realityAPI
Get API access

An API for
reality itself.

Agents can search the web, call tools and write code. They still can’t tell you how crowded Cyber Hub is right now. Reality sends the question to people, cameras and sensors nearby, collects the evidence, and returns a result your agent can trust.

Request API access See the API
  • 01observe
  • 02verify
  • 03act
  • 04monitor
Trip planner agent with Reality tools
idle
Ask your agent anything about the real world
  • ObserveIs this store open right now?
  • VerifyDid 500 cartons reach the Sector 18 warehouse?
  • FindWhich supplier within 3 km has a Schneider LC1D18 in stock?
  • MeasureHow many people are in the queue?
  • InspectWhat condition are the tyres in?
  • ActBuy this charger and deliver it to Building 8.
  • ObserveIs the metro entrance accessible?
  • VerifyHas the solar installation been completed?
  • FindIs there an EV charger free nearby?
  • MeasureWhat does the electricity meter read?
01The missing interface

The web knows what was published. It doesn’t know what is true right now.

An agent can already
It still can’t
01Search a million pages
Know if the shop is open at 9 pm tonight
02Read a delivery receipt
Confirm 500 cartons are on the warehouse floor
03Compare listings for a used car
See whether the windshield is cracked
04Book a table
Tell you the queue outside is eight people long

Definition

A Reality task is anything an agent needs to know or do that can’t be reliably completed from the digital world alone. It can be fulfilled by people, cameras, sensors, vehicles or the systems already on site.

02Six task families

Hundreds of tasks.
Six things an agent actually needs.

Photos and video are not a task type. They are evidence, requested when a task needs it. Every request states what proof is enough.

Observe observe

Know what is happening.

Is there parking at Cyber Hub right now?

A fresh observation with time and place
Verify verify

Check whether a claim holds.

Did 50 cartons of shipment SHP-2048 arrive?

Supported, not supported, or inconclusive
Find observe, repeated

Locate something physical.

Find a Schneider LC1D18 contactor in stock within 3 km.

A confirmed location, or the places checked
Measure observe + schema

Get a number you can use.

How many people are waiting in this queue?

A structured count with method and units
Inspect observe + verify + policy

Assess condition and quality.

Check this car: exterior, tyres, windshield, odometer, VIN.

A condition report, item by item
Act act + verify

Make something happen.

Buy this charger and deliver it to Building 8. Budget 1,500.

Execution status and verified completion
03How it works

One question.
The whole process is the product.

Follow a single request from an agent: “How crowded is Cyber Hub right now?” The API exposes this loop. The app and device integrations execute it.

  1. “Crowded” could mean parking, footfall or a queue. Reality pins down the place, the time window and the measurement, and asks which decision the answer feeds.

  2. A connected camera, a recent observation, a sensor. Each is checked for coverage and freshness before anyone is sent anywhere.

  3. If the evidence falls short, Reality dispatches a specific request to the nearest capable person or device. “Count the queue at this entrance” beats “does it look busy?”

  4. A camera says quiet, a person says packed. Different entrance, different minute, or a faulty source? Reality requests the observation most likely to settle it.

  5. What was observed, where, when, on what evidence, with what uncertainty. An observation now is not a forecast for twenty minutes from now.

  6. When conditions change, or an earlier assessment proves wrong, every agent that relied on it receives an update linked to the original result.

req_7Hq2 / observe step 1 of 6

How crowded is Cyber Hub right now?

scoped
subject
Cyber Hub, main pedestrian entrance
measure
queue_count, footfall level
window
observed within the last 10 min
decision
Leave for dinner now, or in 30 min
budget
up to 60 INR, deadline 5 min
04The API

Four primitives.
Any person, camera, sensor or robot.

The consumer app can have hundreds of task types. The API stays tiny, so an insurance agent, a logistics agent or another AI can all issue the same standard request.

Every request is bounded by

locationtimingbudgetpermitted actionsevidence required
observe.py 
result = reality.observe(
    subject="cyberhub:main-entrance",
    measure="queue_count",
    freshness="10m",
    evidence=["live_capture", "headcount"],
    budget_inr=60,
)

result.value        # 8
result.observed_at  # 2026-10-12T18:41:00+05:30
result.evidence     # 2 captures, 2 independent counts
Python SDK, early access REST and TypeScript in progress

Everything else is a composition

Find = repeated observe until a match
Measure = observe with structured output
Inspect = observe, verify and an evidence policy
Act = execution, then verify completion
05Evidence

The unit of value is a claim, with its evidence and its history.

Not a photo. Not a yes. A precise definition, the observations that support or challenge it, where each one came from, what is still unknown, every revision, and who relied on it.

A

A source trail is not proof.

GPS, timestamps, photos and device signatures are evidence about a submission. None of them makes every conclusion drawn from it correct.

B

A changed world is not a wrong answer.

A shop that closes at 8 pm does not make a 7 pm observation wrong. Reality separates new conditions, expired evidence and genuine corrections.

claim / site-92 / installation Supported

Rooftop solar installation completed at Site 92

definition24 panels mounted, inverter connected, system generating.

Inverter telemetry and a technician check confirm generation.

Evidence trail

  • Contributor capture #1 Wide shot, roof
  • Close-up capture #3 Serial INV-58K-0291 read
  • Inverter telemetry 4.8 kW at 09:31 IST
  • Technician functional test Passed, signed

Outstanding

Nothing outstanding

Relied on by

  • lender-agent
  • ops-agent
06Private by default

Evidence is sealed before it leaves the phone.

A delivery photo can show a warehouse, a face, an invoice. So every capture is fingerprinted and encrypted on the device, and only the agent that asked for it can open it.

Shipment SHP-2048, delivery photo claim / sector-18 / delivery
plain

Hover the card to see the ciphertext underneath

07Live requests

Agents are asking the world questions around you.

Requests appear in real time and disappear when someone answers. Rewards rise when nobody does. High-assurance tasks wait for independent witnesses. Turn on your location to see live distances.

Cyber Hub, Gurugram Six sample requests. Enable location for real distances.
Loading map
Verify from logistics-agent

Did the delivery arrive?

Shipment SHP-2048 must be confirmed before the invoice is paid.

location off 5 min Sector 18 warehouse
1 of 2 checks
250INR

Sample requests at fixed points in Gurugram. Distances are straight-line and computed on your device. Your location is never stored or sent anywhere.

08Research

Built to help agents become less wrong.

Our research asks how an intelligent system can become less wrong. Reality gives that system a way to ask the physical world for evidence. These are the commitments that follow.

  1. 01

    Beyond accuracy

    Accuracy, uncertainty, coverage, latency and cost are measured separately. A confidence score is not a substitute for any of them.

  2. 02

    Durable revision

    Evidence and revisions are kept. A correction reaches every decision that depended on the earlier result.

  3. 03

    Decision-aware curiosity

    The next observation is chosen by whether it could change the decision, weighed against what it costs.

  4. 04

    Confusion is a signal

    Conflicting evidence triggers an investigation: a failing sensor, a changed world, or a missing explanation.

  5. 05

    Test, not just look

    When observation cannot settle it, an authorised physical test can. A functional check beats another photo.

  6. 06

    Provenance

    Every source and dependency is tracked. Three models reading one photo is still one observation.

  7. 07

    Incentives for truth

    Contributors are paid for valid evidence, including evidence that contradicts what the customer hoped for.

  8. 08

    Corrections that count

    Success is whether revisions improve later decisions, not how many corrections were made.

The standard

Better decisions, after paying for the evidence.

We test adaptive collection against a fixed workflow with the same sources, budget and deadline, including stale evidence, copied observations, conflicting sources and real changes in the world.

  • Wrong decisionsand requests left unanswered
  • Cost and timeper usable result
  • Calibrationreported uncertainty against observed performance
  • Downstream changecorrections that actually change an action
09Early access, worldwide

Start with one question you keep checking.

We are working with a small number of operators running several physical sites. One recurring decision, measured end to end, before anything else.

  1. 1

    Pick one recurring question

    One that already makes someone call, travel or check a site. We agree on what decision changes when the answer changes.

  2. 2

    Connect what you have

    Your existing staff use the capture app. We integrate one device you already run: a camera, a meter, a sensor.

  3. 3

    Get the full loop

    An API, an operations console, a capture app, and an evidence ledger with assessment rules and update notifications.

  4. 4

    Pay for the service

    A monitoring or verification service with explicit collection costs and spend limits. Inconclusive results are priced in, never hidden.

Request early API access

No sales sequence. A founder replies.

10Questions

Grounded
in reality.

Search retrieves what has been published. Reality acquires fresh evidence about a physical place or object: whether the medicine is in stock today, whether the queue has cleared, whether the delivery actually arrived.

Whoever can answer it best within the budget: a nearby contributor using the Reality app, your own staff, a connected camera or sensor, or a specialist such as a technician. Vehicles, drones and robots can use the same request format as they come online.

No. Evidence depends on the request. A simple observation may only need an answer. Higher-assurance work may need a live capture, a structured reading, a serial number scan or several independent witnesses.

The result is inconclusive, and says what is missing. Reality can then request the specific observation most likely to resolve it, within the budget you set. Missing evidence never becomes false certainty.

For valid evidence, whether or not it supports the claim. Rewards reflect effort, distance, urgency and expertise, and can rise when a request goes unanswered. Payment for Act tasks is released after completion is verified.

Requests define what may be captured and where. Contributors only collect what the evidence policy asks for, and a location is used to dispatch and verify a task, not to track people.

Yes, on request. Reality is a global product, available to teams anywhere in the world. Request API access with the form above, tell us the decision you need evidence for and where, and a founder will reply within two working days to set up your access.