AI agenti, kteří se starají o váš e-shop.

Builds a session-level GA4 checkout funnel, tests which step worsened against the store's own baseline, localises it by device and channel, and proposes one fix with a way to confirm it.

Calculates break-even ROAS, PNO/COS and target ROAS per category and campaign mix from contribution margin (COGS, fees, fulfilment, returns), on the same VAT and shipping basis as your Google Ads conversion value, with all arithmetic shown.

Turns a Google Merchant Center issue export into a fix plan sorted by lost revenue: root cause per issue (feed, site, account setting or policy), owner, and whether a review is needed. For store owners and PPC specialists.

Translates a catalog export in resumable batches with a mandatory glossary, protected brand and placeholder tokens, locale number formats and per-row QA flags, so content managers know exactly which rows need a human.

Turns a Search Console performance export into three ranked lists for a store: pages losing clicks with the cause split into demand, ranking and CTR, under-clicked queries against the site's own CTR curve, and URLs competing for one query.

Decides facet by facet which filter URLs on an online store should be indexed, consolidated, noindexed or blocked from crawling, using crawl, log and Search Console data, and lists crawl traps with a safe rollout order.

Audits the checkout from cart to withdrawal flow using GA4 step data and a mobile guest walkthrough, ranks friction by severity and reach, and lists EU must-fixes such as the order button and pre-ticked extras.

Maps a store catalog to Heureka.cz/.sk and Zboží.cz XML and idealo offer data element by element, including DELIVERY_DATE, carrier IDs and prior-price rules, and reports every row that would be rejected or badly paired. For Czech and EU e-shops.

Plans SKU-level clearance markdowns from stock, sell rate and exit date, stops at a VAT-correct floor, and shows the EU Omnibus 30-day prior price and correctly based discount percentage for every step, including progressive-markdown rules.

Writes product descriptions and spec lists from supplier data in the store's brand voice, ordering facts by what decides the purchase for each product type, listing missing specs and flagging unsupported or EU-restricted claims.

Turns a master product record into Amazon, Allegro and Kaufland listing drafts: leaf category, title within the channel limit, mapped attributes, GPSR fields and a blocking-gap list, for sellers expanding to marketplaces.

Mines Shopping and Performance Max search terms into collision-tested negatives (match type and level), watch terms and winners to protect, using n-grams and a sample-size rule instead of guesswork; for PPC managers.

Rewrites Google Shopping feed titles and descriptions from verified attributes, policy-clean and marked as AI-generated via structured_title, and proposes missing attribute fills in a reviewable CSV. For store owners and feed managers.

Drafts where-is-my-order replies from order and carrier data, with per-parcel status, only carrier-backed dates, and accurate EU delivery, withdrawal and warranty statements for a human agent to send.

Extracts colour, size, material, gender and other attributes from titles, variants and descriptions into Merchant Center-ready values, with source, evidence and confidence per value and a review queue for conflicts.

Designs a three-mail abandoned cart flow for an ESP with an EU consent check, exit and frequency rules, a margin-tested discount only in the last mail, and a holdout to measure real recovery.

Audits every GTIN, MPN, brand and identifier_exists value in a product feed by GS1 rule (length, check digit, restricted prefixes, variant uniqueness) and lists the fix per row, without ever inventing a number. For feed and catalog managers.

Writes the title, meta, short intro, buying guide and evidence-based FAQ for a store category page from the category's dated product export and Search Console queries, with every number traceable and unsupported claims flagged.

Reviews a product page template against the shopper's real purchase questions and EU price, review and safety-info rules, and returns ranked test hypotheses with metrics for store owners and CRO specialists.

Plans a post-delivery review request flow that asks every buyer the same way, with timing by product type, no gating, and incentive rules checked against Google, Trustpilot and EU review law.

Audits GA4 ecommerce events against Google's schema (currency with value, transaction_id, items, value = price x quantity) and matches purchases to backend orders, returning a prioritised fix list for developers.

Compares your offers with competitor prices you supply, matched on GTIN and on total price with shipping, flags where you are out of the market or leaving margin, and suggests the smallest price move above your VAT-correct floor.

Reconciles backend, GA4 and Google Ads revenue with an order-level join and a line-by-line bridge (VAT and shipping, time zones, consent, gateway loss, test orders, click vs conversion date) for store owners and analysts.

Codes every return to a root cause with evidence, separates withdrawals from warranty and transit damage, flags SKUs with statistically real return problems, and routes each fix to its owner.