Retail & e-commerce · Local & private AI

Local and private AI for retail and e-commerce

Commerce AI work is repetitive and huge. Descriptions, attributes, support triage, review summarization. These are exactly the tasks where per-token pricing turns a useful tool into a line item nobody can defend at the end of the quarter. Self-hosting changes the arithmetic.

Start a buildIndustry overview
03Self-hosted inference · Retrieval on private data
Local & private AI03
Self-hosted inferenceRetrieval on private dataDe-identification
Retail & e-commerceIllustrative

Local & private AI models

Self-hosted inference, fine-tuning, retrieval on private data, and de-identification.

Local & private AI · Retail & e-commerceOne pairing of the matrix
Service
Local & private AI models
Industry
Retail & e-commerce
Covers
Jewelry & fine goods · Salon & beauty retail · DTC brands
Architecture
Unchanged from every other vertical — only the breaks differ
Engagement
A defined build, or run under management

What usually breaks

Before anything gets built.

The same failures recur across retail & e-commerce. Strygon maps the specific system before proposing anything, but these are the ones worth checking for.

Slow storefront

The store is slow enough that ranking and conversion both suffer, and every speed test is a lost sale.

Blind tracking

Attribution runs on client-side pixels that browsers now block, so no one trusts the numbers.

Manual orders

Custom orders and invoices run on emailed PDFs and memory. That works at ten orders and breaks at a hundred.

Recurring in retail & e-commerceObserved pattern · the specific system still gets mapped

The approach

What Strygon does about it.

Local & private AI models, aimed at those failures specifically rather than at a generic checklist.

01Local & private AI
Catalog work at catalog scaleLocal & private AI
01
Part of the scoped build

Catalog work at catalog scale

Descriptions, attributes, and categorization generated in bulk without watching a meter.

02Local & private AI
Triage before a human sees itLocal & private AI
02
Part of the scoped build

Triage before a human sees it

Incoming support messages classified and routed on the business’s own infrastructure.

03Local & private AI
Customer data stays putLocal & private AI
03
Part of the scoped build

Customer data stays put

Order history and support conversations never reach a third-party model.

The approach for this pairingScoped in writing before work starts

What gets built

Concrete deliverables.

Everything in local & private ai models, applied to how retail & e-commerce actually runs.

All of local & private aiIndustry overview
Self-hosted / on-prem LLM and vision deployment
Fine-tuning, evaluation, and quantization pipelines
Retrieval (RAG) over private data
On-device and edge inference
De-identification and policy guardrails
In scopeScoped in writing before work starts

Also for this industry

Other parts of the same system.

Web & commerce
Web & commerce04
Next.js buildsCommerce
Applied hereIllustrative

Web & commerce

A slow storefront loses the sale and the ranking at once, and the two compound. In commerce, speed is not a finishing item; it is most of the conversion rate.

Read the web & commerce read
Acquisition
Acquisition05
Paid search & socialLanding pages
Applied hereIllustrative

Acquisition

Browser-side tracking under-reports badly enough that platform ROAS and actual revenue disagree by a margin wide enough to make the wrong call look correct.

Read the acquisition read
CRM & automation
CRM & automation03
Pipeline architectureSpeed-to-lead
Applied hereIllustrative

CRM & automation infrastructure

Commerce back-offices break on the boring half: orders arriving faster than anyone can process, and customers never contacted again after the first purchase.

Read the crm & automation read
Custom software
Custom software02
Internal toolsData pipelines
Applied hereIllustrative

Custom software

Shopify runs the storefront. It does not run purchasing, or wholesale pricing, or the three systems that disagree about stock.

Read the custom software read

Start

Start with what’s broken.

Send the situation in a paragraph. Strygon comes back with a read on what’s likely wrong and what it would take to fix, before anyone talks about price.

Most builds start withleads dying in an inbox., three half-finished pipelines., follow-up nobody owns., numbers that never agree., four vendors blaming each other.

Or email hello@strygon.com

What to send
A paragraph. What broke, and where it shows up.
What comes back
A read on what is likely wrong and what fixing it takes.
Price
The last conversation, not the first