The AI ​​deployed by Inpulse in more than 1000 chained points of sale

Brice Konda, co-founder of Inpulse, is the first to point this out: if his platform, accessible via App, revisits stock management and supplier orders thanks to its layer of artificial intelligence, it does not intend to replace to its users. “The first step is to provide a decision support tool for the catering back-office, with an agile solution that is easy to set up in multi-sites, in less than 4 weeks, and a little more on the networks of more than 50 points of sale, the teams being formed there by clusters of restaurants »he explains. A small revolution in the field, initiated by dedicated SaaS tools, that of Inpulse, but also of Mapal or Agapio, replacing inventory management software, expensive to develop, to deploy, and until then reserved for very large catering operators.

And it has to be recognized that the democratization of these tools is matched by keen interest from the market, attracted by their promise of better control of the food-cost, and ultimately an appreciation of their gross margins. Inpulse currently serves around a hundred customers in commercial catering, on all network sizes, for a thousand points of sale using its platform. “We are deploying on a wide variety of types, among sushi specialists, burgers, poké bowls, in sandwich shops, boul-pat, in corners in supermarkets and in themed restaurants… In all markets, this need to better manage its margins, which were already there before the health crisis, have increased further in a current context of recruitment difficulties and pressure on the price of raw materials. Management topics are at the top of the pile with just about every operator. »

To lend them a hand, the solution developed by Inpulse first plugs into the establishments’ checkout systems, recovering the sales history there. Data from 20 POS currently integrated which, aggregated with others (weather conditions, sporting events, calendar data, etc.) provides access to the predictive layer of the tool. Daily forecasts of turnover, sales by product or by ingredient, which AI refines by use, learning from on-site sales and delivery orders – Deliverect, RushHour, HubRise, Otter and Uber Eats are integrated -, but which can, as soon as they are deployed, turn out to be surprisingly close to reality. “The sinews of war, and the relevance of this type of tool, is data. The more there are, the better the predictions. This is essential to provide users with restocking suggestions that are as close as possible to real needs. » A partially automated reconversion of orders, to the right supplier and at the right price, which Inpulse deduces both from its activity forecasts and from a theoretical stock, deducted from sales thanks to the recipe sheets.

This feature is key in a restaurant where the complexity of restocking is further increased by recruitment difficulties and the turnover frequently observed in establishments. “The idea is to relieve operational staff with automation, to also allow them to better manage margins thanks to purchases that are closer to needs. » On the side of the network heads, for which the platform was also designed, the tool makes it possible to detect anomalies and other discrepancies between theoretical and actual consumption, to also ensure very tight cost control on their recipe sheets. And rather than making promises about the margin points to be sought there, Inpulse preferred to have them measured by some of its customers: Côté Sushi estimates that it has gained 2 margin points, i.e. €800,000 on a network then at 40 restaurants, Bricktop and its 3 pizzerias in Paris quantifying the gain at 4 points… Interesting, at a time when operators are more attentive than ever to defending their margins. Including independent establishments, for which Inpulse is already considering a dedicated solution. Which, in this upcoming deployment, could go through a fundraiser, possibly for the end of the year.

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *