Agentic Grocery Shopping:
The European Reality.
AI agents are beginning to handle grocery shopping autonomously — finding products, comparing prices, and building baskets without manual browsing. The shift is real. What's missing in Europe is the data layer to make it work. This is what Pepesto builds.
What is agentic grocery shopping?
Traditional online grocery shopping is manual: visit a website, search for each item, compare prices across tabs, add things to a basket yourself. It is online shopping, not intelligent shopping.
Agentic grocery shopping replaces that process entirely. An AI agent — powered by a large language model — receives a goal in plain language and handles the full journey autonomously: querying live supermarket data, reasoning about options, comparing prices across chains, and returning a complete grounded answer.
You say: "Find me the ingredients for pasta carbonara for 4 people, cheapest in Amsterdam, under €10." The agent queries Albert Heijn and Jumbo in real time, accounts for current promotional prices, and responds with facts.
Agentic grocery shopping is when an AI agent autonomously handles a multi-step grocery task — product discovery, cross-chain price comparison, and basket building — using live supermarket data, without requiring manual input at each step.
Why Europe is different from the US.
Instacart built agentic grocery infrastructure for one market with one dominant aggregator. OpenAI Operator launched with US retailers. Europe is structurally different — and that structural gap is where the opportunity sits.
Instacart aggregates major US chains under one API. OpenAI Operator connects to that infrastructure. A US user asking an AI agent to "reorder my weekly shop from Whole Foods" works because the data layer exists.
A European user asking the same question about Albert Heijn, Rewe, or Migros ... well, it's difficult. No equivalent to Instacart exists. No US AI platform has deep integrations with European chains. The agentic grocery gap in Europe is structural, significant, and currently unfilled. Pepesto is the answer.
How Pepesto's knowledge graph makes it possible.
The hardest problem in European agentic grocery isn't building an API — it's entity resolution. The same physical product has a completely different identity in every country. Without a knowledge graph that maps those identities, cross-chain comparison is impossible.
Pepesto's grocery knowledge graph maps product entities across all 27 chains. It was engineered specifically for the European grocery fragmentation problem, drawing on the same entity resolution principles used at search-engine scale.
entity_name: "Spaghetti" — regardless of local naming, enabling cross-chain reasoning.LLM + real-time grocery API + entity resolution.
Four layers work together. Understanding each one explains why agentic European grocery has been hard to build — and why Pepesto exists.
api calls with the appropriate supermarket_domain parameters."ah.nl", "shop.rewe.de", or "tesco.com". Daily updated pricing, promotional prices, unit pricing — normalised. No per-chain parsing in the agent.// Step 1: User intent → LLM intent: "pasta carbonara, 4 people, under €10, Amsterdam"// Step 2: Agent calls Pepesto for AH + Jumbo POST /api/catalog { “supermarket_domain”: “ah.nl” }
POST /api/catalog { “supermarket_domain”: “jumbo.com” }
// Step 3: Identical schema, both responses { “entity_name”: “Spaghetti”, “kg_verified”: true, “price”: 109, // EUR cents “currency”: “EUR”, “price_per_meausure_unit”: “0.34 / 100g” “quantity_str”: “1kg” “accurate_grams”: “1000” }
// Step 4: LLM compares entities, responds “AH total: €8.34. Guanciale on promo. Jumbo: €9.12. Adding AH basket now."
Every major AI platform has the same
European grocery problem.
2025 and 2026 have seen a wave of agentic shopping launches from the biggest players in AI. OpenAI, Google, Amazon, Perplexity, Microsoft — all have built or announced agentic shopping capabilities. None of them have solved European grocery. The data infrastructure doesn't exist in their stacks. Pepesto is that infrastructure.
| Capability | All major platforms (EU grocery) | Pepesto + any of the above |
| Live EU supermarket data | None have EU chain integrations | 27 chains live, daily refresh |
| Pricing accuracy for EU | Unavailable or very hard to get | Live prices including current promotions |
| Cross-chain comparison | Not possible — no data source | Same schema across all 27 chains |
| Multi-currency | No normalised EU currency handling | GBP, EUR, CHF, PLN, BGN, NOK, DKK |
| Product entity resolution | LLM guesses across 13 languages | Knowledge graph maps entities across all chains |
| Integration maintenance | You build and maintain each chain yourself | Pepesto monitors and repairs continuously |
| Time to first EU grocery query | Weeks of integration work per chain | 5 minutes — API key, one POST call |
Who is using agentic grocery shopping today.
The category is early but real. Developers and product teams are already building on top of Pepesto — here are the patterns emerging in the market.
Build agentic grocery for Europe.
Start with Pepesto.
27 European chains. One API. One schema. Pay-as-you-go from €0.05 per request. API key returned in 5 minutes — no approval required.
Questions? orders@pepesto.com · Book a call