Co-founder & CTO @smplrspace · Building The Mesh, a substrate for AI-native teams · Notes at https://teapawt.net/t.co/VP6j6Du1Ua. Scrappy perfectionist, wannabe essentialist.
Singapore
Joined March 2012
- Tweets2K
- Following1.6K
- Followers290
- Likes4.9K
Pinned Tweet
If you're building a crafted product that uses AI to generate work on someone's behalf, you've probably seen it: what it produces doesn't meet the bar the rest of the product sets. A model drifts toward the generic middle of its training data, and that's not where your product lives.
What closed the gap for me wasn't better prompts. It was giving the model my judgment, not just my topic: the standards, the calls I'd make, the things I'd never ship. I've come to think that context, built up, compounds across a team, and that the tools knowledge workers use should help people invest in it, not promise they can skip it.
New post on how I got here, and the question I can't answer yet: whether it holds at a company of hundreds with a product in millions of hands.
t15n.io/where-taste-drops-ou…
See you in Mumbai next week? 🇮🇳
🏬 Own or operate retail malls?
Meet @wanlyntib (CEO) and @tibotiber (CTO) at @MAPIC_India in Mumbai, September 29 - 30, and see how mall teams use interactive floor plans to:
👉 plan tenant mix
👉 track store performance
👉 pitch new concepts
#CRE #Retail
Ooooohh, the new observable is pretty hot ❤️🔥
Short compelling notebook by @paulchaikin: when @CP3 joined a team, their win percentage went up. observablehq.com/@pamacha/wh…
I've been curious whether Fable 5 is worth the cost, so I gave it something really hard.
We had a problem in the @smplrspace 3D viewer that had been sitting in the backlog for four years. On a really complex building, switching floors meant rebuilding the entire scene from scratch, over 20 seconds of waiting every time. We knew how to fix it, but the rewrite was massive and never worth the disruption given the few spaces this impacted.
We shipped it in two weeks with Fable 5 running almost all of it, 106 commits across 15 pull requests. Floor switches on that same building went from 21.6 seconds to 44 milliseconds.
I wrote up the full story, how it works, what failed the first time, and where I think frontier models actually earn their cost.
t15n.io/the-rewrite-we-avoid…
We've seen a meaningful shift in the size and complexity of what customers are running on Smplrspace. With this update, the platform is catching up. Proud of what the team has shipped here 🚀
Smplrspace now handles enterprise operational complexity:
👉 One billing account across organizations
👉 Use case agnostic project scoping
👉 Usage analytics at every level.
➕ A stack of editor and SDK upgrades
Built for teams running more than one of everything.
#proptech
We've had many requests to provide free spaces for demos, pilots, or dev spaces beyond the free tier. This is our answer to it and I think it's going to make a lot of people happy 🎢
You asked for this, so we built it: the Sandbox organization. 🏖️ 🧩 A free, fully functional place to build, test, and demo. Your bill and your production metrics stay exactly as they are.
And soon you'll have more ways to track your usage and organize your spaces.
#proptech
Over the past few months my team moved to working with Claude as our default thinking partner, not just for code but for real decisions and plans. We hit a problem I didn't expect: we were doing good thinking in chat windows, the conclusion got saved, but the reasoning behind it evaporated when the session closed.
So I've been building something to hold onto that reasoning. I call it The Mesh, and I wrote up what it is and where it came from.
Sharing in case it's useful to anyone working the same way. What happens to all that understanding when the tab closes?
→ t15n.io/the-mesh
→ t15n.io/what-i-mean-by-the-m…
This is pretty exciting, because it goes right where I've been heading with the Mesh these past few months.
The key, as the post highlights, is that agents are far better than people at grooming a knowledge base. They spot the stale sections, catch the inconsistencies, check that the cross-references still hold up. That's the upkeep we're all bad at, and agents turn out to be great at it.
File-based formats are more powerful than they get credit for, too. File-based gets us Git and portability for free. OKF is doing this without a graph database, but maybe it could 🤔. RDF or N3 have given us full semantic knowledge bases on plain files for years.
Keen to do a proper deep dive on this soon.
🚨 @Karpathy predicted the power of the "LLM Wiki." Google just formalized it.
Meet Open Knowledge Format (OKF): a vendor-neutral standard for giving foundation models the curated context they need.
I can genuinely see this replacing Notion, Obsidian, or traditional wikis for developer teams, and the reason comes down to bookkeeping.
Traditional wikis fail because humans inevitably abandon the tedious work of updating them.
As Andrej Karpathy pointed out recently, LLMs don't get bored.
They don't forget to update a cross-reference, and they can touch 15 files in a single pass.
OKF standardizes the interoperability layer so agents can actually do that heavy lifting autonomously.
Because the format is minimally opinionated, it doesn't dictate what you write, it just dictates how it's structured. You get:
→ Human-readable documents that live right alongside your code in version control
→ Cross-links that map out complex entity relationships without needing a graph database
→ A system that survives moving between different tools and organizations
There is no complex compression scheme.
No central registry.
If you can cat a file, you can read it.
If you can git clone a repo, you can deploy it.
This is how we stop rebuilding context pipelines from scratch every time a new model drops.
Announcement + spec file in 🧵↓
🚀
We are thrilled to introduce you to the next evolution of web graphics...Babylon Lite...an entirely new engine that we've always dreamed of making! Faster, Smaller, Pixel for Pixel rendering match to Babylon.js.
buff.ly/VtufEl2
#3D #WebDev #gamedev #indiedev #WebDevelopment #gamedevelopment #IndieDevs
A popular prediction says the UI will be dead in two years, replaced by AI agents. I don't think that's right, and not because I doubt the agents. I just don't see the interface and the agent as rivals.
They're two ways to do the same work: hands on the keyboard yourself, or describing what you want to a fast colleague you trust. As the colleague takes the low-judgment work, what's left in the app is the judgment you'd never hand off, so it matters more, not less.
Having both layers doesn't shrink the design work, it widens the canvas. If you're adding an agentic layer to your app, I'm keen to know: how do you carry the opinion and care you put in your app onto the agent?
t15n.io/the-keyboard-and-the…
Smplrspace started as a developer tool. For years, getting your data onto a floor plan meant an engineer writing code.
Now data analysts and BI teams just drag and drop. No code.
I didn't expect to find this as amazing as I do. Kind of feels magical, even for me.
We just shipped a #PowerBI visual. It's live on Microsoft Marketplace.
Point it at your spaces, drag your fields on, and your data renders right on the floor plan. 2D, 3D, or a portfolio map. Select a room and the other visuals filter to it. No code, and a free tier to start.
Working with Claude has come to feel less like using a tool and more like working with a colleague. The part that doesn't transfer is accountability.
The same care that sets up a colleague for good work — the brief, the why, the constraints that aren't visible from the outside — sets up the agent for success. But the output still has my name on it.
This reminds me of a rule from teams I've led: when a bug ships to production, responsibility sits with the PR reviewer, not the coder. It goes where the catch should have happened.
With an agent, the rule still holds. Whether the output is any good, and of any use, is mine to answer for.
t15n.io/the-colleague-i-cant…
Visualization isn't a presentation layer at the end of a process. It's a thinking layer at the start.
I spent fifteen years answering this question in different rooms, and didn't realize it was the same one until I stopped to look.
t15n.io/fifteen-years
CRE keeps trying to solve data fragmentation by adding more tools. The fragmentation is in the data, not the tools. Smplrspace is the coordinate layer the tools run on.
CRE has a data fragmentation problem. Smplrspace adds the missing coordinate layer: unique spatial IDs for every asset, so Power BI, CRMs, and dashboards can run spatial queries across connected data sources.
Plus, your data never leaves your environment.
#BusinessIntelligence
Got this from @npmjs. My side's been ready for months. One problem: trusted publishing doesn't work with reusable GitHub Actions workflows. Open since 2025, no movement: github.com/npm/documentation…
Are they actually fixing this, or just rotating tokens?
Big fan of @sindresorhus's approach: small scope, done exceptionally well. Been trying Velja for 2 weeks, full-featured free tier, gentle nudge every 12h, no time limit. That's how you build trust before asking for money. Kudos sir 🙌