there yet. Not broken code. Just features the agent never built, with no idea it had missed them. A payment flow that led to a blank screen. Data hooks still returning mock data. Auth guards missing on three routes you could navigate to without logging in. None of it was visible without checking the files against the plan. We wrote up the full story — the numbers, the findings, and what it would have cost if any of it shipped. Worth a read if your team is running coding agents 👇 #AICoding #CodingAgents #SoftwareDevelopment #AITools #BrunelAgent

1

Every .NET app makes HTTP requests. But not every app handles them well. If you’re repeating auth headers, retry logic, or logging across services - there’s a better pattern. 👉 Enter Delegate Handlers - a clean way to manage cross-cutting concerns in your HTTP pipeline. If you care about clean architecture and scalable APIs, this one’s worth a swipe. #dotnet #softwarearchitecture #Userware #appdevelopment

7

👨‍💻 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

The experience paradox is real: "We need React experience" but how do you get experience without...experience? The answer: Demonstrate active learning and showcase what you can build right now. This approach helps you bridge the gap between current capabilities and role requirements — and makes you stand out to hiring managers. Your complete guide to sharing skill progress professionally is here, with examples of how you can use our new skills tracking feature to showcase your progress. 👉 https://bit.ly/4mMiSkc

42

Coding agents didn’t remove the bottleneck. They moved it. The teams actually improving delivery aren’t stopping at coding agents. They’re deploying agents across the entire SDLC. We broke down that shift here: background-agents.com Writing code faster doesn’t fix delivery; it increases pressure downstream. More PRs lead to longer reviews. Faster reviews create testing backlogs. Accelerate testing, and deploys begin to queue. The constraint doesn’t disappear, it just relocates. Optimizing each stage individually doesn’t change system throughput. What does change it is restructuring the system. High-performing teams are running agents across the whole pipeline. Dependency updates run on a schedule and open upgrade PRs automatically across every repo. Incident triage agents gather logs, correlate recent deployments, and post root cause summaries before on-call even begins investigating. Routine maintenance and cross-repo changes happen continuously in the background instead of waiting for someone to initiate them. The goal isn’t faster code generation. It’s faster delivery. That requires treating the SDLC as a system and deploying agents accordingly.

11