David Lewis

Paying a CPO $350k a year to groom a backlog is wild, but it happens. As is the scenario when a company is on its 3rd CPO in the last 12 months! There's this assumption that the 1st Product hire must be a Product Leader like a Head of Product or CPO. But in many cases, it makes zero sense to do so. In some cases, it makes sense to hire a Product Leader who can operate in a "Player/Coach" type capacity, especially if the organization is primed for rapid growth and the founding team lacks Product Management expertise. In other scenarios, it may make more sense to hire a Product Manager who will operate as an IC, and then hire a Product Leader when the business is ready. The wrong level of hire or the wrong type of hire is costly for the business and the PM. #hiring #recruitment #productmanagement

286

38 Comments

Your product isn't failing because of bad handoffs it's failing because one specific relationship is broken. Stop fixing your handoffs. The problem isn't all four. It's one relationship between PO and Developer. Watch four people dance to Tauba Tauba. Customer nails it. PO gets it mostly right. Developer struggles. QA delivers something barely recognizable. The dance was perfect when customer showed it. By production it's unrecognizable. Every product team has lived this. But the degradation doesn't happen equally at every handoff. Customer to PO - some loss but the intent survives. PO to Developer - this is where the dance dies. Not because requirements are bad. Because the relationship ends when the tickets start. Developer to QA - testing what was built not what was needed. QA to Production - shipping what passed specs. Most teams try to fix this with better documentation. More detailed tickets. Clearer acceptance criteria. All of that assumes the problem is information transfer. The problem is relationship breakdown. PO writes requirements. Hands them to Developer. Walks away. Assumes the tickets contain everything needed. Developer reads tickets. Has questions about edge cases. Makes assumptions. Implements based on interpretation. Neither talks to the other during implementation because the handoff happened. The relationship ended. The dance was already dying. "User needs instant settlement" becomes "add API with 2 second timeout." Technically matches the ticket. Completely misses why instant matters or what the user actually needs. This isn't a PO problem. This isn't a Developer problem. This is a relationship problem. The handoff mentality treats product development like relay race. PO runs their leg, hands baton to Developer, job done. But product development isn't relay race. It's partner dance. Developer hits technical constraint - needs PO context to make tradeoff decision. But relationship already ended at handoff. Implementation reveals assumption in requirements - needs PO course correction. But PO assumes their job finished when tickets shipped. Edge case appears - needs customer context to decide handling. But neither PO nor Developer feels responsible for maintaining connection. The relationship that matters most is the one that ends earliest. PO and Developer need to stay in conversation through implementation. Not PO dictating requirements. Not Developer working in isolation. Continuous collaboration where context flows both directions. Tradeoffs decided together based on user outcomes. Assumptions surfaced and corrected in real-time. This doesn't mean PO hovering over Developer. It means both treating implementation as shared responsibility not sequential handoff. Start building relationships that survive past the handoff moment. The dance dies when partners stop communicating. Keep the conversation going and the dance survives. What kills the PO Developer relationship in your org?

247

76 Comments

At Testimonial Donut we've been talking to Australian mortgage brokers. Here's what we found: - They use CRMs - They email clients after settlement - They ask for Google reviews - Response rates are steady but not high The problem isn't the ask. It's the follow-up. One or two emails might get ignored. But a smart sequence? Email. SMS. Email. SMS. 4-5 soft custom reminders get more reviews. Mix it up. Stay human. Soft reminders. We built Testimonial Donut to slot into their existing flow. Not replace it. Not complicate it. Just make it work better. Webhook triggers. Zapier connects. Automation follows up. Once they leave a review? Reminders stop. Set and forget. The broker does what they've always done. We handle the follow-up. Follow up. Remind gently. Get moire Google reviews than before. Here's what I've learned: The magic isn't in the big moves. It's in the small systems that compound. Your clients want to give you that review. They just need to be reminded. Set it up once. Never think about it again. We're invisible in your workflow. ------------------------ 👋 I'm Guy, building Testimonial Donut with Farzan M.. We're helping businesses collect more testimonials & Google reviews. Follow along as we share what works (and what doesn't).

12

2 Comments

Kia ora network. If you are wondering what the rise of AI actually means for software engineering, this recent piece from O'Reilly is well worth a read. The guts of it is that software engineering day jobs are shifting. Writing code from scratch is no longer the main event. Instead, developers are stepping up to manage and review AI generated code. AI is a fantastic tool for churning out boilerplate, but it can still be a bit of a loose unit when it comes to accuracy and making things up. Here are the main takeaways from the article. First, reviewing and validating code is now your new baseline. Instead of writing every line, your day-to-day job involves catching the hallucinations and making sure the output is actually solid before it gets anywhere near production. Second, because the grunt work is handled, the big picture is more critical than ever. We need to focus on system architecture, security, and solving real business problems rather than just memorising syntax. Finally, while having almost anyone prompt an AI to build a quick prototype is fun, real production systems still require the hard yards. Proper testing, scaling, and robust engineering principles are what stop the whole thing from falling over. At the end of the day, the fundamental skills still pay the bills. The tools have changed, but good design and clear requirements are what keep projects on track. Link to the full article is below. Keen to hear what developers think. Are you finding your role shifting more towards code review and architecture these days? https://lnkd.in/etrWuXf8 Next week: I'll talk about commercialisation, we are seeing the startup landscape balloon with iterations of the same solutions, the winners will be those who are market led and who have commercialised beyond "check out my cool thing".

40

6 Comments

Product is the only function where engineering AND business both think you work for them. Engineering thinks you're their advocate to the business. The business thinks you're their translator to engineering. In truth, they're both right. And both wrong. I've watched brilliant product leaders navigate this impossible middle ground for years. The best ones don't pick a side. They learn to play both sides. → They speak both languages fluently but translate sparingly → They push back on both sides equally (and get comfortable with the discomfort) → They make the customer the tiebreaker, not politics The tension becomes key to the role. Product managers who try to eliminate it end up as order-takers. The ones who lean into it become strategic leaders. Because the real job isn't keeping everyone happy. It's keeping everyone honest about what actually matters. When engineering says "that's technically impossible," you ask "or just really hard?" When business says "we need this yesterday," you ask "or do we need to validate the assumption first?" The middle ground isn't comfortable. But it's where the best decisions get made. What's your opinion? Is this tension productive or just exhausting?

143

76 Comments

At the risk of sounding arrogant, I'm going to write this post bluntly because the point is that important. I've spent about 8 years building software engineering teams in startups. I'd argue I'm one of the best in Australia at it - hiring software engineers and assembling fantastic startup engineering teams is something that I really excel at. And yet, even from that starting position, over the last 6 months I've been relearning - almost from the ground up - what a good software engineer is. That will continue for at least the next year. If you haven't meaningfully changed how you assess - or even define - what a good software engineer is in the past 6 months, your definitions and assessments are outdated. Startups are in pole position to adapt and carve out an advantage. But that also means they'll be punished most for failing to adapt, and you need to be adapting now. This is only half the problem though. When you do hire engineers who are excellent by the new definition, they require a company that's ready and set up to receive them - and most aren't. That means product, design, and the entire roadmap and the speed in which it's executed changing to accommodate what's now possible in engineering teams. It's a business-level shift, affecting how the entire company operates, not just a change to your hiring process. This has knock-on effects to funding, growth expectations, salaries, metrics, and more. Everything is changing, and it's changing now. It will continue to change, probably even more rapidly. But this isn't about predictions. It's already happening. First movers realised this 12 months ago. I realised it 6 months ago. The vast majority are yet to realise it. You should be realising it by the end of this post.

39

21 Comments

At Wednesday's Xero TechTalks, amazing AI specialist Marek Poliks from LaunchDarkly highlighted how runtime control can improve the secure and flexible deployment of AI-powered features. He addressed common deployment challenges such as prompt updates, model switching, evaluations, and implementing guardrails. Marek underscored how LaunchDarkly's AI Configs platform facilitates safer experimentation, instant rollbacks, and configuration-driven releases.

39

Everyone made jokes about Atlassian acquiring the Browser Company... But it’s actually genius. It's part of a larger acquisition strategy all about AI. CONTEXT Atlassian has been v successful with acquisitions like Trello & Loom. They see things others don't. I loved Brian Balfour's Arc analysis: https://lnkd.in/era8Xn3y Since then, Atlassian also acquired DX and Cycle. And 2 years ago, they made the huge acquisition of Loom. So let's look at it from that larger angle: 1. LOOM ($975M, 2023)     This was their first big splashy acquisition. And it made sense - async video was becoming critical for remote work. But the deeper play was about AI. Loom's videos get automatically transcribed, summarized, and turned into searchable docs. Loom + AI makes every standup, demo, and feedback session becomes structured data that feeds back into product decisions. 2. BROWSER COMPANY ($610M, 2025) Everyone thought this was weird. But think about it - where does all work actually happen? The browser. Arc and Dia aren't just browsers, they're AI-powered workspaces with intelligent tab management, workflow organization context switching. Atlassian's bet is AI will fragment browsers into B2B vs B2C. Most desktop browsing is already work-focused, but Chrome was built for consumer web. Atlassian is positioning for the AI-native work environment layer. 3. CYCLE (Undisc, 2025) This one's sneaky smart: https://lnkd.in/eYtYQ7Ut Customer feedback is complete chaos - scattered across Slack channels, support tickets, sales calls, and user interviews. Cycle uses AI to turn that into actionable signals, then feeds it directly into Jira Product Discovery. No more guessing what customers want or synthesizing feedback from 15 sources. 4. DX ($1B, 2025) The biggest acquisition. Developer productivity analytics powered by AI insights. As companies dump millions into AI coding tools like Copilot, they desperately need to know: is this actually making us ship faster? DX measures exactly that. #AI #AtlassianPartner #ProductStrategy TAKEAWAYS These 4 acquisitions show how Atlassian is trying to rebuild how knowledge work actually happens. They are building the entire stack for AI-native work. Each acquisition amplifies the others: Loom and Cycle give you insight → Arc/Dia help you orchestrate → DX measures. It's a genius play. In 2025, everyone needs to have an AI product strategy: https://lnkd.in/egemMhMF And Atlassian is taking that seriously. Build your own: https://lnkd.in/e8EUQHAW P.S. If you enjoyed this, you'll love my podcast with Brian: https://lnkd.in/e9aNsCbs

399

40 Comments

🚨 Big news: Yump has officially joined forces with Bright Labs! We’re now part of one of Australasia’s largest Craft CMS agencies, combining our award-winning UX and accessibility expertise with Bright Labs’ full-service capabilities. Together, we’ll continue delivering exceptional websites and performance-driven digital marketing solutions for not-for-profit organisations — now with even more resources, services, and industry reach. Learn more here 👉 https://lnkd.in/gCnguvzH #YumpDigital #BrightLabs #CraftCMS #UXDesign #WebDevelopment #DigitalMarketing #NFP #Accessibility #AgencyLife

65

8 Comments

We're increasingly hearing from our Australian customers that they care deeply about data protection and want their data stored locally too. We’re delighted to announce local data residency for Figma in Australia and excited to have National Australia Bank (NAB) as one of the first customers to be a part of this initiative. In relation, we’re rolling out new governance features for our enterprise customers. If security in product design and development is important to you too, check out our blog for more information. Special thanks to Lance Thornswood. https://lnkd.in/enePHUjR

91

6 Comments

Most people use LinkedIn search wrong. Here’s how to shortcut to the actual decision-makers, without sending 500 connection requests into the void. 1. Start with the right title pattern For hiring decisions in ANZ tech: “Head of Engineering” “CTO” “Delivery Manager” (for project-heavy orgs) “Talent Acquisition” (if targeting internal recruiters) Avoid single words like “Manager”, its too broad. Use "Engineering Manager" instead . 2. Use the ‘Current Company’ filter If you know your target employer, this removes ex-employees from results. Bonus: Combine with ‘Location’ = “Australia” or “New Zealand” to strip out overseas HQ noise. 3. Layer the ‘Connections of’ filter Find people already connected to someone in your network. Your mutual can warm up the intro, far better than cold messaging. But don't forget to check if you have any 1st degree connections. One client just booked a interview at Qantas with this tip alone last week. 4. Check the ‘Posted on LinkedIn’ filter Decision-makers who post are far likelier to reply. Comment on their recent post before sending a DM, builds context fast and you are going in their digital proximity. In the our tech market, speed to the right conversation is what wins. These filters let you bypass endless scrolling and start talking to the people who can actually say “yes.” Caveat: Not everyone will accept your connection and that’s fine. The key is to keep doing this consistently, and make your approach specific and personal. Please do me a favour and don’t blast the same copy-paste message to 50 people. Think about what kind of note would make you want to reply, then write that.

20

5 Comments

Was curious about how much ShopDev Alliance has grown, not just internally but some of the benefits that we've chosen to exploit being a Slack workspace. We've got just over 30 channels in SDA and the cool thing is that we have almost DOUBLE Connect channels than internal channels - For those that don't know what Connect channels are in Slack, these are external connections to our workspace. This is so developers in our space can create connections with other workspaces - like other app devs, merchants, or service providers. For many, this means they were able to get rid of their own personal Slack they were using (I know I did). Just really cool to see it's not just about internal connections, though that is the biggest appeal of the Slack by far, but it's still helping developers connect to all the other spaces.

22

TOP PROMISING STARTUPS IN ANZ (RECON LIST) Looking to join a rocket ship before it takes off? ANZ doesn't produce many unicorns, but when we do, they tend to be exceptional: Canva, Atlassian, SafetyCulture, Xero, RocketLab, Halter, Crimson Education. Here are 27 of the most promising startups in Australia/New Zealand: Relevance AI – AI-native platform for building agents and workflows (AUS)​ Instant – One‑click checkout and retention for ecommerce brands (AUS)​ Zepto – Real‑time payments infra for account‑to‑account transactions (AUS)​ Neara – 3D digital‑modelling software helping utilities map (AUS)​ Myriota – Satellite IoT connectivity for ultra‑low‑power remote sensors (AUS)​ Minikai – AI software layer to streamline disability and workflows (AUS)​ Shopfront – AI‑driven re‑commerce infrastructure for listings (AUS)​ Kardinia Energy – Deep‑tech developer of lightweight, printable solar film (AUS)​ GOTERRA – Automated insect systems turning food waste into function (AUS)​ VXT – AI VoIP Phone System for Law Firms (NZ) Harrison.ai – Health‑AI platform providing decision tools with hospitals (AUS)​ Sendle – Carbon‑neutral parcel‑delivery service for ecommerce merchants Linktree – Link‑in‑bio infrastructure for creators and brands (AUS)​ Go1 – Corporate learning platform aggregating online training content (AUS)​ Partly – Auto‑parts data and ecommerce infrastructure (NZ)​ LawVu – Legal‑operations software platform for in‑house legal teams (NZ)​ SurfLab – AI‑powered personalised surfing‑coaching platform (ANZ)​ VetNotes.com – AI scribe and workflow assistant for veterinary practices (AUS)​ Spacium App – AI agents to help manage household and startup tasks (ANZ)​ AreaSpec – Platform matching workforce rosters with childcare capacity (ANZ) Zipline AI – Platform to build AI and ML‑powered products in days (AUS)​ Chattrly – AI patient‑intelligence for insights from electronic health records EngageNow – B2B marketplace using AI to match companies with services Tracksuit – Brand tracking and performance analytics platform (NZ) Auror – Retail crime intelligence and prevention platform (NZ) Let me know if I’m missing any below!

820

61 Comments