Todayagenticux.de
The agent gets stuck.
Product facts are hard to extract.
Policies contradict the cart.
Variants and bundles lack decision rules.
Checkout limits appear too late.
AI agents buy differently than people.
I check whether AI assistants recommend your shop and whether agents can select products and prepare a purchase. You get documented findings and clear changes for your team.
Stephan Lucka · 10+ years product & ecommerce UX
From the studyAugust 2026 · German report
Which shops does an AI assistant name?
- buying questions
- 115
- answers
- 345
- retailers named
- 837
Agent-readyagenticux.de
The path is verifiable.
Evidence is easy to cite.
Rules stay consistent.
Comparisons show clear criteria.
The purchase can be completed.
What changes when the buying path is clear.
Be found. Make buying possible.
Two checks answer two different questions. Start with the one that matters to your shop.
01 · AI visibility
Does your shop get recommended?
I ask relevant buying questions and check whether your shop appears, which other retailers are named, and which sources support the answer.
You get: questions, answers, sources, and specific gaps.
See the visibility study02 · Buying flow
Can the agent move forward?
I give an agent a buying task: find a product, check variants, and prepare the cart. Every unclear or blocked step is recorded.
You get: the test record, observed drop-offs, and changes for your team.
Review my buying flowAgent Journey Mapping01
See exactly where AI agents drop off in your shop.
I test how AI agents move through your shop from search to checkout, then turn the weak spots into a ranked fix list: missing context, missing proof, broken rules, and blocked completion.
Tactic
We run one real buying task and record every point where the agent guesses, stops, or needs help. Findings are ranked by impact.
Outputs
Agent Intent Mapping
Shop Crawl Diagnostics
Decision Path Modeling
Task Completion Analysis
Policy Visibility Review
Agent Journey Playbooks
"Agents rarely fail everywhere. Usually one path breaks first: search, product proof, policy trust, or checkout."
Audit note
Agent Journey Mapping
Structured Commerce UX02
Product pages where AI agents don't have to guess.
I turn product, category, comparison, and policy pages into clear evidence surfaces so agents can extract facts directly instead of guessing from persuasive copy.
Tactic
We check whether product facts, comparison criteria, and policies are easy to find, consistent, and clear enough to use.
Outputs
Product Evidence Architecture
Comparison UX
Schema & Content Requirements
Policy Surface Design
Assistant-Readable Copy
AI Search Readiness
"Good product copy is not always good agent input. Claims need proof, structure, and clear decision rules."
Audit note
Structured Commerce UX
Agent Checkout Readiness03
Checkouts that don't strand AI agents.
I find the moments where agents still need a human: shipping, returns, bundles, accounts, payment, and product rules. Then we make those steps explicit enough to complete.
Tactic
We test shipping, returns, accounts, payment methods, and exceptions in the order an agent needs them to complete a purchase.
Outputs
Checkout Friction Audit
Cart & Account Flow Review
Shipping Rule Clarity
Returns & Warranty UX
Payment Path Diagnostics
Agent Completion Dashboards
"Checkout fails when rules appear too late: availability, accounts, payment options, return windows, or bundle limits."
Audit note
Agent Checkout Readiness
Trust & Evaluation Systems04
Trust signals ChatGPT & co. will recommend.
I make reviews, specs, guarantees, pricing, availability, and support claims verifiable so agents can recommend you for the right reasons.
Tactic
We compare the product page, cart, policies, and FAQ. Every contradiction becomes a clear task for the responsible team.
Outputs
Trust Signal Inventory
Review & Proof Architecture
Availability Messaging
Guarantee Clarity
Support Path Design
Agent Evaluation Tests
"Agents need proof they can verify. Reviews, specs, guarantees, and policies have to tell the same story."
Audit note
Trust & Evaluation Systems
Test. Find. Repair.
One real buying path shows where the agent stops. That exact point gets fixed first.
I test one real buying task across search, PDPs, policies, and checkout.
- 01Pick one buying task that maps to real revenue
- 02Run it end to end, the way an agent would
- 03Record every point where it stalls or guesses
The map
The same shop, read two ways.
A shop can be clear to people while agents miss essential facts. The audit checks both views: what can be found, what can be verified, and where the purchase stops.
Map my shopHuman UXAgent UX
Example profile. Your shop is assessed in the audit.
Why agenticux.de exists.
agenticux.deThe idea
Why agenticux.de exists.
I am Stephan Lucka. For more than ten years I have worked on product and ecommerce UX. agenticux.de applies that work to AI agents: I check whether a shop gives them the facts, rules, and next steps they need to justify a recommendation and complete a purchase.
- One real buying path.
- One clear breakpoint.
- Fixes your team can check.
Typical breakpoints
"Agents do not drop off because the page looks bad."
They drop off when product facts, policies, comparisons, or checkout rules stop lining up.
Agentic UX checkTypical breakpoint
Our research
What the tests show.
AI answers · 5 August 2026
Many retailers appear only once.
423of 837 retailers named only once
115 buying questions produced 345 answers. The study shows how the named retailers vary between responses.
One assistant, one measurement day. The test measured mentions, not purchases.
Study, method, and results (German)Product data · 7 July 2026
Access determines what can be read.
9of 100 fashion shops blocked retrieval
87 of the 91 reachable shops included a price in their HTML. Further checks examined product data and cart signals.
The test examined HTML and files. It did not attempt a complete purchase.
Read the technical study (German)Check first. Then build.
Start with one path.
The first 30 minutes clarify your goal, the scope, and next steps.
Stephan Lucka
Agentic UX Consultant
10 daysto clarity
In 10 days you get the findings for one agreed buying task: observed drop-offs, their causes, and the three changes to make first.
Recorded test with the task and agents used
Findings with evidence, cause, and a proposed change
Visibility checks across major AI assistants
Schema audit for Product, Offer & Policy
30-day check-in after going live
Audit fee fully credited toward a Sprint
FAQ
I test one buying task from an AI agent's perspective and show where the purchase becomes uncertain.
Insights.
Field notes from working with AI agents in real shops, and what you can do before your competitors read it.
Agentic Commerce15 min read
May 5, 2026
Agentic UX: why shops need to be readable for agents
A practical, SEO and AEO optimized guide to agentic UX for ecommerce teams: what changes when AI agents parse, compare, trust, and complete purchases.
Read article
AEO15 min read
May 5, 2026
Answer Engine Optimization for ecommerce
A practical answer engine optimization ecommerce guide: how shops structure pages, evidence, schema, and answers so AI systems can cite them.
Read article
Let’s find the first drop-off.
Discuss your shop and goal
Choose one task to review
Agree on scope and fee before starting
Send me your shop URL and your question. In the first call, we will decide which review would help you.
Stephan LuckaYour contact for the review