Zheng Liu

2025 DESIGN IN TECH REPORT SXSW TAKEAWAYS https://lnkd.in/gsp5SYTH - AI is not replacing designers but transforming how design is done. - AI experimentation is becoming significantly cheaper and faster. - The “Agent Era” is shifting AI from models to task-completing agents. - UX is evolving into AX, reducing UI in favor of direct AI execution. - AI UX best practices are emerging to improve trust and usability. - AI automation introduces risks that require responsible governance. - Human adaptability is key to thriving in an AI-augmented future. --- 2025 #DesignInTech Report super-long version, unedited version goes live on Tuesday: 📺 https://lnkd.in/gsp5SYTH This year's SXSW was a special surprise as I had the chance to meet my all-time hero ... Cookie Monster! This year's report: https://lnkd.in/gsp5SYTH Past 10 reports: https://lnkd.in/eq5kFQY

391

11 Comments

Waymo for Delivery Turns out the Waymo of delivery is --- DoorDash! DoorDash's Dot delivery bots are a BIG deal. They have an autonomy stack with sensors, hardware, and software similar to that of robotaxis, no extra fees to merchants or consumers (vertical integration is key for this IMO, standalone robot companies without demand cannot survive), are capable of going down busy streets. get better every single day with reinforcement learning from millions of deliveries, and have a much better form factor (not to mention much lower cost) than Waymo for delivery. Transporting humans is an expensive proposition and the robots should be built with this in mind; transporting your food and groceries should be designed from scratch with very different design principles. (eg: how does the vehicle behave in a crash? in the case of Waymo, it must protect its passengers. For autonomous delivery bots, they should sacrifice their food :)) This could very well be the Waymo moment for delivery. Congrats Stanley Tang and team! Can't wait to try it out soon 🙏 Link: https://lnkd.in/gECNPEQ9

439

10 Comments

Today marks an exciting step forward in the future of AI-powered commerce. Instacart is now the first grocery partner to launch an app directly inside #ChatGPT, and the first to enable embedded, end-to-end shopping and #InstantCheckout within a conversational interface. This integration is built on the Agentic Commerce Protocol and leverages Instacart’s real‑time retail graph, our dynamic product catalog of 2B+ product instances, and our deep retailer integrations. Together, these enable ChatGPT to translate natural language intent into accurate, in-stock, real-world grocery decisions. A parent can start with a simple question like “I need quick dinner ideas for a busy weeknight.” Within seconds, ChatGPT can suggest a recipe, pull in the exact in‑stock ingredients from their preferred local store, factor in dietary preferences like nut-free or gluten-free, and create a ready-to-review cart. With Instant Checkout built directly into the conversation, they can complete the order securely,without switching apps. A few highlights: Real-time inventory intelligence: Our platform continuously models local availability and pricing across nearly 100,000 stores. When ChatGPT recommends an item, it’s grounded in what’s actually on the shelf — not theoretical data. Personalized, detail-aware shopping: Instacart’s decade of preference and fulfillment signals (brand nuances, sizing, dietary needs, substitution behavior) helps the agent tailor recommendations that reflect how people truly shop. Instant Checkout built natively into the conversation: Using the Agentic Commerce Protocol and partnering with Stripe, we enabled secure, familiar payment flows inside ChatGPT for the first time. It’s a seamless bridge from language to action. Scalable architecture for agentic shopping: Grocery is a high‑dimensional, high‑variance domain. Our fulfillment intelligence, catalog systems, and retailer integrations make it possible for AI agents to operate reliably at scale — even as inventory and consumer preferences shift minute to minute. This launch is more than a new interface for ordering groceries. It’s a preview of how AI agents will connect digital intent to physical fulfillment across the economy. Instacart’s real-time grocery engine is uniquely positioned to power these experiences, and we’re investing deeply to make shopping more predictive, connected, and effortless. Huge thanks to the teams across Instacart and OpenAI who made this possible. The next chapter of agentic commerce is just getting started. https://lnkd.in/gtN_XcZY

709

34 Comments

↗️ Launching Spur Environments!! Run any test. Any storefront. Any device. All at once. Retail engineering teams know the pain. - The promo banner works in staging — but breaks in prod. - Checkout looks perfect on desktop — but overlaps on mobile. - Data differences between environments make test results unreliable. You spend hours duplicating the same tests for every site, every region, every viewport — just to make sure customers get a consistent experience. But what if one test could handle it all? Now it can. With Spur Environments, you can run a single test across every environment, device, and configuration — instantly. Here’s how it works 👇 ① One Test, Infinite Contexts Write a test once. Spur automatically adapts it across staging, prod, and regional storefronts — accounting for environment data, config, and viewport differences. ② Eliminate Environment-Specific Maintenance Stop cloning and editing tests for each environment. Update once, and every storefront stays in sync. ③ Consistent CX, Everywhere From checkout to product pages, Spur ensures every test run stays stable across devices, browsers, and geographies — so you can trust your customer experience. ④ Accelerate Rollouts Teams are achieving environment-wide coverage faster than ever — catching issues before they hit customers, and shipping confidently to every market. — This feature launch is a fundamental step in the direction of building the platform that powers the fastest moving brands to launch stress-free and confidently. Big shoutout to the engineering team for making this possible. Specifically, Ani Emmadi for leading the effort on this one 🚀 more coming!!

48

1 Comment

Exploring Google’s new Antigravity IDE An agent‑first, AI‑powered coding environment built around Gemini 3 that feels like a great option among current AI-augmented IDEs like Cursor, Windsurf, and Claude Code. You can run autonomous agents to plan features, write code, run tests, and even drive a real browser session, all from a VS Code style interface. The best part: Antigravity is currently free in public preview for individual developers on it’s definitely worth a weekend test drive. https://lnkd.in/gJdqJgt7 #antigravity

25

1 Comment

Silicon Valley venture capitalists are shifting focus to industries like accounting, customer service, and property management—sectors once considered too "boring" or low-margin to attract significant investment. What's driving this shift? It's not AI itself but how it's applied. By automating repetitive tasks and improving processes, AI is transforming these industries, enabling them to boost efficiency and profitability. These changes highlight a critical truth: real business value from AI comes not from the technology alone but from applying it strategically to solve everyday challenges. From my experience leading digital transformation, it's clear that any business—no matter how mundane—can unlock growth through innovation and the effective application of AI. A recent article in The Wall Street Journal highlighted this:

16

Platform engineering is the future 🚀 Developers are at the core of everything we do at PagerDuty 💚 Our latest collaboration with Spotify for Backstage brings PagerDuty’s industry-leading Incident Management directly into Backstage, the open platform reimagining developer experience. With the new PagerDuty Plugin for Backstage, devs can see incidents, on-call context, and service ownership right where they already work : Less dilution ; More focus. Let’s go 🎉 Read more 👇 https://lnkd.in/gnsq8Nr9

28

2 Comments

Figma’s stock surged ~250% today, closing at $115/share after pricing its IPO at just $33. That’s not a victory lap. That’s a signal the underwriters got it wrong. When a stock jumps that much out of the gate, it means the company may have left billions on the table. That money went to early traders and institutions instead of the business itself. This raises real questions: - Why are we still using outdated pricing models for tech IPOs? - Would a direct listing or auction-style model have captured more value? - Are traditional underwriters misaligned with founders and long-term investors? Figma built a generational product. But Wall Street's playbook might need a rewrite.

44

4 Comments

Lower your expectations. Putting together my talk for Jin Budelmann's new AI Coding meetup, I've been realising that the more I dgaf about whether AI is going to be useful, the more useful I've been finding it. I now chuck ideas at Claude Code or loop Cursor into a Slack incident channel with absolutely zero expectation that the output will be relevant, accurate or useful. A lot of the times it isn't. More than once it's been revolutionary. Exhibit A: I looped Cursor's Slack agent into an incident channel, with zero confidence it would work. In the background it came up with a bizarre implausible root cause while we were discussing what was going wrong. I ignored it. Maybe 60 minutes later it turned out Cursor was correct, and the PR it had generated was 80% ready to go. I've also started asking agents to build features or solve bugs with the absolute certainty that I'll throw away the results. I continue with other work and occasionally look at the AI outputs. When the results are great, I'm pleasantly surprised and feel like I get work for free. #AI #Agents

62

14 Comments

A year ago, if you had asked me whether our engineering team at Owner.com was truly adopting AI, not just dabbling, but actually integrating it into how we work, I would’ve said no. We had a few early adopters playing with tools. A few GitHub Copilot installs. Some LLMs used here and there. But nothing close to an org-wide shift. That’s all changed. And while we’re still evolving, I’ve seen what it takes to start moving the whole team forward. First: hire people who are curious. We’ve put deliberate effort into bringing on engineers who are open to trying new things, who see AI as an accelerant, not a threat to quality or process. It’s subtle, but it matters. You can’t force this mindset into a team. You build it from the start. Second: find your champions. Every org has a few folks who tinker in their off hours, who show up to standups with something cool they just built. Identify them. Support them. Make them visible. They’re your internal force multipliers. Then: show up yourself. I’ve made a point of using AI in my own day-to-day work. Whether it’s brainstorming in ChatGPT, automating routine writing tasks, or even building simple prototypes, when the team sees leadership leaning in, it shifts the culture. Boldness is contagious. We’ve also created real space for experimentation. Our internal AI hackathons have surfaced some of the most creative, practical tools we now use: automated code generation for straightforward tasks (more on this exciting stuff in a future post), auto-generated tests, and internal bots. Not all of them made it to production, but they created momentum. And that momentum changes everything. On measurement - we’re still figuring it out. Traditional metrics like velocity and story points fall short. We’ve started thinking about time saved, friction reduced, and AI-assisted workflows adopted, but I’d be lying if I said we’ve nailed it. It’s an active conversation, and we’re learning as we go. Throughout it all, we’ve kept our engineering guardrails. Code reviews still matter. Design quality still matters. AI is there to raise the ceiling, not lower the bar. This isn’t about chasing hype. It’s about giving your team better tools, faster feedback, and more creative freedom. The orgs that figure this out now are going to be miles ahead six months from now. There’s no playbook yet. But that’s the opportunity: you get to write your own. At Owner.com, we’re not waiting for someone else to figure it out. We’re leading from the front and taking the future into our own hands.

144

10 Comments

Part 2 of my ongoing series From Vibe Code to Production-Ready is live! This one dives into Claude, Gemini, and MCPs (Zen + Serena) -- not just as shiny tools, but as a framework for moving from quick prototypes into reliable systems: Command patterns, context, and the practices that make "vibe coding" scale. Part 1 explored why vibe coding eventually breaks down, and Part 2 is the bridge: where experimentation starts becoming engineering. Part 3 will explore agent command patterns: Compacting, Context, and Claude Code Ops Read here: https://lnkd.in/e8-dnaYC #AI #MCP #ClaudeAI #Gemini #AgenticAI #Engineering #ProductDevelopment #VibeCoding #FromPrototypeToProduction #TechLeadership

10

𝗧𝗵𝗲 𝗿𝗲𝗮𝗹 𝘀𝘁𝗼𝗿𝘆 𝗯𝗲𝗵𝗶𝗻𝗱 𝗙𝗶𝗴𝗺𝗮 𝗠𝗮𝗸𝗲 𝗶𝘀𝗻'𝘁 𝗮𝗯𝗼𝘂𝘁 𝗔𝗜 𝗿𝗲𝗽𝗹𝗮𝗰𝗶𝗻𝗴 𝗱𝗲𝘀𝗶𝗴𝗻𝗲𝗿𝘀. 𝗜𝘁'𝘀 𝗮𝗯𝗼𝘂𝘁 𝗿𝗲𝗶𝗺𝗮𝗴𝗶𝗻𝗶𝗻𝗴 𝘄𝗵𝗮𝘁'𝘀 𝗽𝗼𝘀𝘀𝗶𝗯𝗹𝗲 𝘄𝗵𝗲𝗻 𝗽𝗿𝗼𝗱𝘂𝗰𝘁 𝘁𝗲𝗮𝗺𝘀 𝘁𝗵𝗶𝗻𝗸 𝗶𝗻𝗳𝗿𝗮𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲-𝗳𝗶𝗿𝘀𝘁. David Kossnick's deep dive on building Figma Make reveals something fascinating: the breakthrough didn't come from throwing more AI models at the problem. It came from years of foundational work on Figma Sites that nobody was thinking about as "AI preparation." Here's what caught my attention as someone who's spent two decades building scalable platforms: 𝐓𝐡𝐞 𝐈𝐧𝐟𝐫𝐚𝐬𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞 𝐏𝐚𝐫𝐚𝐝𝐨𝐱: While everyone's racing to bolt AI onto existing products, Figma spent years building the "boring" stuff. Deterministic code generation, design-to-web translation, new rendering systems. That unsexy infrastructure work? It became the rocket fuel for their AI breakthrough. 𝐓𝐡𝐞 𝐏𝐫𝐨𝐭𝐨𝐭𝐲𝐩𝐞-𝐅𝐢𝐫𝐬𝐭 𝐑𝐞𝐚𝐥𝐢𝐭𝐲: "Prototypes are the new PRD." This hits hard. In our rush to write perfect specs, we're missing that AI development has flipped the validation model. Build, test, and learn faster than ever before. 𝐓𝐡𝐞 𝐓𝐞𝐚𝐦 𝐒𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞 𝐑𝐞𝐯𝐨𝐥𝐮𝐭𝐢𝐨𝐧: Kossnick's team had designers writing system prompts and PMs prototyping. Not because roles don't matter, but because AI tools are dissolving the rigid boundaries that slow us down. The future belongs to teams comfortable with fluidity. 𝑩𝒖𝒕 𝒉𝒆𝒓𝒆'𝒔 𝒕𝒉𝒆 𝒌𝒊𝒄𝒌𝒆𝒓: their evaluation process wasn't some fancy ML pipeline. It was designers in Slack channels scoring outputs 1-4, then scaling that human judgment through FigJam boards and contractor networks. Sometimes, the most sophisticated AI products are built with surprisingly simple feedback loops. The lesson for us platform leaders? Stop trying to predict which AI features users want. Build the infrastructure that lets you iterate fast when they tell you. What patterns are you seeing in your own AI product development? The companies getting this right aren't necessarily the ones with the best models. They're the ones with the best platforms for rapid experimentation. #AIProduct #PlatformThinking #ProductDevelopment #CloudInfrastructure https://lnkd.in/gjrSwYS5

19

2 Comments

I spend hours a week speaking with retail leaders. Too many are submerged in a sea of managing integrations versus actual innovation. This isn’t a new problem–I’ve seen it throughout my career. But there’s a new urgency to solve it. We’re calling it the hidden "fragmentation tax". While some brands remain tangled in integration webs, others like Mejuri have transformed by unifying their commerce operations. Their Chief Digital Officer (Rohit Nathany) now spends time "thinking about customers and growth, rather than worrying about technical debt." What's your experience? Is your tech stack enabling innovation or stifling it? Read my latest blog with Eduardo Frias and Peyman A. N. where we outline the true cost of fragmented commerce systems. https://lnkd.in/ebP4VuX8

68

11 Comments