Any real-world signal → agent-drafted, human-approved rules → provable, hash-chained ROI.
The thesis, in full ▸
Kairos is the moment layer for marketing. Signal adapters normalize any real-world feed - weather, air quality, cricket, traffic, markets - into one contract; an agent drafts governed rules from a plain-English goal; a human approves every one; a single deterministic engine executes by explicit priority with suppressions, hysteresis and dwell; and every decision lands on a tamper-evident hash chain. Any signal in. Provable ROI out.
The whitespace
Every incumbent is a weather company with an ad product - the gaps are the columns.
| Platform | Signals | Rule authoring | Governance | Attribution |
|---|---|---|---|---|
| WeatherAds | Weather only | Manual - self-serve campaign forms | None published | Campaign reports / exports |
| WeatherFX (The Weather Company) | Weather (+ location context) | Manual - managed enterprise setup | Enterprise contracts; no decision ledger | Media lift reports |
| WeatherTrigger | Weather only (API triggers) | Manual - developer integration | None published | Report exports |
| Tomorrow.io | Weather-centric intelligence | Manual - insight & alert workflow builder | Ops audit basics; no evidence chain | Ops alerting - not ad incrementality |
| KAIROSTHIS DEMO | Any signal, one adapter contract - weather, AQI, cricket, traffic, markets today; anything normalizable tomorrow | Agent-drafted from a plain-English goal → human-approved; every edit logged | Hash-chained evidence ledger · named approvals · explicit priority & suppressions · fail-safe creative | Holdout incrementality - control subjects prove lift, the ledger proves why |
Basis of comparison ▸
Incumbent cells summarize each vendor's public positioning as of mid-2026; specifics vary by contract. The claim is categorical, and fair: weather-only signal coverage, manual rule authoring, and no tamper-evident decision governance anywhere in the category.
The platform in one line
The India wedge
Three high-attention signals; zero incumbent coverage.
Cricket - the national moment
Hundreds of millions watch one chase at once - no incumbent trigger platform models it.
AQI - India's context signal
Delhi past AQI 300 shifts health, mood and commerce within the hour - read live, per city.
Monsoon - the unlocalized playbook
A four-month, city-by-city season - per-city rain rules with hysteresis, not snow-day playbooks.
The wedge, in full ▸
Cricket - the national moment. A boundary burst in a chase is watched by hundreds of millions at once, and no Western trigger platform even models it. Kairos treats an India win-probability spike as just another signal - with a market-crash suppression rule so celebration is never tone-deaf.
AQI - India's context signal. When Delhi crosses AQI 300, health, mood and commerce all change within the hour. Kairos reads live air quality per city and lets brands respond responsibly - or go quiet - under the same governed rules as every other signal.
Monsoon - the unlocalized playbook. Global weather-marketing tools are built for snow days and heatwaves, not a four-month monsoon that arrives city by city. Kairos runs per-city precipitation rules with hysteresis, so creatives don't flip-flap at the edge of a drizzle.
Why now
The category is old; the unlock is new.
Why now, in full ▸
Agents crossed the authoring threshold. LLMs can now reliably draft a constrained rule AST from a business goal - the expensive part of trigger marketing (authoring and upkeep) just collapsed. The new bottleneck is trust, which is why Kairos is governance-first: propose → approve → evidence.
Signals are commoditized; decisions are not. Keyless public APIs and cheap event feeds mean data access is no longer the moat. The defensible layer is the one that turns raw signals into governed, deterministic, provable decisions - and that layer barely exists today.
India's moments are unserved. Cricket, AQI and monsoon are among the highest-attention context signals anywhere, in a market where media efficiency matters most - and every incumbent in the category is weather-only and built elsewhere.
What's real in this demo
LIVE·open-meteoweather & AQI, per citySIMcricket · traffic · marketsSCRIPTEDno-key agent fallbackIf it isn't chipped LIVE, it isn't live.
Provenance, in full ▸
Weather and air quality are fetched live, per city, from Open-Meteo's keyless public APIs. Real numbers, real timestamps.
Cricket, traffic and market feeds are deterministic, clearly labeled simulations (a scripted T20 timeline, an IST rush-hour curve, a seeded NIFTY-shaped walk). Each is swappable for a licensed feed without touching the rule layer.
When no LLM key is configured, the Studio agent falls back to a deterministic non-LLM planner - labeled SCRIPTED, never passed off as a model.
The engine, the approvals workflow and the hash-chained evidence ledger are fully real - deterministic TypeScript plus append-only Postgres, verifiable from the Ops surface. The “Frosty” brand, its campaign and all spend figures are fictional demo content. Every datum on every surface carries one of the chips above; if it isn't chipped LIVE, it isn't live.