commerce intelligence infrastructure

One normalized product brain
behind every store you build.

The Aether ingests product links, supplier feeds, affiliate URLs and API data into a single normalized system — then cleans it, matches the same product across merchants, scores every offer, and pushes only approved products into your sites, SEO pages, dashboards, and AI agent workflows.

Product Brain Core
sources in → normalize · match · score · gate → approved out
01
Ingest
links · feeds · APIs
02
Normalize
clean & structure
03
Match
same product, all merchants
04
Score
7-signal model
05
Gate
approval + compliance
06
Publish
sites · agents · pages
What it is

A reusable backend commerce layer,
not another store builder.

Most tools rebuild ingestion, link tracking, and scoring for every new site. The Aether does it once — as modular MCP servers and a clean data model — so every AI-built business you launch plugs into the same intelligence.

Ingest anything

Product links, supplier feeds, affiliate URLs, user-added products, and partner APIs flow into one staging system. Licensed feeds and official APIs — not scraping.

Normalize & match

Rules-first cleaning with LLM fallback structures titles, specs and identifiers — then matches the same product across merchants using GTIN/MPN-aware identity resolution.

Score every offer

A configurable 100-point model rates reviews, price, margin, SEO potential, trend potential, risk and availability — so the best offer surfaces automatically.

Gate with approval

Compliance filters and approval gates mean nothing flagged ever goes live. approve is unreachable while any compliance flag is open.

Trackable redirects

Every approved product gets a clean /go/<slug> redirect with link-health checks and attribution baked in.

Publish everywhere

Push approved products into websites, SEO and comparison pages, dashboards, and AI agent workflows — all from the same source of truth.

Architecture

Modular MCP servers over a shared spine.

The core IP is the product / offer / link / score data model. Everything else is a small, composable server — so future apps connect to Shopify, Stripe, analytics, affiliate networks, supplier APIs and custom storefronts without touching the core.

offer_matching
Identity resolution across merchants
product_scoring
Configurable 100-pt signal model
compliance_filter
Rule-driven approval blocking
redirect_tracking
/go/<slug> + attribution
affiliate_link
Network-aware link builder
link_health
Nightly dead-link checks
price_tracking
Price & availability history
agent_permissions
Scoped API access for agents
pipeline/run_gate.py
# LOAD → MATCH → COMPLY → SCORE → PUBLISH
from aether import pipeline

result = pipeline.run_gate(
source="supplier_feed",
thresholds={"approve": 70, "hold": 50},
)

# every flagged item is held, never approved
assert result.flagged_approved == 0

pipeline.publish(result.approved)   # → sites · pages · agents
Who it's for

One layer, three audiences.

Developers

Build on the spine

  • Python + FastMCP + FastAPI, Postgres/Supabase backed
  • Single DB contact point via a clean store adapter
  • Scoped agent permissions and audit logs on every action
  • Drop-in MCP servers — no core rewrites
Investors

Own the data model

  • Defensible IP: normalized product / offer / link / score graph
  • Reusable across an unlimited portfolio of sites & stores
  • Licensed, compliant sourcing — not scraping risk
  • Single-tenant today, multi-tenant on the roadmap
Operators

Launch stores faster

  • Approved products flow straight into pages & comparisons
  • Automatic best-offer selection and price tracking
  • Link health checks keep catalogs from going stale
  • Human approval gates keep you in control
Trust by design

Approval gates aren't optional.

Every action has a user approval gate, audit log, API permission check, link-health check and compliance filter. The gate is independently verified — and proven on real data.

1,068
offers processed
800
product identities
479
approved & published
0
flagged items approved
User approval gate on every publish
Immutable audit log of all actions
Compliance filter blocks approval
Scoped API permissions per agent
Automated link-health monitoring
Independent gate verification
The intake engine

The File Brain feeds the spine.

Before offers are matched, scored, and gated, raw data has to become clean records. The File Brain is the ingestion layer: drop in feeds, files, or links and it scans, extracts, cleans, structures, analyzes, and validates them into the canonical model — raw kept separate from analysis, code never executed.

Open the full File Brain dashboard ↗
Get started

Power every store you build from one commerce brain.

Early access for developers, design partners, and investors exploring the The Aether infrastructure layer.