Open source project

Open source environmental sensing, grounded where you are.

Climate hazards are getting more local. The infrastructure to sense, interpret, and act on them at the site level — with trust, privacy, and honest uncertainty — is what we're building.

OESIS fuses indoor and outdoor sensor readings, site context, public hazard feeds, and nearby shared evidence into one trust-scored view — with confidence levels and reasons shown for every output.

Every output shows what was observed, what was inferred, and how strong the evidence is — so you can check the reasoning, not just trust a score.

Smoke, heat, floodTrust-scored evidenceYour data stays yoursAdvisory, not alerts

System pillars

Three rules that hold across every release

The tools and software will evolve, but these three principles are permanent.

  • Parcel firstThe product thinks about your whole site, not only a regional forecast.
  • Private by defaultSensors and readings belong to the parcel operator unless you explicitly share.
  • Shared by choiceAnything broader than your site is optional and controlled by policy you can see.

Who this is for

Find your path

Start where you are; depth stays one click away.

Audience path

Parcel operators and neighbors

Short reads first

Start with the problem gap and parcel-first framing, then governance for ownership, sharing, and claim limits.

Audience path

Builders and integrators

Plain language, then specs

Evidence layers, modes, and advisory outputs in plain language—then hardware, software, and schemas on the open release hub.

Audience path

Partners and pilots

Align expectations, then engage

Follow the roadmap, pilot scope, and how to plug into deployment—after the short Why it matters / How it works spine so expectations stay aligned.

Why this matters long term

The gap between regional data and your site is growing

Climate-related hazards are becoming more frequent and more localized. Public alerts tell you what is happening regionally — but not what it means for your specific site, your indoor air, or your next decision as an occupant.

NowOne parcel, real sensors

Prove the pipeline works end to end: sensor readings through inference to a parcel view with trust scores and honest uncertainty.

NextNeighbor-aware, privacy-first

Test whether nearby parcels can share coarse signals without exposing private data — making each parcel smarter without central surveillance.

LaterNeighborhood-scale intelligence

Communities understand shared risk collectively, with governance that keeps each occupant in control of their own data.

EventuallyComplements official systems

Parcel-level evidence complements emergency services, insurers, and utilities — better ground truth for institutions, better context for occupants.

Each step only works if the previous one earns trust through field use. That is why the current scope is deliberately narrow.

Active development

Version arc

v0.1 through v1.0 are acceptance-gated runtime lanes — each is a real working build with its own test suite. The next build target is v1.5, bridging from measurement to intervention.

0.1–1.0Implemented

Reference pipeline through fielded parcel system

Six acceptance-gated runtime lanes: reference pipeline, indoor+outdoor sensing, flood coverage, trust scoring and node registry, governance enforcement, and extended support objects (house state, intervention, verification).

What this adds

Bench-air + mast-lite + flood-node, node registry, five-factor trust scoring, consent/revocation/retention/export enforcement, house state and intervention tracking, contrastive explanations, divergence analysis.

1.5Next build target

Measurement-to-intervention bridge

Adds the minimum bridge from hazard sensing to building state, action, and measured outcome reasoning, including indoor response and outage-aware support surfaces.

What this adds

Introduces response and verification support surfaces without implying full automation or mature controls compatibility.

2.0Future direction

Block intelligence

First sparse shared intelligence across nearby parcels.

What this adds

Introduces block-scale derived intelligence without abandoning parcel ownership rules or the parcel-first operating model.

3.0Future direction

Neighborhood network

Neighborhood coordination becomes a first-class product layer.

What this adds

Extends from block-scale inference to broader neighborhood coordination, richer shared signals, and clearer network behavior.

4.0Long-range direction

City federation

Federated local systems connect at city scale.

What this adds

Connects larger areas without collapsing local control, parcel ownership, or neighborhood-level governance.

Near term

Current pilot focus

The field pilot moves the program from reference docs to real sites in one defined local geography. It focuses on repeatable installs, structured feedback, and operational learning. Geography, goals, and partner paths are on the pilot page.

Latest progress

v0.1–v0.5 scope and site alignment

April 2026

The site now describes active v0.1–v0.5 engineering lanes: reference pipeline, indoor+outdoor sensing, flood coverage, trust scoring, and governance enforcement, while keeping packet-status boundaries explicit. Next: pilot geography with partners and field milestones as they land. Project note.

Organize

Help shape the pilot

We are organizing community partners, pilot sites, builders, and funders. Pick a path on Get involved—or start with the Pilot page if you represent a site.

Top