How Hearth works
Hearth helps homeowners understand their home — its systems, its history, the area around it, and the things that need attention but rarely make it onto a to-do list. We do this by combining what you give us (documents, photos, records of your home's systems) with public data sources you might not have known to check.
This page is where we show our work. It explains what Hearth does, how we organize information about your home, and how we arrive at the findings you see on your dashboard. Everything Hearth tells you about your home comes from somewhere, and you should be able to see exactly where.
What Hearth covers
Emergency procedures
Where to shut off water, gas, and electrical in an emergency — short videos you record once, kept one tap away on your dashboard for the moment you need them.
Inventory
A record of your home's systems and major items — appliances, mechanical systems, exterior assets, and property like vehicles — with photos, receipts, registrations, and service history attached.
Maintenance
A maintenance plan built from what's actually in your inventory: recurring tasks, renewal reminders, and the reasoning behind each one, adjusted for the conditions around your home.
Habitat
What's around your home — environmental conditions, public records, things in your area you should know about. This is where the habitat modules live.
Inventory
A record of the major systems, appliances, and notable items in your home — the things that have a model, a service history, or a reason to be remembered — plus property like vehicles and pets. Each item lives in a room (rooms are seeded for every house and can be renamed, added, or removed), and the full inventory is browsable in one place, grouped by type.
How items get in
The fastest path is a photo. Hearth’s Smart Uploader takes a picture of a nameplate or label, uses AI to work out what it’s looking at, pulls the printed facts off it (manufacturer, model, serial, plus specs like capacity or fuel type), and pre-fills an entry you confirm or edit before saving. If the photo looks like something you already have — a second shot of the same furnace — Hearth offers to add it to the existing item rather than create a duplicate, and an identical re-upload is caught before anything is stored. Items can also be entered by hand. Vehicles work the same way from a registration or insurance card: Hearth reads the VIN, decodes the year, make, and model automatically, and notes the expiration date.
What attaches to an item
More photos, and documents: service receipts and invoices (up to five pages), registration and insurance cards, and similar paperwork. Hearth reads each receipt — vendor, date, line items, total, any serial numbers it mentions — and uses those serials to suggest which item it belongs to. On the item’s page, every document shows both the original pages and a “what we found” summary of what Hearth read off it.
What Hearth does with it
Each item has a detail page with the captured facts as pills, its photos, and an on-demand “Research this model” panel that uses an AI web lookup for what’s generally known about equipment like yours — typical service life, common maintenance, things to watch for — with source links so you can verify. Alongside research, a separate reasoning model tries to decode the manufacture date from the serial number; Hearth records that date only when the decode is high-confidence, so an uncertain guess never lands on your page. A History section lists completed maintenance alongside install and purchase milestones.
What it’s for
The inventory is the foundation the rest of Hearth leans on. Research feeds the maintenance plan (see below), renewal documents create renewal reminders, and reports draw on it — the Water Quality Report is live today; a pre-listing export, a contractor brief, and warranty tracking are planned.
Maintenance
Hearth builds a maintenance plan from the specific items in your inventory rather than a generic checklist, and explains every task it schedules. Tasks arrive two ways: an AI pass that turns what Hearth has researched about an item into a schedule, and renewal reminders created directly from documents that carry an expiration date.
Where tasks come from
When you research an item, Hearth hands what it learned — the expected service life, the manufacturer’s maintenance guidance — together with the item’s install date, any service receipts on file, and the habitat findings for your home to a reasoning model, which returns a structured set of recurring tasks: what to do, how often, and why. Most items land between four and eight tasks. The plan builds automatically after a successful research run and can be rebuilt any time; rebuilding replaces the AI-generated tasks but never touches reminders that came from your documents.
The second path involves no AI at all. When a document with an expiration date is attached to an item — a vehicle registration, an insurance card — Hearth writes a renewal task due on that date. Upload next year’s card and the old task is marked done with a new one chained after it.
How the schedule is anchored
Each task’s first due date is grounded in something real when possible: last year’s furnace-service receipt anchors next year’s, an install date anchors the first filter change, and only when neither exists does the schedule start from today. Tasks tied to using the appliance rather than the calendar — checking rinse aid, clearing the lint screen — are kept as “every time you use it” practices instead of being forced onto a monthly interval.
How your home’s conditions change the cadence
Habitat findings don’t create tasks; they adjust ones the item already needs, and the adjustment is recorded on the task. Water is the clearest example. When your utility’s annual water quality report lists hardness, iron, or manganese, Hearth classifies the water (soft, moderate, hard, very hard) and the plan shortens the intervals for water-touching equipment — the water heater’s anode rod, softener resin, dishwasher rinse aid, faucet aerators. A flood-zone finding makes a sump pump test more pressing. A unit near the end of its expected service life gets tighter intervals too. Every adjustment appears under “Why this task” with the finding it came from.
What Hearth leaves out
Not everything a manual mentions is a recurring task. Installer setup steps — leveling, anti-tip brackets, initial clearances — are one-time work the installer did and are left out entirely. “Call a technician if the igniter is slow” has no honest interval, so it becomes something to notice while you use the appliance rather than a date on a calendar. Work you’d realistically do in one session is merged into a single task instead of one per sentence of guidance. Gas, propane, and oil appliances always get an annual clearance-and-combustibles check even when their guidance omits it.
Where tasks show up
The dashboard’s “On your plate” panel shows anything overdue plus the next handful of upcoming tasks, each labeled with its appliance; the full list for an item lives on that item’s page. Opening a task shows the instruction, the due date, the reasoning — where the task came from, why this cadence, what adjusted it — and its history. Marking a task done or renewed closes it and schedules the next occurrence from the date you completed it; renewal tasks offer the issuer’s term lengths (one or two years for a Michigan registration, six or twelve months for insurance) or let you enter the exact new expiration.
Habitat module methodology
One section per habitat module Hearth runs against your home. Each section names the data source, lays out exactly how a finding gets its severity, and links back to the upstream authority so you can verify.
Radon zones
Hearth checks the EPA Map of Radon Zones for your county. The EPA published this map in 1993 and republished it most recently in June 2024. It groups every U.S. county into one of three zones based on predicted average indoor radon concentrations, before mitigation.
| EPA Zone | EPA's predicted indoor radon | Hearth severity |
|---|---|---|
| Zone 1 | Greater than 4 pCi/L (above EPA action level) | Concern |
| Zone 2 | Between 2 and 4 pCi/L | Caution |
| Zone 3 | Less than 2 pCi/L | Favorable |
Important caveat
The EPA zone is a county-level predicted average. An individual home in a Zone 3 county can still test high; a home in a Zone 1 county can still test low. EPA recommends testing every home regardless of zone. Hearth surfaces the zone as a starting point — your actual radon level can only be known by testing your specific home.
Superfund proximity
Hearth checks EPA's Superfund Enterprise Management System (SEMS) for every NPL-relevant Superfund site in your state, measures the straight-line distance from your home to each one, and surfaces the closest sites that fall within Hearth's proximity model. The data is live — every check pulls current EPA records.
How we classify — proximity tiers
Each Superfund site in your state is sorted into a tier based on distance and listing status:
| Distance from home | NPL statuses included |
|---|---|
| Tier 1: 0 – 0.5 miles | Any (Final, Proposed, Part of NPL, Deleted) |
| Tier 2: 0.5 – 2 miles | Final or Proposed only |
| Tier 3: 2 – 5 miles | Final only |
| Beyond 5 miles | Not surfaced |
How we map tier and NPL status to severity
| Tier | NPL status | Hearth severity |
|---|---|---|
| Tier 1 | Final or Proposed | Concern |
| Tier 1 | Part of NPL or Deleted | Caution |
| Tier 2 | Final or Proposed | Caution |
| Tier 3 | Final | Neutral |
| No qualifying sites within 5 miles | — | Favorable |
How we label each site for your situation
Severity grades a site’s distance and listing status on the dashboard. Separately, Hearth gives each site — and the finding as a whole — a label that says how relevant it is to someone deciding whether to act: Worth acting on, Worth knowing, or Informational. When EPA hasn’t published enough for a confident read, the label is suppressed rather than shown as false reassurance.
The label weighs distance, NPL status, the most serious contaminant EPA lists at the site, and — if you shared them during onboarding — your water source and whether you have a basement. A groundwater contaminant two miles away is Worth knowing for most homes but Worth acting on for a well user; a vapor-forming contaminant within half a mile escalates the same way for a home with a basement. Unknown answers keep the label at its conservative value — skipping onboarding never inflates it.
Portfolio summary
When nearby sites exist, a language model writes a two-to-four sentence summary from the structured facts (names, distances, statuses, contaminants, labels): what’s there, what stands out, any pattern across sites. It is forbidden from claiming your property is or isn’t contaminated — that’s for EPA, your utility, and licensed testers. The summary is generated once per check and stored; if the model call fails, the finding ships without it.
How Hearth's model relates to EPA's guidance
EPA uses 1-mile and 3-mile rings in its community-involvement work. Hearth’s 0.5 / 2 / 5-mile tiers are our own synthesis, informed by that practice, because living half a mile from an active cleanup is a different experience from living two miles away. Tier classifications in a Hearth finding are Hearth’s, not EPA’s.
A note on precision
EPA publishes one representative point per site, even for sites that stretch for miles along a river or across several parcels. For those, the distance Hearth computes can be off by miles, and the finding says so. The Allied Paper / Portage Creek / Kalamazoo River site is the local example.
Site contact and documents
Each nearby site links straight to its document library on EPA’s site profile. When EPA has assigned a Community Involvement Coordinator — the contact for public questions, as opposed to the technical project manager — Hearth shows their name, email, and phone; when there isn’t one, it says so.
Recommended actions for your situation
Beneath the summary, Hearth lists concrete next steps matched to your setup. Well users see a prompt to test for the relevant contaminants; city-water customers see a prompt to read their utility’s water quality report; a basement near a site with volatile chemicals within half a mile prompts a vapor-intrusion check. When your water source or basement is unknown, Hearth withholds these rather than guess — you can fill them in any time from Home details.
Contaminant summaries on each site card
Each site card carries a plain-English line — “Heavy metals,” “PCBs and dioxins, heavy metals” — grouping EPA’s listed chemicals into a few consistent categories so you can triage without opening the site. Sites with no published contaminants show no line.
Sites we filter out
Hearth only shows sites where EPA has published a contaminant inventory. Some database entries are sub-records of a parent listing with no details of their own; those are dropped, and the finding’s activity log names what was filtered.
Inside a single site
Opening a site shows its name and address, the precision caveat when relevant, what makes it distinct within your nearby set (closest, only active cleanup, and so on), how its contaminants typically spread (groundwater, vapor, and so on), the EPA contact and documents, the full contaminant list, and a note on where the data came from and when it was last checked. The pathway explanations are fixed text written once per pathway, not generated per site, so the framing stays consistent.
FEMA flood zones
Hearth checks FEMA's National Flood Hazard Layer for your home's coordinates and returns the official FEMA flood zone designation for your property. FEMA's flood zones come from local flood studies and are the basis for federal flood insurance requirements and floodplain management regulations.
| FEMA zone | What it means | Hearth severity |
|---|---|---|
| Zone X (minimal hazard) | Outside both the 100-year and 500-year floodplains | Favorable |
| Zone X (shaded — 0.2% annual chance) | In the 500-year floodplain; flood insurance not required but recommended | Neutral |
| Zone D | Possible but undetermined flood hazard — FEMA hasn't formally studied this area | Caution |
| Zones A, AE, AH, AO, AR | 100-year floodplain — Special Flood Hazard Area; federal flood insurance required for federally-backed mortgages | Concern |
| Zone AE with FLOODWAY subtype | Regulatory floodway — the active channel that carries flood flows | Critical |
| Zones V, VE | Coastal high-hazard zone — flooding plus wave action | Critical |
| No FEMA flood map data for the area | FEMA's digital coverage doesn't include this location (about 10% of U.S. addresses, mostly rural) | Neutral (with note) |
About base flood elevation (BFE)
For homes in detailed AE-zone studies, FEMA publishes a Base Flood Elevation — the elevation that floodwater is expected to reach during a 100-year flood. When we have this number, Hearth surfaces it. This is the elevation reference used in flood insurance pricing and elevation certificates.
How Hearth's model relates to FEMA's classifications
The flood zone codes (X, AE, V, etc.) are FEMA's. The severity mapping is Hearth's interpretation — most of it is straightforward (an AE-zone home is in a Special Flood Hazard Area, period), but the X-shaded-versus-unshaded distinction is one we surface intentionally because most consumer-facing flood lookups collapse both into "Zone X" without explaining the difference.
Water Quality Awareness
Hearth turns the drinking-water data your utility already reports — EPA’s record of the system serving your address, its compliance history, lead and copper results, and the utility’s annual water quality report — into a plain-language picture of what’s in your tap water, what it means, and which treatment addresses it.
| Situation | What it means | Hearth severity |
|---|---|---|
| private_well (you told us) | Your home is on a private well or shared private system. EPA doesn't monitor these — testing is on you. | Neutral |
| cws_unmapped | You said you're on city water, but EPA's map doesn't pinpoint a utility at your address (about 1 in 7 U.S. addresses). Without knowing the utility, Hearth can't read its compliance data. | Neutral |
| City water — nothing detected | No active health-based violations, every lead and copper sample on file is below the detection limit, and — when a report is on file — nothing in it reaches a caution level. | Favorable |
| City water — something detected | Any detected lead or copper, even well below the federal action level; any PFAS detection; or a contaminant at 80% or more of its federal limit. | Caution |
| City water — over a limit | An active health-based violation, lead or copper at or above the federal action level, or a reported contaminant at or above its federal limit. | Concern |
| City water — can't tell | EPA's compliance feed couldn't be read, or no lead and copper samples are on file — no positive signal either way. | Neutral |
| non_community | Served by a non-community system (school, campground, small business). No annual report is required; we lean on the compliance feed. | Neutral |
| stale | EPA returned an unexpected or inactive record for your utility. We'll retry on the next refresh. | Neutral |
Where the data comes from
Four EPA sources plus one from you. EPA’s service-area map identifies the utility at your address; EPA’s inventory record describes it; the SDWIS compliance feed lists violations; Lead and Copper Rule sampling gives the 90th-percentile lead and copper results. The fifth source is your utility’s annual Consumer Confidence Report, which a homeowner uploads once and Hearth reads into structured data shared with every household on that utility. If an EPA feed is unavailable, the finding says it couldn’t check rather than implying a clean record.
Why we trust your onboarding answer over EPA’s map
EPA’s map covers about six of every seven addresses; the gaps are mostly rural fringes, recent annexations, and parcels served by a city utility but mapped just outside its boundary. If you told us you’re on city water, we believe you even when the map comes up empty — calling a known city-water customer a private well would be the wrong answer.
When the map misses
If your exact point isn’t inside any utility’s boundary, Hearth checks a 500-meter circle around it. When every utility nearby is the same one, we use it and mark the match as inferred; you can confirm it or enter your utility’s ID to correct it. When several utilities are nearby, we don’t pick one.
How severity is decided
Two axes from EPA — compliance and lead/copper — plus a third from your utility’s report when one is on file. Concern wins immediately: an active health-based violation, lead or copper at or above the action level, or a reported contaminant at or above its limit. Otherwise any detected lead or copper, any PFAS, or a contaminant at 80% or more of its limit is caution — federal limits are regulatory thresholds, not health-safety ones, so a detection is worth knowing about at any level. Favorable requires all three axes clean, and a clean record with no samples on file stays neutral. Monitoring and reporting violations are paperwork between EPA and the utility and don’t move the severity.
Your utility’s annual report
Upload the Consumer Confidence Report once and Hearth extracts the measured data — every detected contaminant with its level and limit, the lead and copper distribution, PFAS monitoring, and any free-testing offer — and ignores the brochure content. Each contaminant is matched to Hearth’s reference for a plain-language description and an EPA link. From the detected list, a remediation matrix shows which filter technologies address which contaminants and recommends the combination with the best coverage. When the report lists hardness, iron, or manganese, that feeds your maintenance plan (see the Maintenance section above).
Boil water advisories
For the cities it watches, Hearth checks the official advisory source every 30 minutes and records each boil water advisory as it is issued, updated, and lifted. Your water finding and the public water page show the most recent few, oldest to newest, with an advisory that is still in effect called out. Where a city’s own page can’t be read from the cloud, Hearth watches a local newsroom’s feed instead and says so — that source typically runs one to three hours behind the city. Advisories that apply across the whole water system are treated as district-wide; ones that name a few streets are recorded as localized. Notifications are district-wide only for now.
Year-over-year trends
With more than one year’s report on file, Hearth lines each contaminant up across years and shows which way it’s moving — an arrow next to the number, a sparkline at three or more readings, and a chart of every reading (hover the sparkline, or tap the chart icon) with each year’s level printed beside its point, against the limit the report stated that year. We compare the two most recent years with a measured value; a change within about 10% reads as stable so ordinary measurement noise doesn’t look like a trend.
Every trend states how many readings it rests on and which years. A year that doesn’t list a contaminant leaves a gap, never a guessed value — a missing row could mean not detected or simply not printed, so Hearth never claims a new or cleared detection it can’t prove. A contaminant still under its limit but at half the limit and rising is escalated to “worth knowing” so it surfaces on your dashboard; it never escalates further while it stays under the limit.
About the data and our limitations
Hearth's findings are interpretations of public data, not authoritative determinations. We pull from EPA, FEMA, Mapbox, and other public sources, and we apply our own classification logic to turn raw data into something useful. If you're making a decision that has legal or financial weight — buying flood insurance, contesting a property assessment, filing an environmental claim — please use Hearth as a starting point, not a substitute for talking to a professional in that field.
When the underlying data updates, we update with it. When our classification logic changes, this page changes too. Anything Hearth tells you about your home, you can trace back to its source through the activity log on each finding.