[{"data":1,"prerenderedAt":241},["ShallowReactive",2],{"blog-article-en-mcp-model-context-protocol-restaurant":3,"blog-locales-mcp-model-context-protocol-restaurant":225,"blog-related-en-mcp-model-context-protocol-restaurant":234},{"id":4,"title":5,"body":6,"category":213,"categoryColor":214,"date":215,"description":216,"extension":217,"meta":218,"navigation":219,"path":220,"readTime":221,"seo":222,"stem":223,"__hash__":224},"blog\u002Fblog\u002Fen\u002Fmcp-model-context-protocol-restaurant.md","MCP (Model Context Protocol): what it changes for a restaurant",{"type":7,"value":8,"toc":203},"minimark",[9,13,20,25,32,35,39,42,45,79,82,86,89,108,112,115,142,146,155,159,192],[10,11,12],"p",{},"You may have already asked an AI assistant to analyse your sales. The scenario is often the same: export a CSV file from each tool, clean it, paste it into the conversation, then start again the following week. The answers are interesting, but the work is tedious and the data already out of date.",[10,14,15,19],{},[16,17,18],"strong",{},"MCP (Model Context Protocol)"," was designed to avoid that.",[21,22,24],"h2",{"id":23},"mcp-in-one-sentence","MCP in one sentence",[10,26,27,28,31],{},"MCP is an ",[16,29,30],{},"open standard"," that lets an AI assistant connect to tools and data sources in a structured way: instead of sending it files, you give it controlled access to the tools themselves.",[10,33,34],{},"In practice, an \"MCP server\" exposes what a tool can do: look up orders for a period, read the menu, get the history of a price. The AI assistant calls these functions when it needs them, with data that's always up to date.",[21,36,38],{"id":37},"why-its-useful-for-a-restaurant","Why it's useful for a restaurant",[10,40,41],{},"A restaurant selling through delivery juggles many sources: each platform's tablet, the POS, the menu, the order history, disputes. Answering a simple question — \"why did my Tuesday evening sales drop?\" — means cross-checking several of these sources.",[10,43,44],{},"With an MCP server connected to your tools, the assistant can:",[46,47,48,55,61,67,73],"ul",{},[49,50,51,54],"li",{},[16,52,53],{},"read your sales"," by dish, by platform and by time slot;",[49,56,57,60],{},[16,58,59],{},"retrieve price history"," and see when a change had an effect;",[49,62,63,66],{},[16,64,65],{},"cross-check with events"," (holidays, weather, matches) to explain a variation;",[49,68,69,72],{},[16,70,71],{},"spot profitable time slots"," and those that aren't;",[49,74,75,78],{},[16,76,77],{},"suggest adjustments"," to prices, menu or opening hours.",[10,80,81],{},"All without manual exports or copy-pasting.",[21,83,85],{"id":84},"what-mcp-doesnt-do","What MCP doesn't do",[10,87,88],{},"MCP isn't magic:",[46,90,91,98,105],{},[49,92,93,94,97],{},"it doesn't replace ",[16,95,96],{},"data quality",": an incomplete history gives incomplete analyses;",[49,99,100,101,104],{},"it doesn't decide for you: the AI ",[16,102,103],{},"suggests",", you validate;",[49,106,107],{},"it only gives access to what the server exposes: it's the tool's publisher that defines the scope.",[21,109,111],{"id":110},"precautions-to-take-with-your-data","Precautions to take with your data",[10,113,114],{},"Plugging an AI into your tools means entrusting it with sensitive data: revenue, customers, prices. Before enabling a connection, ask these questions:",[116,117,118,124,130,136],"ol",{},[49,119,120,123],{},[16,121,122],{},"Who has access to what?"," Access should be limited to your restaurant's data, and authorised by you.",[49,125,126,129],{},[16,127,128],{},"Where does the data go?"," Find out how it's processed and hosted, and whether it complies with the GDPR.",[49,131,132,135],{},[16,133,134],{},"Can you cut off access?"," You must stay in control of the connection.",[49,137,138,141],{},[16,139,140],{},"Is customer data protected?"," A sales analysis generally doesn't need your customers' personal details.",[21,143,145],{"id":144},"and-where-does-pepprio-fit-in","And where does Pepprio fit in?",[10,147,148,149,154],{},"Pepprio has built its own ",[150,151,153],"a",{"href":152},"\u002Fen\u002Ffonctionnalites\u002Fagent-ia","AI agent and MCP server",". With your consent, they connect to your Pepprio tablet, your POS, your delivery platforms and your history to analyse your menus, orders and prices, and suggest improvements. They're available to independent restaurants as well as brands and chains, with analysis per brand and per site, and only access the data of the restaurant that authorises them, in line with the GDPR.",[21,156,158],{"id":157},"in-short","In short",[46,160,161,168,175,181],{},[49,162,163,164,167],{},"MCP connects an AI assistant ",[16,165,166],{},"directly"," to your tools, without exporting files.",[49,169,170,171,174],{},"For a restaurant, it lets you analyse sales, prices, time slots and events ",[16,172,173],{},"at the source",".",[49,176,177,178,174],{},"The AI suggests, ",[16,179,180],{},"you decide",[49,182,183,184,187,188,191],{},"Always check the ",[16,185,186],{},"access scope"," and ",[16,189,190],{},"GDPR compliance"," before connecting your data.",[10,193,194,195,187,199,174],{},"Also read: ",[150,196,198],{"href":197},"\u002Fen\u002Fblog\u002Ffixer-prix-livraison-historique-ventes","setting delivery prices using your sales history",[150,200,202],{"href":201},"\u002Fen\u002Fblog\u002Fdata-driven-restaurant-tableau-bord","running your restaurant with data",{"title":204,"searchDepth":205,"depth":205,"links":206},"",2,[207,208,209,210,211,212],{"id":23,"depth":205,"text":24},{"id":37,"depth":205,"text":38},{"id":84,"depth":205,"text":85},{"id":110,"depth":205,"text":111},{"id":144,"depth":205,"text":145},{"id":157,"depth":205,"text":158},"Management","is-link","2026-09-27","MCP lets an AI assistant connect directly to your tools. A simple explanation, concrete use cases for a restaurant and the precautions to take with your data.","md",{},true,"\u002Fblog\u002Fen\u002Fmcp-model-context-protocol-restaurant","6 min",{"title":5,"description":216},"blog\u002Fen\u002Fmcp-model-context-protocol-restaurant","KioRINPCKOgaz33JLy9YHApwsOOaBt5Fx81rXp4pAxw",[226,227,228,229,230,231,232,233],"ar","ca","de","en","es","it","fr","pt",[235],{"path":236,"title":237,"description":238,"category":213,"categoryColor":214,"date":239,"readTime":240},"\u002Fblog\u002Fen\u002Fmulti-etablissements-gestion-centralisee","Managing multiple restaurants: technical architecture for scaling without pain","When you move from 1 to 3, 5, or 10 locations, your technical stack should scale smoothly. Here's the architecture that allows for seamless scaling.","2025-06-25","7 min",1790617967114]