laravel-ai-price-intelligence

laravel-ai-price-intelligence — enterprise competitor price & product intelligence for Laravel

Latest Version on Packagist
PHP 8.3+
Laravel 11 | 12 | 13
License Apache-2.0

laravel-ai-price-intelligence is the competitor price & product intelligence engine laravel/ai is missing — self-hosted in your own stack.
Send it your SKUs and the countries to watch; it discovers competitor listings, matches the right
product with an AI cascade, scrapes price/stock/content, normalizes multi-currency time-series, runs
forecasts and anomaly detection, and emits signed webhooks — open-source, EU AI Act-ready, advisory
by design. Your ecommerce keeps every pricing decision.

New here? Read this page top to bottom

In five minutes you’ll know exactly what this package is, the problem it solves, why it beats every
black-box SaaS or DIY scraper alternative, and where to click next. Every other page goes deeper —
this one gives you the whole picture.

What it is — in one minute

Ecommerce teams pay expensive, black-box SaaS (Netrivals, Competitoor) to watch competitor prices,
assortments and content — renting back their own market data and never owning it. laravel-ai-price-intelligence
brings that capability into your Laravel stack, open-source and self-hostable.

You send the SKUs, the countries to watch, and either the competitor URLs or let the
AI search discover them. From there a single pipeline does the rest:

  • Match the right competitor product — a cascade of GTIN → MPN → normalized name → embedding →
    optional vision LLM, each result carrying a confidence band with a human-review queue.
  • Scrape & normalize — JSON-LD/OpenGraph extraction, generic HTTP + Browsershot and dedicated
    marketplace adapters (Amazon, eBay, Google Shopping, Farfetch, Idealo, Trovaprezzi), with
    multi-currency FX into one base currency and stored time-series.
  • Reason & notify — forecasting, anomaly detection and GDPR-safe review sentiment, then
    HMAC-signed webhooks + Laravel events (price drop/raise, undercut, stock-out) your app reacts to.

In one line: the competitor-monitoring brick for Laravel — discover, match, scrape, forecast
and alert on competitor prices at ~500k-SKU scale, from inside your own app, while you keep the
pricing decisions.

The problem it solves

Every team running competitor monitoring hits the same wall: the data lives in someone else’s SaaS,
the matching is a black box, and you pay per SKU to rent your own market intelligence. Here is the
gap this package closes.

Without laravel-ai-price-intelligence With laravel-ai-price-intelligence
You rent a black-box SaaS (Netrivals, Competitoor) and never own your market data. A self-hosted, Apache-2.0 engine runs in your Laravel app, on your DB — you own every row.
Product matching is opaque: you can’t see why a competitor listing was paired with your SKU. A transparent matching cascade (GTIN → MPN → name → embedding → vision LLM) with a confidence band and a human-review queue (auto ≥85 · review 60–84 · reject <60).
One scraper breaks on a marketplace and the whole run dies. Marketplace adapters with API paths that fall back to scraping on any missing key or failure — a run never breaks.
Prices arrive in mixed currencies with no history. Multi-currency FX normalization to a base currency and time-series observations for price, stock and promos.
“Insights” are static dashboards; no forecast, no anomaly signal. Built-in forecasting, anomaly/price-error detection and GDPR-safe review sentiment, each flagged is_ai_generated.
Scraping 500k SKUs causes a thundering herd and OOM exports. Chunked jobs + adaptive backoff on dedicated queues, cursor pagination, exact SQL facets and streamed CSV export — engineered for ≈500k SKUs.
Compliance is an afterthought — robots.txt, PII, AI disclosure all on you. EU AI Act-ready by design: decision log, human-in-the-loop matching, robots.txt policy, PII redaction.

Who it’s for

Ecommerce & retail

Watch competitor prices, stock and assortment across countries, then feed signed webhooks into your dynamic pricing, MAP enforcement and merchandising — without handing your catalog to a SaaS.

Marketplaces & brands

Track buy-box and listings on Amazon, eBay, Google Shopping, Farfetch, Idealo and Trovaprezzi with dedicated adapters, plus visual matching for catalogs where GTINs are missing.

Pricing & category teams

A match-review queue, price-history charts, forecasts, anomalies and a weekly AI narrative — the analyst cockpit Netrivals/Competitoor don’t give you self-hosted.

Platform & compliance engineering

Multi-tenant (single-DB or DB-per-tenant), config-toggle everything, EU AI Act decision log and GDPR-safe pipelines — the governance layer to run competitor intelligence in-house.

Why it’s different — the moats

Most tools either rent you competitor data or make you build a scraper. This package does
the whole pipeline, self-hosted, and goes further than anything in the Laravel ecosystem.

Transparent matching cascade

GTIN → MPN → normalized name → embedding → optional vision LLM, every result carrying a confidence band. Auto-confirm ≥85, review 60–84, reject <60 — and you can see exactly which signal matched.

Visual matching with vision LLMs

When GTIN and MPN are missing, a vision-LLM match-judge compares product imagery — a moat neither Netrivals nor Competitoor offers.

Marketplace adapters that never break

Amazon (SP-API/Keepa), eBay (Browse API), Google Shopping (SerpApi), Farfetch (retailed/Apify) and generic JSON-LD — each defaults to scraping and falls back to scraping on any failure, so a run always completes.

Engineered for ≈500k SKUs

Cursor pagination (OFFSET-free), exact SQL facets, streamed CSV export (OOM-safe), daily aggregates + partition-ready time-series, and chunked jobs with adaptive backoff on dedicated queues.

AI signals, not just data

Statistical forecasting (OLS trend + confidence interval), detrended-residual anomaly + price-error detection, narrative, content-gap and promo analysis — pluggable and toggleable per feature.

EU AI Act-ready by design

Native disclosure (is_ai_generated), an Art. 12 decision log, human-in-the-loop matching, plus an optional bridge to padosoft/laravel-ai-act-compliance.

Boundary-respecting & advisory

It is the intelligence engine: it emits signals via HMAC-signed webhooks and events. The optional repricer is off by default and advisory-only — your ecommerce keeps every pricing decision.

Pluggable everywhere, sovereign-ready

Every stage is an Interface + Driver with null-object optional deps. Run the deterministic fake LLM/embeddings with zero keys, or switch to laravel/ai — including the EU-sovereign regolo provider.

Open-source, self-hosted, you own it

Apache-2.0, runs in your Laravel app and your DB. No per-SKU SaaS bill, no black box, no vendor lock-in.

See it: the intelligence cockpit

A companion web admin panel ships separately as
padosoft/laravel-ai-price-intelligence-admin
— a React 19 + Vite + TypeScript + Tailwind console driving every endpoint: dashboards, the
match-review queue, price-history charts, forecasts and anomalies, the weekly AI narrative,
assortment/content gaps, the repricer rule builder, alerts inbox, webhooks and an EU AI Act
compliance view. It consumes this package’s API directly — no mocks.

laravel-ai-price-intelligence — web admin panel dashboard

laravel-ai-price-intelligence vs. the alternatives

Capability laravel-ai-price-intelligence Manual monitoring SaaS price trackers (Netrivals / Competitoor) DIY scrapers
Price/stock across multiple countries ✅ ➖ ✅ ➖
Transparent AI matching cascade + confidence ✅ ❌ ➖ ❌
Visual matching (vision LLM) ✅ ❌ ❌ ❌
Marketplace adapters with scrape fallback ✅ ❌ ➖ ➖
Price forecasting + anomaly detection ✅ ❌ ❌ ❌
Engineered for ≈500k SKUs (cursor/facets/stream) ✅ ❌ ➖ ➖
EU AI Act-ready by design ✅ ➖ ❌ ❌
Self-hosted — you own the data ✅ ✅ ❌ ✅

Legend: ✅ built-in · ➖ partial / extra cost / manual · ❌ not available.

How it fits together

Your ecommerce sends a catalog and targets; the package discovers, matches, scrapes and normalizes
into a time-series, runs the AI layer, and emits signed webhooks your app reacts to.

flowchart LR A[Ecommerce catalog] --> B[Targets: SKU x country] B --> C[Discovery] C --> D[Matching cascade] D -->|low confidence| R[Human review queue] D --> E[Scraping + adapters] E --> F[Normalized observations] F --> G[AI layer: forecast / anomaly / sentiment] G --> H[Alerts + signed webhooks] H --> I[Host pricing logic]

The package is boundary-respecting: it produces intelligence, while the host application keeps the
final pricing, margin, order, checkout and inventory decisions.

Start in 30 seconds

  1. Install the package

    composer require padosoft/laravel-ai-price-intelligence
php artisan vendor:publish --tag=price-intelligence-config   # optional
php artisan migrate

    The default LLM/embeddings driver is fake — deterministic and offline — so the package works out
    of the box with no API keys and CI never makes a live call.

  2. Issue an API key and sync your catalog

    use Padosoft\PriceIntelligence\Models\ApiKey;

[$key, $plaintext] = ApiKey::issue($tenantId, 'ecommerce-sync', ['*']);
// store $plaintext securely — it is shown only once

Http::withHeaders(['X-Api-Key' => $plaintext])->post("$base/api/v1/catalog/products:bulk", [
    'products' => [[
        'external_id' => 'SKU-123', 'gtin' => '8001234567890',
        'brand' => 'Acme', 'model' => 'X1', 'name' => 'Acme X1 64GB',
        'our_price_cents' => 19900, 'currency' => 'EUR', 'base_country' => 'IT',
    ]],
]);

  3. Create a target and react to signals

    Http::withHeaders(['X-Api-Key' => $plaintext])->post("$base/api/v1/targets", [
    'product_external_id' => 'SKU-123', 'country' => 'IT', 'frequency' => 'daily',
]);

// Signed webhook listener in your app
use Padosoft\PriceIntelligence\Services\Webhooks\WebhookSigner;

Route::post('/webhooks/price-intel', function (Request $r) {
    abort_unless(WebhookSigner::verify(
        $r->getContent(), config('services.pi.secret'), $r->header(WebhookSigner::HEADER, '')
    ), 401);

    match ($r->input('event')) {
        'price.dropped', 'undercut.detected' => MarginOS::reevaluate($r->input('data')),
        default => null,
    };
});

→ Quickstart · → Installation · → Integrating an Ecommerce

Batteries included for AI-assisted development

This repo ships AI batteries — a CLAUDE.md working guide, an AGENTS.md workflow contract and
invocable .claude/skills/ encoding the TDD loop, the docmd docs-sync discipline and the
LESSON/PROGRESS logging rules. Open the package in Claude Code, Cursor, Copilot or Codex and your
agent already knows the house rules.

Where to go next

Quickstart

Install, sync a catalog and create your first monitoring target in minutes. Open →

Product Matching

The matching cascade, confidence bands and the human-review queue, explained. Read →

Scaling Large Catalogs

Cursor pagination, exact facets, streamed export and daily aggregates for ≈500k SKUs. Explore →

Package facts

Composer padosoft/laravel-ai-price-intelligence · PHP ^8.3 · Laravel 11 | 12 | 13 ·
LLM via the official laravel/ai SDK (deterministic fake default) · Apache-2.0 ·
GitHub · Packagist

next Quickstart