Writing
Field notes from the studio.
Posts about the engineering decisions behind real projects — and the trade-offs that don't survive a tweet.
2026-08-17
iPhone LiDAR Indoor Mapping: The Industries Using It and the Accuracy They Get
How iPhone LiDAR builds digital maps of indoor spaces, which industries run it in production - construction, real estate, insurance, facilities, warehousing - the measured accuracy at room and building scale, and the pipeline that turns a scan into a usable map.
2026-08-17
iPhone LiDAR Roof Measurement: What Works, What Doesn't, and the Real Accuracy
Why iPhone LiDAR cannot scan most roofs from the ground, how smartphone roof measurement actually works - guided photos and photogrammetry - the published accuracy numbers, and where LiDAR still earns its place in roofing and claims workflows.
2026-08-17
LiDAR + LLMs: Scan-to-Report Is the Next Enterprise Mobile Feature
Scan-to-report pairs mobile LiDAR capture with a language model so a spatial scan comes out as a finished deliverable - a claim narrative, an inspection report, a maintenance checklist. No iOS category leader owns the pattern yet. Here is the architecture, the on-device question, and who should build it.
2026-08-06
iPhone vs Terrestrial Laser Scanning: An Honest Comparison for Surveyors
iPhone LiDAR vs terrestrial laser scanning compared on accuracy, range, speed, and cost - what the peer-reviewed data says, where each one wins - from survey sites to forestry - and how to match the instrument to the job.
2026-08-06
RoomPlan in Production: Limits, Pitfalls, and Workarounds
What Apple's RoomPlan API actually does in a shipping product: the documented operating limits, the exceedSceneSizeLimit and wall-semantics pitfalls, and the engineering workarounds that make scans usable downstream.
2026-07-29
ARKit + LiDAR Business Applications: The 2026 Industry Map
Where iPhone LiDAR actually earns money in 2026: construction, insurance claims, real estate, dimensioning, field service, healthcare - the problems solved, the published numbers, and the pattern every successful product shares.
2026-07-29
iPhone LiDAR Accuracy in 2026: What the Peer-Reviewed Data Actually Says
Measured iPhone LiDAR accuracy from peer-reviewed studies: objects, rooms, buildings, clinical scans - plus the hard physical limits and why the app's processing pipeline matters more than the sensor.
2026-07-29
Scan-to-CAD/BIM on iOS: From RoomPlan Mesh to Revit-Ready Models
What it takes to turn an iPhone RoomPlan or LiDAR scan into CAD and BIM deliverables: RoomPlan's real limits, the USD-to-IFC format gap, the cleanup pipeline, market pricing, and build timelines.
2026-07-29
How Insurance Carriers Cut Claim Cycle Time With Smartphone Capture
How property insurers use smartphone photo and LiDAR capture to cut claim cycle time: the published numbers from Hover, Polycam, and magicplan, the Xactimate export economics, and what it takes to build this capability.
2026-07-20
March 2027 Is Closer Than You Think: Android's AppFunctions Mandate
Google has set a deadline: by March 1, 2027, apps in six major categories must expose AppFunctions. What the mandate covers, what it means architecturally, and what to do this quarter.
2026-07-20
The AI Feature That Beat the Cloud: Inside Kakao Mobility's Decision
A teardown of the best public head-to-head test of cloud vs on-device AI: what Kakao Mobility benchmarked, why on-device won on business metrics, and what the method teaches any product team.
2026-07-20
Your AI Bill Shouldn't Grow With Your Success
The unit economics founders skip: what a popular AI feature costs on cloud APIs, why on-device inference costs $0 per request, and how much of a real workload can actually move to the phone.
2026-07-20
HIPAA, GDPR, and the Architecture That Makes Lawyers Relax
Why processing sensitive data on the user's device changes the compliance conversation for health, finance, and privacy-first products - with the real examples and the honest limits.
2026-07-20
What On-Device AI Still Can't Do (And Why That's Fine)
The limits vendors skip: context windows, memory gates, thermals, and the quality gap - and how to design a hybrid architecture that respects them instead of discovering them in production.
2026-07-19
On-Device AI in 2026: What Actually Ships
A practitioner's state of the union: the free system LLMs Apple and Google now give every app, real benchmark numbers on real phones, what it costs, and what still belongs in the cloud.