TRANSPARENCY

How we work

Smart home buyers deserve to know exactly how a recommendation site operates — and what runs underneath it. Here's the whole story, before you ask.

Everything here is research. We don't do hands-on reviews.

Every page here is a research profile: compatibility facts compiled from manufacturer documentation, certification registries (Matter/CSA, Z-Wave Alliance, FCC filings), and published testing. Profiles carry no verdicts, no rankings, no stars — because we haven't held the product, and we won't pretend otherwise. We've decided against publishing hands-on reviews at all: the internet has plenty of opinions, and our value is different — verifiable, sourced, dated facts about what works with what. Our build system mechanically blocks review language on every page — honesty here is enforced by software, not policy.

Every fact has a source and a date

Our database records where each claim came from and when it was last verified — you'll see the date on every page. Records go stale; we re-verify on a schedule, and we track products through their whole lifecycle, including the end: discontinued and bricked products stay in the library with warnings, so nobody buys a dead platform secondhand. (Ask our checker about the Sutro pool monitor.)

What "we haven't covered it" means — and doesn't

The library is what we've verified, not what we believe exists or works. So when the checker comes back empty, that is a statement about our coverage, not about the product: it may work perfectly. We'd rather show you an honest gap than a confident guess, because a guess from us is worth exactly what a guess from anyone else is worth. There are two ways to close that gap. Tell us what's missing and it goes into the research queue for the free library. Or, when our paid Setup Check report opens, order one — it researches every product on your list, whether or not we've already profiled it. Paid research feeds the free library: the general product facts get published, and nothing about the customer does.

The engine underneath — why the data layer is the real asset

The database behind this site is not a product list. Firmware ships, integrations break, subscriptions appear behind features that used to be free, and products get discontinued or bricked — so a catalog that only grows is a catalog that silently rots, and a confident, well-sourced, stale claim reads exactly like a true one. What we're building instead is a knowledge graph — the same kind of structure that powers Google's information panels and Wikidata — with one addition those don't make mandatory: provenance on every claim. Each fact knows where it came from, when it was last checked, and what has to be re-checked when the world moves. A provenance-tracked knowledge graph of the connected home, maintained by AI agents under human ownership.

It's designed to do three things. Discover — continuously look for products and changes we haven't seen yet, and turn messy web pages into typed, sourced facts rather than loose prose. Monitor — because every claim cites its source, a cited source is one we can watch; provenance here isn't just a receipt for how a fact was established, it's the hook that makes the fact checkable again tomorrow. Flag changes — surface the day a fact moves, so a firmware update that quietly breaks a HomeKit pairing shows up as a flagged record instead of a stale page nobody caught. Nothing the agents produce ships on trust: a claim without a cited source fails the build, review-style language is blocked mechanically, and the standards every agent works to are set — and owned — by a human.

Where this stands today: the graph is kept current by agent-assisted research under human-owned standards and software-enforced sourcing rules, and discovery has run as a bounded pilot. The monitoring and change-flagging loops are being built in stages, and a "verified" date is a record of when we last looked — not a guarantee nothing has changed since. We won't describe the catalog as self-maintaining until it is; that promise is the whole point, so we don't get to fake it.

Built on real research, run like living infrastructure

We didn't invent provenance modeling. W3C PROV, RDF-Star, and nanopublications are decades of careful work on saying "here is a claim, and here is where it came from," and our schema borrows from all three. What the standards don't cover is what happens after a fact is written down — and facts about smart home products change constantly. So we added the discipline modern technology teams use to run systems that change: discovery (find what's new), change detection (notice when a recorded fact stops being true), and mapping how everything connects.

Two gaps we had to fill ourselves: the standards have no first-class way to say "we checked, and the evidence doesn't exist" — yet an honest unknown is often the most valuable answer to a compatibility question — and they treat provenance as a backward-looking audit trail, where for us a cited source is the hook that makes a fact checkable again tomorrow.

What it means for you: every fact carries a source and a date, the checker says "unverified" instead of guessing, dead products stay in the library with warnings, and when a fact changes, the page is built to tell you.

AI-assisted, human-owned

We use AI agents to compile research, maintain the database, and draft pages. Every page is published under human ownership and accountability; we publish no hands-on reviews, and AI is never used to fake experience. The honest advantage: AI lets us maintain more accurate, fresher data than a team many times our size — which is the entire point of this site.

How we make money

Today: we don't — we currently have no affiliate relationships and no ads. The plan: affiliate commissions when you buy through links on our pages (always disclosed, on every page that has them), and licensing our compatibility dataset to businesses and AI platforms. Two promises that will outlive the beta: commissions never change a compatibility fact, and we cover products that pay us nothing on exactly the same terms as products that will. But that's just the business model. Our bigger dream is a smart home that's easier to live with, not just easier to buy — one you can enjoy without it becoming a project you maintain.

The mission

"Complexity is inevitable, confusion is not." Home automation's complexity — five ecosystems, five protocols, thousands of devices — is a given. Our job is making sure the confusion isn't: everything we build must make a smart home simpler to choose, simpler to set up, and lower-maintenance to live with. When we fail that test, tell us.