Robbie Tilleard

👀

@aakashgupta Watch out. Once you start looking into the Sydney housing market, you can’t stop. You start mentioning it to friends, speaking with excitement at dinners. People nod. People from Sydney grab you and share their stories. It never ends.

Sydney has 70 beaches within city limits. The coastline stretches 240 kilometers from Palm Beach in the north to Royal National Park in the south. The harbor alone has 317 kilometers of foreshore. More than 40% of the metro area is green space or national parkland. For comparison: Los Angeles has 75 miles of coastline but most of it is private or industrial. New York City has 520 miles of coastline but try swimming at most of it. London has zero beaches. Tokyo has artificial ones. Sydney somehow built a metro of 5.3 million people around one of the most dramatic natural harbors on the planet and kept the coastline almost entirely public. The entire 6-kilometer Bondi to Coogee walk sits on sandstone cliffs above the Pacific and costs nothing. The trade-off: median house price approaching $2 million AUD. The total value of residential property in NSW alone is $5.4 trillion. That single state’s housing stock is worth more than the entire GDP of Japan. The scenery explains the price. The price explains why a generation of Australians is moving to Brisbane, Perth, and Adelaide. Those three cities just made the biggest jumps in the 2026 World’s Best Cities rankings. Perth climbed 15 spots. Brisbane climbed 11. Australia is running a natural experiment in what happens when livability pricing pushes an entire generation to the next tier of cities. The answer: those cities start looking like Sydney did 20 years ago.

@TrySpiral Perhaps I need to upgrade?

@TrySpiral Hello - love the concept. Feedback form isn’t working and I can’t scroll on the settings page in iOS. Was looking for the connect account area and couldn’t find it.

Spiral shipped a lot this month. Here's the rundown: 🐦 X/Twitter style: Connect your account. Spiral analyzes up to 1,000 tweets—weighted by engagement—and builds a voice guide. One click on the Styles page. 🌐 Style from any URL: Paste a page, site, RSS feed, or sitemap. Spiral pulls up to 20 posts and builds your style. No copy-pasting. 🏢 Workspaces: Separate styles, knowledge, and chats by workspace. Invite your team so everyone writes in the same voice. ✨ Editor: Faster across the board. Remembers your last style, autoscrolls, smarter attachments, granular progress. 🔧 Stability: Fixed silent delays in long writing jobs and conversation compression slowdowns. Much smoother now.

@steveruizok Don’t you think the final end state will be a single scrolling window with persistent apps available only if needed? It’ll then be hard for the folks that can’t combine into one properly connected experience - where you have to manage different windows

Congrats to perplexity for being the first to (almost) ship this. Unfortunately this is going to be the only type of product anyone releases for the next six months

@trq212 /btw that’s huge

We just added /btw to Claude Code! Use it to have side chain conversations while Claude is working.

When chat and natural language emerged as the dominant design pattern of AI, I was skeptical it could be the endpoint. I was wrong. It's clear that natural language combined with apps / tools that persist is the ideal design pattern for daily work. Our work will all happen in one window, driven by our initial prompt - rather than the 50 apps / tabs we all have open today. Imagine starting with confirming your tasks, a document opens already drafted, you make a few updates, you send the drafted email, and then cycle through your @SlackHQ messages. All in one workflow without context switching or pressing 100 buttons to do the work. The same pattern is possible across any knowledge work. Technology can bring clarity rather than the distraction and switching we all often experience today. Humans remain in the loop but over time as the agents get better they move up layers of abstraction and are surfaced core decisions when needed. The question becomes whether @Microsoft and @GoogleWorkspace can move to the vision fast enough. Or whether the challengers like @NotionHQ, @salesforce (mostly with Slack), @Superhuman, @OpenAI, and @claudeai can achieve it faster and hoover up the most innovative companies into their ecosystems. Given the shape of the future is coming into view, it's gonna by an exciting time for productivity apps.

@lenjaminbeonard @Ocado Gonna automate my long black order to Trade next

@rtilleard @Ocado Perhaps the most Robbie tweet I’ve ever seen

Hey @Ocado - any chance of an MCP any time soon?

@jasonlk Not yet anyway - it’ll come in the next 12/18 months though

We had an AI agent stop ingesting data four months before we noticed. No errors. No alerts. Just stale outputs wearing the mask of something functional. You can't train an AI agent and then just … go away.

@GeoffreyHuntley @shreyas AE ships feature request during demo call, closes deal 🍾

@rtilleard @shreyas AEs fall into steaks and handshakes still remains super valuable but an AI that uses agents will mog AEs that don’t ghuntley.com/real

The nirvana of the online weekly shop is it can be automated Here is an opensource skill that allows anyone to use @claudeai Code to fill their basket on @Ocado @martinwolf_ don't tell me AI can't increase the UK's productivity 😂 github.com/rtilleard/ocad…

@jasonlk Feels like we’re Gulliver with all the little people climbing all over us! We’re still capable but they can overwhelm us if they swarm

Those of us who are very deep on AI Agents are wildly more productive than before. Already. Today. It's not just engineers. What's less clear is if we can keep this pace up. It's a serious cognitive load to process so many more outputs per unit time. I'm already tired from all our AI Agents. They never stop. And it's only March.

@theshre You’d have to ask @doanythingapp !

@rtilleard Do you know the team? Not a fan of virtual demos but maybe we do an exception :)

What app/workflow recently blew your mind? Saw a demo this week that blew my mind. Got me thinking about doing a demo night in London with apps / workflows that will blow people's mind... and this app was pretty good last time I wanted reccs! 🙏🏽

@lennysan @shreyas Is this….Lennybot?

@rtilleard @shreyas Best reply

The new product management

The companies that have gone deep and been smart have created better models - Agree and Handshake has been pivotal to the jump we’ve all experienced. I wonder how many ppt decks need to be reviewed and graded before the model is good, how many finance models need to be observed and rated before it’s good. It feels hard but repeatable. There is a long runway of tasks and expertise will matter in a crowded market but I’m already impressed by basic Claude Code for finance and ppt - it’s better than a lot of what I’ve seen in large corporates over the years. Will there be significant further marginal gains as people catch up - I’m not sure, it must reach a plateau, at least for a lot of knowledge work

@rtilleard How will it become commoditized?

Larry Ellison is right. Models are trained on the same data. The differentiator now is expert human feedback applied to real workflows. Doctors, lawyers, engineers. People who actually do the work. Observing how models fail in production, providing the cognition to fix it, and feeding that signal back into training. That loop is the new moat: observe failure → extract expert judgment → build verifiers → post-train → deploy → observe new failures. Each cycle makes the model better at harder, longer-horizon tasks that general training data never covered. The companies that own that loop will power the next generation of AI.

The CX and Ops leaders of London are hungry for AI and @claudeai code Fun event last night where 40% of the room (ok, in a straw poll) installed CC for the first time and ran analysis. Hit me up if you’re interested in the next one. Thanks for hosting us @techspace!