Choosing the process
Together we pick the process that repeats most often, has a clear definition of "done" and has its data in your systems. We define what the AI does on its own and what it does only after a person approves.
AI for restaurants and cafes
Bookings arrive through Messenger, reviews land on Google Business Profile and Wolt, sales live in Lightspeed, and supplier invoices reach SAP Business One a month late. AI assistants and agents take over this routine under your rules, and your data stays in the EU and is available only to those who need it.
Built for ISO 27001NIS2 DORA
- Reservation and enquiry assistant
- Menu and allergen assistant
- Review analysis and replies
Bookings in Messenger, sales in Lightspeed, invoices in SAP Business One, and everything between the tables
- 24/7 Messenger booking enquiries answered at night too
- 1 queue Messenger, email and website in a single queue
- Minutes supplier invoice in SAP Business One without manual entry
- EU guest data stays in Lithuania (EU) or an EU region
A restaurant or cafe day starts with messages: someone on Messenger asks whether there is a table for six, an email arrives about a birthday, a Google Business Profile review from last night's guest complains about slow service, and a Wolt rating about cold food. Meanwhile the kitchen receives goods from a supplier, the delivery note sits on a shelf, and the supplier invoice will reach SAP Business One a month later, when the bookkeeper keys it in by hand. Staff answer allergen questions from memory, and shifts are scheduled by gut feeling, even though Lightspeed holds every hour's receipts. All of this works until the manager takes a day off or a new employee joins.
AI takes over the repetitive parts here: it answers booking enquiries on Messenger and by email and passes to the manager only what it cannot resolve itself, answers staff questions about dish composition from your approved recipe cards, collects reviews from Google Business Profile, Wolt and Bolt Food and prepares reply drafts, turns supplier invoices into ready-to-approve entries in SAP Business One, and builds a weekly occupancy forecast from Lightspeed sales. MDP CLOUD implements these solutions with Claude, ChatGPT or Microsoft Copilot and connects them through an API, file export or email flow, depending on your system versions. We start with security: we define who can see guest data, where it is stored (in the MDP CLOUD data centre in Lithuania (EU) or an EU region) and in which cases the agent must stop and wait for a person.
What we implement for restaurants and cafes
The 5 solutions we implement most often in this industry. Each one starts with a single process, has a human approval step and is measured.
Reservation and enquiry assistant
On a Friday evening a guest messages on Messenger asking whether there is a table for six. The agent checks free tables in Lightspeed or your booking system, confirms or suggests another time, and records the booking where the manager sees it. Groups, events and special requests go to the manager with a summary of the conversation.
We build it with: Claude agent via API + Messenger, email and a Lightspeed or other booking system integration
Menu and allergen assistant
A guest asks whether the Bolognese sauce contains celery, and a new waiter runs to the kitchen. The assistant answers from your approved recipe cards, exported from Lightspeed or your costing system, and from the allergen tables, always citing the source. Recipes are changed by the chef; the assistant never creates them.
We build it with: ChatGPT Custom GPT with recipes from the recipe cards of Lightspeed or another costing system
Review analysis and replies
Reviews are collected from Google Business Profile, Wolt and Bolt Food ratings, and Facebook and Instagram, classified by topic (service, kitchen, prices, cleanliness, waiting time), and a reply draft in your tone is prepared for each. Every Monday the manager receives a summary in Teams of what keeps recurring. A person publishes the reply; the agent never posts publicly.
We build it with: Copilot Studio agent + Power Automate with Google Business Profile and Wolt review collection, summary in Teams
Supplier invoices and goods receipt
The chef photographs the supplier's delivery note on receipt, and the invoice arrives by email. Recognition extracts lines, quantities and prices, matches them against the order and the contract price list, and prepares a purchase entry in SAP Business One or Xero. Discrepancies (quantity, price, missing item) are flagged for review. The bookkeeper approves a prepared entry instead of keying it in again.
We build it with: Document recognition + integration with SAP Business One or Xero via API or import files
Occupancy summaries from POS data
From Lightspeed sales by hour, the flow of bookings and Wolt orders, and the events calendar, we prepare next week's occupancy forecast and a suggested staffing need by hour. The manager sees why exactly that many people are suggested and builds the rota. The system assigns nothing and changes nothing without your decision.
We build it with: Claude via API + Lightspeed sales export, report in Power BI
How we implement it
From the first conversation to a working solution, with measurement and ongoing support.
Access and data
We put access to files and systems in order, give the AI its own identity with the least privileges it needs, turn on an action log and choose where the model runs: your Microsoft 365, Amazon Bedrock in an EU region or our GPU cloud in Lithuania (EU).
Build and test
We build the solution with Copilot, Claude or ChatGPT, whichever fits the task, connect it to your systems and test it on historical documents and enquiries until accuracy reaches the agreed threshold.
Launch and support
We launch with a person in the loop, review the logs and exceptions and adjust the rules. Once the first process runs reliably, we move on to the next one on the list.
Are your processes different?
These solutions are a starting point for a conversation, not a complete list. Tell us which processes repeat in your company, and in one conversation we will tell you what is worth handing to AI and what is not.
Frequently asked questions
Can an AI assistant make a mistake when taking a reservation on Messenger?
It can, which is why we implement it with clear rules: it confirms only what it sees as free in Lightspeed or your booking system, and it passes groups, events and unclear enquiries to the manager. Every booking is visible in your usual system, so it can be corrected in seconds. We start with a pilot period in which a person reviews all of the assistant's replies.
Does it work with Lightspeed and SAP Business One?
Yes, these are among the most common systems at our restaurant customers. We connect Lightspeed through its report export or API, depending on the version and your vendor's agreement, and SAP Business One through its API, import files or ERP hosting that we manage ourselves. Recognised invoices reach SAP Business One as prepared entries that the bookkeeper approves. We choose the integration method after a check, and we do not claim an official partnership with these systems.
Where are guest booking data and Messenger conversations stored?
Your choice: in the MDP CLOUD data centre in Lithuania (EU), in a Microsoft EU region or in Amazon Bedrock in an EU region. We define how long conversations are kept, who can see them and how GDPR requirements are met. Models are not trained on your guest data, and the Messenger conversation history stays on your Facebook page.
Related services
AI for retail
A supplier sends a thousand products with no descriptions for PrestaShop or WooCommerce, customers ask about stock in Lightspeed and parcels with…
AI for transport and logistics companies
Orders from Trans.eu, Outlook and customer portals are retyped into CargoWise by hand, CMR photos travel through Viber, and DKV invoices are keyed…
AI for wholesale
Orders arrive as emails, Excel sheets, PDFs and through EDI from retail chains, customers call about stock and contract prices, and every month…
AI for accounting firms
Client invoices arrive in Outlook as phone photos and PDFs, bank statements are keyed in by hand, and month-end close starts with reminder e-mails.
Expertise confirmed by certifications
This service is delivered by specialists with valid international certifications. Their expertise is confirmed by independent organisations, not by us.
-
Artificial Intelligence Security Professional
AI systems are deployed with protection against data leakage and model abuse.
-
ISACA Advanced in AI Security Management
The AI security programme and its controls are managed following the ISACA methodology.
-
ISACA Advanced in AI Risk
AI risks are identified and managed like any other IT risk.
Let's discuss an AI solution for your restaurant or cafe
Tell us which processes repeat most in your company. Within 1 business day we will tell you where it makes sense to start and how long it would take.
We reply within 1 business day