parsance_

01 · Early access

AI agents are buying.Can they read your store?

The check will make your store readable for ChatGPT, Perplexity and Gemini, and keep it readable: findings, fixes, ongoing maintenance.

Audit 07/2026 · 0 of 20 German stores fully readable for AI ↓

B2B offer · Emails only for scan and launch · Privacy policy

02 · Why now

Product search is moving into chat

+393 %

Q1 2025100
Q1 2026493

Index · Q1 2025 = 100

AI traffic to online retail within one year (US retail, Adobe, Q1 2026).

+42 %

2025-38 %
2026+42 %

Shoppers arriving from AI recommendations convert better than other traffic (US retail, Adobe). A year earlier they trailed by 38 %.

Protocol status

OpenAI and Stripe defined ACP as a commerce protocol; Google and Shopify defined UCP as a second one. Shopify and Shopware ship first feed features. What goes into those feeds is still on you.

ACP · OpenAI + StripeUCP · Google + Shopify

03 · Sample · 2026-07-02

We audited 20 German stores. By hand.

Established German stores across categories, checked against the signals an AI agent reads. Those signals became the six criteria below. A sample of 20, not representative, but clear.

0/20
fully readable for AI agents
0/20
shipping & returns readable for agents
9/20
no machine-readable catalog
2/20
effectively block AI crawlers
Store 01–20present
Access
18/20
Catalog
11/20
Schema
9/20
GTIN
5/20
Rating
3/20
Stock
10/20
Shipping
0/20
llms.txt
5/20
presentmissingunreadable (catalog closed)

Protocol 2026-07-02 · n = 20 · manual · anonymized

We don't name names. Whether your store shares these gaps is what the check will tell you at launch.

04 · The agent's view

What an agent sees of your store

Your customers get the full page: images, reviews, price. An agent only reads the data fields underneath. Three finding classes from the audit plus the target state, switchable.

What the shopper sees

Classic Tee

Size M · White

€89.00Add to cart

What the agent reads

JSON-LD · schema.org/Product

"name""Classic Tee"ok

"price""89.00 EUR"ok

"availability""InStock"ok

"gtin"—MISSING

"shippingDetails"—MISSING

"returnPolicy"—MISSING

"aggregateRating"—MISSING

Readable: 3 of 7 fields

05 · The plan

The operating layer for your store's AI visibility

Visibility dashboards show you the problem. Fixing it has been your job so far. This layer will take over both and keep the state. Fully autonomous, with approvals where you want them. This is what it looks like, with sample data.

01 · Check

Every finding with evidence

30 checks in six categories, from bot access to return details. Every row carries its measured value, not just a verdict. Whatever fails points straight to its fix.

02 · Repair

Approve or auto

Every fix with a before/after diff. Approval one by one, per 1-click, or permanently on auto, set per fix type. Every application stays reversible.

03 · Maintain

The ongoing operation, provable

Product data drifts with every theme update and every new product. The layer will detect drift, heal it within minutes and write every step to the log. The ACP feed as a service, not a file.

04 · Report

What it returned

Every month a report with score history, fix log and your store's AI revenue. As a web view and as a PDF, automatically on the 1st.

06 · The audit matrix

Six criteria. Thirty checks.

These six criteria are what the layer works by. Each one bundles several individual checks, 30 are planned for the scanner. And every criterion comes with a fix. That, too, comes from the layer.

Six criteria · 30 individual checksproduceScore 0–100Fix plan

  • robots.txt is your store's front door. If it says “Disallow”, the agent turns around before it has seen a single product.

    Planned individual checks

    • GPTBot / OAI-SearchBot allowed
    • PerplexityBot allowed
    • ClaudeBot allowed
    • Bingbot + Google-Extended allowed
    • Bot protection tested with real bot clients

    Fix · robots.txt clearance, 1 click with preview

  • Under every product page sits a record only machines read. If it has gaps, the agent is missing exactly the field your customer asks about.

    Planned individual checks

    • JSON-LD Product present
    • Record technically valid
    • Price + currency
    • Availability
    • GTIN / MPN + brand
    • Images in the record
    • Variants mapped correctly
    • Merchant identity (Organization)

    Fix · complete JSON-LD, automatic and theme-update-proof

  • “What does shipping cost? How do I return this?” are buying questions. If the answers are not in the record, the agent recommends the store that answers them.

    Planned individual checks

    • Shipping cost on record
    • Delivery time on record
    • Return window machine-readable
    • Return method machine-readable

    Fix · one-time wizard, then served permanently

  • Your stars are visible on the page. Without markup they still do not exist for the agent — 128 reviews then count for exactly nothing.

    Planned individual checks

    • aggregateRating present
    • Value + count consistent with the page
    • Markup technically valid

    Fix · reads your review app, automatic

  • Agents do not read page by page — they pull the catalog as a record. If there is none, your assortment stays invisible to them.

    Planned individual checks

    • Open product catalog (products.json)
    • Product sitemap current
    • ACP feed present + valid
    • Price + currency consistent
    • Stock per variant
    • In the Bing index (basis of ChatGPT search)

    Fix · ACP feed generated + kept in sync

  • Many stores assemble their pages in the browser. Agents often do not render — they see the empty version. And llms.txt is the guide that shows them the way.

    Planned individual checks

    • Core content readable without JavaScript
    • llms.txt present
    • llms.txt maintained, not just the default
    • Response times + status codes for bots

    Fix · llms.txt generated + maintained

07 · Planned pricing

Maintenance is the product. The scan stays free.

Provisional prices, net plus VAT. Not yet bookable, binding at launch.

Scan

€0

Stage 1 · Findings

  • Score and report
  • Stays free

Basic

€29/mo

Stage 2 · 1-click fixes

Everything in Scan, plus one-click fixes:

  • Schema
  • GTIN markup
  • Feed generation

Maintenance

The product

€79/mo

Stage 3 · Auto

Everything in Basic, plus:

  • Continuous fixes
  • Feed operation
  • Drift monitoring
  • Monthly report with AI visits and revenue

Agency

€199/mo

Stage 4 · Auto × shops

Everything in Maintenance, plus:

  • Multiple stores
  • White-label reports

Details on the pricing page

08 · Questions

The obvious questions

  • The audit methodology exists (six criteria, applied to 20 stores), so does the product interface you see above — running on sample data — and this waitlist. The scanner and the automation behind it are in development. The rest arrives at launch.

  • We'll name a date once we can keep it. The waitlist hears it first.

  • The feed carries whatever your product data contains. If GTINs, shipping details or review markup are missing, it carries exactly those gaps. Stores with active platform features were incomplete in our audit too.

  • Once, yes. But product data drifts: new products, theme updates, price changes. And what the engines read keeps changing. That is what the maintenance tier is for: holding the state, not just creating it.

  • Store URL and email, nothing more. Stored in Frankfurt (EU), used only for the waitlist and launch updates. Details in the privacy policy.

  • Merchants running their own online store, and agencies managing stores. A business-only offer (B2B), consumers aren't the audience.

  • The scan stays free. The planned paid tiers are listed above; they become binding at launch.

09 · Early access

The waitlist is open

Signing up takes a minute. The earlier you join the list, the sooner your store gets the fixes at launch. Until then, the free scan measures the six criteria above for your store. And the views under “The plan” show what the check makes of them.

B2B offer. We'll only email you about your scan and the launch. Details: Privacy policy