Roadmap

Building in public, one honest phase at a time.

This page says what exists, what's next, and — unusually for a roadmap — what we're deliberately not building yet. It gets updated when reality does.

The phases

From bench to production.

  1. Phase 0 · Bench validationComplete
    • · ESP32-S3 drives a real thermal printer: USB enumeration, print, cut
    • · Live paper / cover / drawer status over DLE EOT — with known:false honesty when a printer won't answer
    • · :9100 network-printer emulation, including status replies
    • · Local HTTP API + mDNS discovery + runtime config (PUT /config, NVS-persisted)
    • · Digital twin, conformance suite, dev harness — all in CI (119+ TS tests, ASan C suites, firmware booted in an emulator per push)
  2. Phase 1 · PilotIn progress
    • · Remaining transports behind the same driver seam: RS-232 serial, Ethernet printers
    • · Headless provisioning (first-boot setup without a laptop ceremony)
    • · 5–10 dev-board nodes running local-only in 2–3 pilot restaurants
    • · Repository opens — protocol, firmware, twin, conformance suite
  3. Phase 2 · Custom PCBPlanned
    • · Four-layer boards per variant (Wi-Fi, Serial, Station Hub, PoE)
    • · Reliability hardening from pilot findings
    • · Developer SDK
  4. Phase 3 · First production runPlanned
    • · Commercial manufacturing run
    • · FCC / CE certification
    • · Emulation compatibility profiles for direct-to-restaurant sales

Deliberately not built

The cloud layer is a decision, not an omission.

A hosted control plane (remote access, fleet dashboards, OTA at scale) is genuinely useful — and genuinely dangerous to build too early, because it puts a service in the middle of a product whose whole claim is that nothing sits in the middle.

So it's gated on demand, not on a date: it gets built when pilot restaurants or integrators actually ask for off-site access or fleet management. The seams for it already exist in the protocol — parked, outside the boot path, with nothing depending on them.

When it does exist, it will be optional by architecture. A print job on the LAN never routes through it. That constraint is written into the open-source stance and enforced by the design.

Also not built yet, stated plainly: vendor-protocol emulation (ePOS XML, CloudPRNT), compatibility profiles, any dashboard, serial and Ethernet printer transports, custom PCBs, and the waitlist backend on this very site. When a page here mentions one of these, it says "roadmap", not "feature".

Follow along

Pilots and early access.

The most useful thing right now: a restaurant willing to run a pilot node, or a POS team that wants the simulator in CI.