Be the answer AI assistants give, and the brand they buy.

AI assistants increasingly do the shopping. We manage your answers, product data, agent channels and marketing against your revenue and margin.

Shoppers now ask an assistant before they search

More shoppers ask ChatGPT, Gemini, Perplexity or Copilot what to buy. The assistant reads product data, reviews and web pages, then names a few options. Missing or poorly described brands are not in the answer, and few marketing teams know how often they are cited.

In January 2026 Google announced its Universal Commerce Protocol for agent purchases, first with eligible US merchants. In March 2026 OpenAI dropped Instant Checkout; merchants send feeds and keep their own checkout. OTTO opened an agent endpoint. So readiness matters more than any single bet.

Every marketplace, social platform, retail media network and assistant wants its own feed, creative and budget. Done by hand, the long-term target gets lost. AI can now run it all, if it is tied to the numbers that matter: margin, customer value, return on spend.

What we build

  • Answer engine visibility

    We test your buyers' questions in ChatGPT and Gemini, track where you are cited, and fix what assistants rely on.

  • Agent-readable product data

    Your catalogue cleaned and enriched in your PIM, then published as feeds to Google, ChatGPT shopping and your marketplaces.

  • Agent endpoints and checkout

    Catalogue, prices and stock opened to AI agents through MCP, with checkout ready for new protocols as they launch.

  • Marketing automated across platforms

    Agents create assets for each channel, launch campaigns and adjust bids within your limits. Your team approves brand-sensitive work.

  • Goals managed against ROI

    You set revenue, margin and return targets. Agents shift budget between channels and report weekly on what moved the numbers.

  • AI traffic measurement

    AI-referred visits, orders and revenue tracked as their own channel, so you can compare them with paid search and marketplaces.

How it works

One loop runs continuously, from what assistants say about you to your revenue figures.

  1. Measure how assistants see you

    Each week agents ask the major assistants your buyers' questions and rank every gap by revenue at stake.

  2. Fix the data at source

    Each gap is fixed where the fact lives (PIM, shop or CMS), reviewed, then published to every feed.

  3. Open the agent channels

    Feeds, endpoints and checkout go live where assistants support them. By default, shoppers finish in your own checkout.

  4. Run marketing towards the goal

    Agents run campaigns and move budget to what returns most, within spending limits and brand rules you approved.

  5. Report and reallocate

    Weekly reports show citation share, AI revenue and return against target. Agents propose shifts; larger ones need approval.

How we roll it out

  1. Discovery

    3 weeks

    We measure how assistants answer your buyers today, audit product data and feeds, and agree targets and the baseline.

    • AI visibility baseline
    • Product data and feed audit
    • Targets and ROI model
  2. Build

    8 to 10 weeks

    We fix data gaps, publish feeds and set up marketing agents with spend limits. One market and category launch first.

    • Live feeds and agent endpoint
    • Marketing agents in your accounts
    • AI traffic reporting
  3. Run

    Ongoing

    We run the loop with your team, add markets and new channels, and review results against targets monthly.

    • Weekly performance report
    • Monthly ROI review
    • Channel and protocol updates

In production elsewhere

Public examples from other organisations, not unus clients, showing what retailers already do as AI changes shopping.

Public case studyOTTOOnline retail · Germany · 2026 Opened an MCP endpoint and a ChatGPT plugin 19 million+ products via @OTTO
The situation
OTTO sells a broad range online. Its voice shopping assistant, live since March 2026 and built on Google technology, delivers up to 37% higher order values than classic search, OTTO says. Next: serving shoppers who start in external AI assistants.
What they deployed
With Otto Group partner one.O, OTTO built an MCP endpoint for AI agents. First use: a ChatGPT plugin where @OTTO gives advice on over 19 million products. Checkout stays on OTTO's site. It has tested ChatGPT ads since September 2026.
The outcome
OTTO calls the plugin its first practical use case for agentic readiness and keeps checkout on its own platform. It describes a future in which agents check stock and prices in real time, place orders and adjust invoices in the ERP system.

A standard agent interface on your own catalogue lets you learn in the new channels now, without handing over checkout or the customer.

Source: OTTO newsroom, "OTTO: Mit ChatGPT-Plugin und -Ads zur Agentic Readiness", October 2026
Public case studyZalandoOnline fashion retail · Germany · 2025 Campaign imagery produced with AI in days, not weeks 90% lower imagery costs
The situation
Zalando sells branded clothes, shoes and accessories in 25 European markets. Trends now spread and fade fast on social media, but producing imagery for its app and website took six to eight weeks. Campaigns could not keep pace.
What they deployed
Zalando produces app and website imagery with generative AI, fast enough for short-lived trends. Around 70% of its editorial campaign images in Q4 2024 were AI-generated. It is also building AI-generated digital twins of models for product pages.
The outcome
Zalando says generative AI cuts the time to produce imagery from six to eight weeks to around three to four days, and reduces costs by 90%. It adds that the AI content drives greater customer engagement.

Automated creative pays off when it makes marketing faster and more relevant to customers, not when it simply produces more of the same.

Source: Reuters, "Zalando uses AI to speed up marketing campaigns, cut costs", May 2025
Public case studyUlta BeautyBeauty retail · United States · 2026 Optimised for chatbots, kept checkout on its own site Double the conversion and intent
The situation
Shoppers increasingly ask ChatGPT and Gemini for recommendations. Chatbots answer detailed questions instead of ranking pages on keywords, so retailers rethink product descriptions. Ulta also wants purchases on its own site, where it collects data for future sales and loyalty.
What they deployed
Like Walmart and Wayfair, Ulta reworked its site for chatbot searches. With Google, it is bringing carts and Ulta Beauty Rewards into Gemini. Its head of digital and e-commerce prefers sales on Ulta's site: other platforms always cost something.
The outcome
Ulta says it is seeing double the conversion and intent from shoppers who find its products through Gemini and ChatGPT.

Be present and accurate in the assistants, but design the journey so the sale, the data and the loyalty relationship come back to you.

Source: Reuters, "Retailers tap AI shopping traffic but fight to keep customer data", August 2026

These are public deployments by other organisations, summarised from the sources linked. They are not unus clients.

Built for European rules

Customer data stays yours

Feeds carry product data only. Customer data stays in your systems under GDPR and nDSG; agents run in the EU or Switzerland.

Transparency when AI talks

Our assistants and agents tell customers they are AI (EU AI Act, Article 50) and hand over to staff on request.

Accurate claims, clear advertising

Product facts come only from approved PIM and CMS data. Paid placements are reported separately from organic visibility; new claims need approval.

Human control of spend

Agents work within budgets and brand rules you set. Bigger changes wait for sign-off; the log shows who or what did what.

Questions buyers ask

Is AI-assisted shopping big enough to matter for us yet?

It depends on your category and markets, so we measure it before you invest. Public data points in one direction: Adobe Analytics, reported by Reuters in August 2026, found that AI-referred US shoppers generated 41% higher revenue per visit than those from traditional channels. Discovery gives you your own baseline: how often you are cited, what assistants say, and how much traffic and revenue they send today.

Can customers buy from us inside ChatGPT or Gemini?

Not everywhere, and the options change often. Google has announced checkout in AI Mode and Gemini through its Universal Commerce Protocol, starting with eligible US retailers. OpenAI moved away from its Instant Checkout in March 2026 and now lets merchants keep their own checkout. We make your data and checkout ready for these routes, and switch them on where they are available and make commercial sense.

Do we lose the customer relationship to the AI platforms?

Not if the journey is designed for it. Many retailers, including OTTO and Ulta Beauty above, use assistants for discovery and route the purchase to their own site or app, where they keep customer data and loyalty. We build it that way by default. If you choose in-assistant checkout for a channel, we connect it to your payment provider and loyalty programme so orders still land in your systems.

Can we pay to be recommended?

Organic answers cannot be bought. They depend on the data, content and reviews the assistant can find and trust, which is what answer engine optimisation improves. Some assistants now sell ad placements in certain markets; OTTO, for example, has tested ads in ChatGPT since September 2026. We treat paid placements as a separate channel with its own budget and reporting.

Which systems do you need access to?

Your PIM or product catalogue, the shop system, stock and price data from the ERP, your analytics and CRM, and the ad and marketplace accounts we will manage. We work inside your accounts and environments, with access scoped to each task. You also name one owner on your side who sets targets and signs off on changes above agreed limits.

How do you measure return on investment?

We agree the targets and baseline in Discovery: for example AI-referred revenue, citation share for priority questions, return on ad spend and contribution margin by channel. AI-referred traffic is tracked as its own source in your analytics and CRM. Each month we compare results with the baseline and the cost of running the programme, and decide together where the next euro or franc goes.

Bring one process. We'll bring the loop.