Review: 2025–2026 Watchlist — Newcomers, Comebacks, and Corpses

Review: 2025–2026 Watchlist — Newcomers, Comebacks, and Corpses

Part of the A3DP remote-monitoring series. Verified 2026-10-23.

One-liner: The live edge of this category: an AI-assistant MCP server, a dead farm tool, a vanished Klipper startup, and one genuinely interesting open question — does anyone actually benchmark AI failure detection?

1. OctoEverywhere remote MCP server — the actual newcomer (2026)

The most interesting recent feature in the whole survey isn't a printer feature — it's plumbing for AI assistants. OctoEverywhere now offers a free remote MCP (Model Context Protocol) server for your printers (https://octoeverywhere.com/mcp): connect Claude, ChatGPT, Gemini, Grok, or a local model, and it can read live print status, pull webcam snapshots, access files, diagnose errors — and optionally control the printer, gated by "fine-grain security tool permissions" with "no public internet exposure".

Why it matters: it's the first credible "ask your AI what's wrong with my print" path that doesn't require port forwarding or DIY plumbing. It also quietly changes the monitoring question: if your LLM can pull status + snapshots on demand, the push model (alerts) gains a pull sibling. Worth a follow-up piece on its own.

Also from the same org: bambuddy (Bambu-focused companion, org repo pushed 2026-10-14) — OctoEverywhere is expanding beyond Klipper/OctoPrint into the vendor-locked ecosystems.

2. MattaOS — the one that got away (Feb 2026 → gone)

Matta-Labs/moonraker-mattaos appeared in GitHub search results on Feb 24, 2026 with a pitch that reads like a category-killer: "Connect your Klipper printer to MattaOS and join us in creating AI-powered closed-loop control for 3D printing. As a bonus, securely access and control your machines from anywhere." Closed-loop AI control — not just detecting spaghetti but reacting to it — is the grail this whole series circles.

As of 2026-10-23 the repo is deleted from the Matta-Labs org, and matta.ai now sells factory AI inspection (Sentry defect detection, Gauge dimensional measurement, Tally, Trace) — an industrial pivot. We could not find a formal announcement ending the consumer Klipper product (UNVERIFIED). Lesson: the "AI closed-loop control" idea is real and companies are chasing it — but the consumer-side attempts are fragile.

3. OctoFarm — the corpse worth knowing about

Self-hosted OctoPrint farm manager: 358★, last push May 2023 (GitHub API, checked 2026-10-23). It still shows up in "best print farm software" lists around the web, and OctoEverywhere even advertises OctoFarm remote access. It works, but it is unmaintained. If a reader is choosing a farm stack today, SimplyPrint's Print Farm tier (or a maintained self-hosted alternative) is the defensible recommendation.

4. Printer-vendor AI — context, not contenders

Bambu Handy (4.4★ on Play), Creality's on-device AI, Prusa Connect: all real monitoring ecosystems, all locked to their own printers. For a Klipper-first audience they're context: the vendors proved that buyers want failure detection enough to ship it on-device — which validates the third-party tools' premise. Obico's own blog (July 2024) compares these systems; vendor source, no independent benchmarks.

5. The open question the whole industry skips: nobody benchmarks this

We searched specifically for independent, quantitative evaluations of AI print-failure detection — Gadget, Obico, SimplyPrint, vendor systems: false-positive rates, detection latency, precision/recall on common failure modes. Nothing rigorous exists as of 2026-10-23. What exists: vendor claims, vendor-vs-vendor blog posts, and a decade of community anecdotes (including r/3Dprinting's beloved false-positive memes).

That's the hole an honest blog series can own: "every accuracy claim in this space is unverified by anyone other than the seller." It's also a series asset — a we ran 100 prints through all three detectors follow-up would be the first real benchmark in the niche. (See the open-questions section of survey.md.)

Watchlist verdicts