Blog
Notes from the work.
Discovery calls, technical decisions, and the things we’re learning by talking to hardware teams. Written by the people doing the work, in public, while it’s fresh.
18 posts
Editor’s pick
Why we're writing this
A short, honest note about what this space is for, and what it isn't. Notes from the work, in public, written by the person doing it.
Recent
What happens when the AI gets the integration wrong
Most AI-generated-driver pitches stop at "the model figured it out." That works 70% of the time. The other 30% is what actually decides whether the platform works in production. Here is the engineer-in-the-loop fallback for the failures the verification layer cannot catch.
From device manual to live cloud data, with no firmware engineer in the loop
A walkthrough of what actually happens between dropping a device datasheet PDF onto SCADABLE and the first event landing in your cloud. The drag-and-drop is real. The mechanism underneath is a five-stage pipeline we built so the customer doesn't have to know it exists.
Hardware teams shouldn't have to become IoT experts to ship a connected product
A statement of where SCADABLE is going. Every connected hardware team hits the same six walls in the same order, and none of those walls have anything to do with the product they are actually trying to ship. The vision is to make that whole layer disappear.
We shipped ESP32 support, and it took a year longer than Linux
A founder-voice walkthrough of the four hard problems that turn ESP32 from a $5 hobbyist board into commercial firmware infrastructure. The 520 KB of RAM, the missing filesystem, the FreeRTOS scheduling model, and the factory-line cert provisioning flow.
ESP32 plus Modbus TCP, from sensor to cloud in 40 lines of Python
Most ESP32 tutorials wire up a DHT22 and publish to test.mosquitto.org. This is the version where the ESP32 is a Modbus TCP master polling a real PLC, and the data ends up in your production cloud through a unified API.
MQTT vs Modbus TCP, when to use which (and when to bridge)
A field guide to picking between MQTT and Modbus TCP for industrial and connected-product fleets, with a decision tree based on fleet size, regulatory posture, and whether you are retrofitting or building greenfield.
How much does an embedded engineer actually cost a hardware startup in 2026?
The salary aggregators tell you base. They do not tell you the 12-month all-in cost, the time-to-fill burn, or the single-point-of-failure risk. Here is the honest model.
The Verification Layer: How to Trust AI-Generated Integration Code in Production
The bug isn't that LLMs hallucinate. It's deploying their output without a verification layer. Here are the six concrete checks that catch AI mistakes before they reach a regulated fleet.
How we generate Python device classes from datasheets (and the failure modes we still see)
A walk-through of SCADABLE's pipeline that turns datasheet PDFs into runtime-verified Python device drivers, including the LLM mistakes we still see in production.
Embedded engineering is the next AI-native service
YC named insurance, accounting, compliance and healthcare admin as the verticals AI is now ready to replace. We think embedded engineering belongs on that list, and the platform underneath the AI is what makes it possible.
IoT Middleware vs Library: Why Your Protocol Translation Layer Doesn't Scale
Middleware that translates between heterogeneous device protocols grows linearly with every new protocol you onboard. A library shape doesn't. Here is the architectural reason, and when each is the right call.
Why Your Hand-Rolled BLE Multi-Device Pipeline Stops Scaling Around 10 Devices
If your ESP32 BLE central works fine at 3 peripherals and falls apart at 10, the problem is structural, not a tuning issue. Here is what actually breaks.