Travis Hummel

I’m always incredibly proud of our engineering team here at Opal, but am truly excited about what we've recently accomplished. In recent months, we've been working on two areas of the product that are deeply impactful for our customers: Plans and Gem AI. Plans is such an important part of how teams in Opal organize major initiatives, stay aligned, and bring strategy to life. Seeing our team continue to strengthen that foundation has been exciting in its own right. On top of that solid foundation, with Gem AI, we've built something that is not just innovative, but genuinely useful to solve real problems that our customers face every day. This video from Opal Co-Founder @David Gorman shows both of these capabilities truly shine when brought together!  

38

4 Comments

Sau Sheong's recent piece(https://lnkd.in/gTbfKw3i) spoke about supervisory engineering being the core skill in agentic engineering. In it he showcased how it empowered GovTech teams building things nobody thought a small team could build. It got me reflecting on my own experience building agent skills, and I penned a closer angle on the same idea. The agent does the execution. You design the loop it runs on. That loop is where a new kind of engineering work lives now. https://lnkd.in/gPK2G6ru

5

1 Comment

Excited to share my first post on the Airbnb Tech Blog. I'm really proud of the work our team has been doing on observability, and this post digs into some of it. We faced a major migration of 300,000 alerts to Prometheus alongside growing alert noise. The noise looked like a culture problem. It wasn't. The post covers how we rethought our Observability as Code platform, introducing Change Reports and bulk backtesting to let engineers validate alert behavior at scale before deployment. Development cycles collapsed from weeks to minutes. Alert noise dropped 90%. We completed the migration. And the culture shifted. https://lnkd.in/eixNKqJG

89

3 Comments

"Just keep doing what you're doing." So many engineers have talked to me about receiving this feedback from their manager. In many cases, the question they asked that led to this feedback was a very generic, "Hey, do you have any feedback for me?" How do we get better feedback? Here are two ways to get better feedback from your manager: 1. 🎯 Give a goal, a timeline, and ask about what you should change. "Hey Julia, I want to get promoted to the next level in the next 6 months, what do I need to change about what I'm doing today to achieve this?" Even if the question is still open-ended, this works better since it's more clear about what you want to achieve and you frame a time constraint. 2. 📅 Ask for specific feedback from a specific event. "Hey John, do you have any feedback on the clarity of my communication from the Tuesday status meeting?" This works because it's specific so your manager can give you more targeted feedback. What's your best tip for getting better feedback from your manager? Drop it down below 👇 #growth #feedback #bespecific

24

10 Comments

If you’re not actively experimenting with AI in your real work yet, this is a good nudge to start. Re-sharing this essay that Amit Fulay posted: https://lnkd.in/gVdYKvkG It’s one of the clearest "zoom out" takes I’ve read on how fast AI capability is compounding and why the perception gap is becoming risky. Curious: what’s one workflow you’ve already changed because of AI?

3

1 Comment

I find it ironic when companies who want RTO for "close collaboration" also want to hire overseas workers. Spinning up a whole team is one thing for which there are some benefits beyond financials (such as "follow the sun" support schedules.) However, if it is adding to an existing team, suddenly you have remote and asynchronous team members. Which can be worse for them because they are not privy to all of the "close collaboration" that happens amongst the team that are all co-located. So the only obvious solution once you start adding team members who are half the world away is to have everyone WFH. (Note: there is a little sarcasm in the "the only obvious" quantifier.)

20

2 Comments

I don’t know how AI will impact the software development job market over the next few years, but it sure has turbocharged the hot takes market.

54

4 Comments

What a fascinating time to be a builder. In my current role, I’m fortunate to see the Software Development Life Cycle across hundreds of teams as they work with Checksum.ai for CI/CD-ready, full end-to-end AI testing. 2026 is going to be a wild! About a year ago, most customers were looking for solutions to augment their QA automation teams as coding velocity was increasing significantly. About nine months ago, customers started asking for solutions that directly augmented their software developers, enabling them to write CI/CD ready end-to-end tests. Recently, customers are asking for solutions that augment their army of coding agents autonomously: AI coding agents working in parallel with AI quality agents as they build features and fix bugs, with developers signing off after quality check has passed. This… is… exciting... 🚀

29

I recently attended APIDays Melbourne and it’s one of the first conferences that truly feels current with today’s AI transformation. Here are my key takeaways: - APIs and AI are converging faster than our security models can keep up. - APIs have always been deterministic: predictable, testable, controlled. - AI systems are probabilistic: adaptive, contextual, and ambiguous. - The challenge now is designing architectures that combine both structured APIs that AI agents can safely and intelligently interact with. - As AI agents start invoking APIs autonomously, standards like Model Context Protocol (MCP) and Agent2Agent (A2A) are emerging. They bridge reasoning (AI) with execution (APIs), embedding observability and compliance into agent workflows. - The next evolution is Adaptive API Security fingerprinting devices, establishing persistent identities, and rerouting malicious traffic instead of blocking it. Static rules are no longer enough; intelligent systems must learn and respond in real time. -APIs are becoming the nervous system of AI ecosystems. - Security, reliability, and intelligence are now continuous, adaptive, and inseparable.

19

Luckily, I was NOT impacted by the recent layoffs at Atlassian. However, some colleagues I’ve interacted with in the past were not as fortunate. While this news was concerning, the reason I am posting today is different. Ever since the news broke, I have seen many people on LinkedIn fear-mongering—claiming that these jobs were cut because they became redundant due to AI, and that if you don't "learn AI now," it will be too late. Honestly, I strongly disagree with this sentiment. We are living in a world of rapid technological and geopolitical change, and companies are adapting to these shifts in their own ways. While this is sad news, I am confident that Atlassian, and other companies, will start hiring again based on their long-term growth strategies. I know for a fact that AI, even today, cannot perform even 10% of what the impacted individuals were doing with any level of consistency. So, while I believe it's important to keep up with change and never stop learning, I do not subscribe to this narrative of fear. To those impacted: your skills are valuable, and your experience goes far beyond what any tool can replicate. #Atlassian #Layoffs #AI https://lnkd.in/dZN2ytsF

50

Happy to be speaking with Susana de Sousa about how the landscape have shifted in the world of Support, and what hiring looks like in this new AI era. Sign up to join the conversation!

69

6 Comments

This is the best part IMO: TLDR: even with agents this is really hard stuff. “There are a whole bunch of reasons I’m not scared that my career as a software engineer is over now that computers can write their own code, partly because these things are amplifiers of existing experience. If you know what you’re doing, you can run so much faster with them. [...] I’m constantly reminded as I work with these tools how hard the thing that we do is. Producing software is a ferociously difficult thing to do. And you could give me all of the AI tools in the world and what we’re trying to achieve here is still really difficult. [...]”

3

Following on from building Estii — still one of the most polished and thoughtfully designed SaaS products I’ve seen come out of a two-person, self-funded team — my brother and Dominic are now getting ready to launch OpenSurf. Fraternal bias (and immense pride) aside, the potential here is enormous. OpenSurf tackles a problem many of us have experienced firsthand: how brittle and unreliable “automation on the web” tends to be, especially once things get even slightly complex. At a practical level, think of OpenSurf as something akin to an instruction manual for websites — providing structure, intent, and guardrails so AI can actually understand what it’s interacting with, rather than blindly clicking and hoping for the best. And as someone who regularly provides parental tech support on using online tools, the possibilities in that space alone are pretty exciting. This quick demo is well worth a watch.

20

2 Comments

You need to check out this demo! 🚀 🛑 Stop typing out bug reports. Just talk through them to ship it! My team has spent the last 9 months heads-down building this, and it perfectly showcases the incredible power of structured, AI-powered workflows. We’ve brought the best pieces of the Atlassian ecosystem together - Loom, Jira, and Rovo - to deliver a massive productivity boost. Instead of endless chat threads to reproduce a bug, you just hit record. Loom captures the context, Jira builds the dev-ready ticket, and Rovo Dev opens the PR - all with a simple screen record! So incredibly proud of what the team has accomplished here. Watch the magic happen below! 👇

10

What is the next frontier of engineering software? This excellent article from Maxwell Dylla, one of our founding engineers at Cosmon breaks it down. https://lnkd.in/gDKqjBHU My top 3 takeaways were: 1. Engineers are stuck on "workflow trivia" instead of deep thinking. Brilliant minds waste time wrestling with software, installs, and handoffs, rather than on the actual problem-solving they were trained for. We've confused operating software with doing engineering. 2. Engineering knowledge is fragmented and fails to accumulate. Insights are lost in local folders, personal scripts, or between projects. As a result, organizations repeatedly "pay twice" for the same understanding and fail to build on past lessons. 3. The future is not better tools, but systems for "compounding knowledge." The next frontier is creating an environment where every analysis and test builds a "living body of understanding," allowing the entire organization to grow wiser with each project. The point about "workflow trivia" really resonated with me, having wrestled with bad software for too many years. A must-read for anyone in the tech and engineering space. What do you think will be the biggest change in engineering in the next 5 years? #Engineering #FutureOfWork #AI #Leadership

19

As someone with ties to the Australian tech scene and an ex-Atlassian employee, I’m naturally deeply shocked by the recent layoffs. This will have a significant impact on the industry in the coming months. It’s not the time to “praise Atlassian for their sincerity” or criticise their work-from-anywhere policy. This is a time for empathy and to remember that such things can happen to anyone at any time. This means finding a balance between being prepared and not letting anxiety consume us. Stay strong everyone. I’m sure brighter days are ahead for all of us.

65

Recently we shared our AI Engineering Maturity Framework. We have been using this internally for a year now to track the industry and guide the evolution of AI Engineering for our 8000 engineers. David Hong has now posted an article which speaks to our Level 3 agent orchestration framework, Project Coral. This framework allows engineers to offload multiple types of 'toil' tasks, so they can focus on direct customer value. Don't forget to follow the CommBank blog, we'll be sharing more frequently. https://lnkd.in/g6Sv5c7U https://lnkd.in/gciAvhTh Gavin Munroe Rodrigo Castillo Brendan Hopper Martha McKeen, GAICD

43

"It's 2025 and we're all adding AI features to our apps. But the tech moves so fast - what solid ground can you actually build on? This talk will focus on one of the best established patterns: building chat UIs that accept natural-language instructions. This includes: Getting started with Microsoft. Extensions.AI and Aspire (without needing any paid service) Ingesting, indexing, citing, and visualizing your custom business data sources for retrieval-augmented generation (RAG) Teaching AI how to retrieve data from your APIs and invoke your app's actions on behalf of a user Consuming and building Model-Context Protocol (MCP) servers for easy interop across AI systems A peek into potential future tooling that simplifies the development process This is a fast-paced deep dive into how to do it - it's coding demos all the way through! By the end you'll hopefully have a strong sense of which parts are relevant to your scenarios and how to wire them together into something genuinely useful for your users." #ndcmelbourne #ndcconferences #developer #ai #dotnet #web #api

1