Ask instead of search
A plain-language question replaces filters, exports and pivot tables.
Let's talk AI
AI that knowsyour portfolio.
The GET17 assistant answers your questions from your data and the product documentation. The GET17 MCP server opens that same data to Claude, ChatGPT, Copilot or any compatible assistant. Either way, AI sees what the user is allowed to see, and nothing more.
Built-in assistantMCP serverAccess rights enforced
Example in the application● online
Rank the properties by 2025 EBITDA, from lowest to highest.
Here's the ranking of your 12 properties. Hôtel Les Tilleuls comes last with 2025 EBITDA of €412k; the highest is above €2.1M. Each line links to the property's record.
For the bottom 5, which CAPEX requests are pending?
Four requests are pending, for €152,300 excl. VAT in total. The largest is the “Heat pump replacement” at Hôtel Les Tilleuls (€48,600 excl. VAT), on the agenda of the next committee. Here are the links to each request.
Ask your question…
What AI changes
Fewer clicks,more answers.
Your data is already structured in GET17. AI lets you query it the way you'd ask a colleague who knows every file.
From documents to data
Findings in a PDF inspection report become items to track; quotes and invoices are captured by OCR.
From voice to text
Voice notes recorded during an audit are transcribed, then summarized into the recommendation.
Your AI, your tools
With the MCP server, the assistant your teams already use reads your GET17 data directly.
The GET17 assistant
A built-in assistantthat answers with your numbers.
Available from any page of the application, the GET17 assistant relies on three sources: your records, the calculation engines behind your dashboards, and the GET17 documentation. When the answer isn't there, it says so.
KPIs and statistics
Cash, bank balances, EBITDA and revenue, rents, loans, projects, CAPEX, energy, equipment, work orders, audits, preventive maintenance: the assistant queries the same calculation engines as your dashboards, by period and by scope.
“Compare actual and budgeted 2026 EBITDA by company.”
Search your data
Properties, CAPEX requests, contracts, suppliers, inspections: it finds the records the user can see and links to each one.
“Which auto-renewing maintenance contracts reach their notice deadline in the next 3 months?”
Help from the documentation
It answers how-to questions from the GET17 documentation, module by module, and shows where the information comes from.
“How do I freeze a project budget?”
Document reading
The “Extract findings” action turns a PDF inspection report into findings to close out. “Add by OCR” pre-fills an order's quotes and invoices.
Audit voice notes
Recordings made on mobile during an audit are transcribed, then summarized into a short, factual comment on the recommendation.
Hand-off to support
If the answer isn't in its sources, the assistant offers to open a support ticket and shows you the summary before sending it.
Fictional example. Answers are calculated on the signed-in user's scope.
The GET17 MCP server
Your GET17 data,in the assistant of your choice.
Do your teams already use Claude, ChatGPT or Copilot? The GET17 MCP server gives them read-only access to your data, with the user's permissions.
Your AI assistantClaude · ChatGPT · Microsoft Copilot · Any MCP-compatible client
GET17 MCP serverTools, permissions, logs
GET17Your data, your permissions
What is MCP?
The Model Context Protocol is an open standard that lets an AI assistant use external tools. GET17 publishes its own: the assistant picks the right tool, GET17 runs the request and returns the result.
Search records
Properties, companies, contracts, suppliers, CAPEX requests, projects, audits: anything the user can see, as a list or in detail.
Filter precisely
GET17 tells the assistant which criteria each type of data supports: dates, statuses, amounts, categories, properties.
Query the calculation engines
Performance, cash, loans, rents, projects, CAPEX, energy, equipment, work orders, audits, maintenance: the figures behind your dashboards.
Compare and analyze
By period, property, company or brand, with ratios per occupied room, per guest or per unit of revenue.
A link to every record
Each result links to the matching record in GET17, so you can check it or carry on in the application.
Read-only
The MCP server reads your data and never creates, changes or deletes anything.
MCP-compatible assistant● online
Give me an update on Hôtel Les Tilleuls: open work orders, overdue maintenance and CAPEX in progress.
GET17 · Search · Property “Hôtel Les Tilleuls”
GET17 · Search · Open work orders · 4 results
GET17 · Calculation engine · Preventive maintenance · 2 overdue, 3 open findings
GET17 · Search · CAPEX requests in progress · 1 result
Here's the update: four work orders are open, one of them for more than 15 days. Two maintenance tasks are overdue and three findings remain open. The “Heat pump replacement” request (€48,600 excl. VAT) is awaiting the committee's decision.
Which equipment breaks down most often there?
GET17 · Calculation engine · Work orders by equipment · last 12 months
Plumbing fixtures come first with 14 work orders over the last 12 months, ahead of air conditioning (9) and door locks (6). Each work order can be opened from its link in GET17.
Ask your question…
Fictional example. The MCP server is read-only: every call reads GET17 with the user's permissions.
Turn on the MCP server
An administrator enables it for your GET17 platform, chooses the roles or users allowed and the tools exposed.
Add GET17 to your assistant
Register GET17 as a connector in Claude, ChatGPT, Copilot or any MCP-compatible client.
Use your personal API key
Created from “My API keys” and approved by an administrator, it gives the assistant your permissions and your property scope, nothing more.
AI by role
Every job,its own questions.
A few examples of what each role asks the GET17 assistant, or their own assistant connected through MCP.
Executives
“For each hotel, what's left of 2025 EBITDA after rent and loan repayments?”
“Prepare the next CAPEX committee: listed requests, amounts, properties and the EBITDA of each.”
DiscoverCFO
“What are each company's 2026 cash needs (projects, loan repayments, equipment replacement) against its current bank balance?”
“What's each company's debt-to-EBITDA ratio (outstanding principal at end-2025 ÷ 2025 EBITDA)?”
DiscoverAsset Manager
“What's each hotel's rent-to-revenue ratio for 2025?”
“Which hotels have exceeded their 2025 FF&E reserve?”
DiscoverTechnical Director
“Which properties have overdue or unrecorded maintenance?”
“How many findings are still open, by supplier?”
DiscoverProject Manager
“Which ongoing projects have more than €100k left to pay?”
“Which purchase orders are waiting for an e-signature or a level 1, 2 or 3 approval?”
DiscoverSite Manager
“Which work orders at my property have been open for more than 15 days?”
“Which equipment at my property is worn out or reaches end of life before 2028?”
DiscoverTechnician
“Which plumbing requests have guests made this month?”
“Which equipment breaks down most often at my property?”
DiscoverLegal Counsel
“Which lease financing agreements expire before the end of 2027?”
“Which capital increases or reductions have taken place in the last 3 years?”
DiscoverProcurement Manager
“Which suppliers have missing documents despite more than €50k ordered over 24 months?”
“Which suppliers are in insolvency proceedings or have ceased trading?”
DiscoverAI by industry
Every industry,its own priorities.
The same tools, applied to what matters in your business.
Hospitality
Connect operations and assets: occupied rooms, energy, CAPEX and FF&E reserve.
“What was the electricity cost per occupied room, by property, in 2025?”
Restaurants
Track equipment breakdowns and safety inspections at every restaurant.
“How do work orders break down by equipment family and resolution time?”
Serviced residences
Stay on top of a housing portfolio: repairs, inspections, refurbishment works.
“How did the number of work orders change month by month in 2025, by residence?”
Healthcare
Meet every regulatory deadline and document every inspection.
“What share of maintenance is up to date at each facility?”
Franchise networks
Compare the sites in a network and track audits and investment site by site.
“Which sites haven't been audited in more than 24 months?”
Industry & logistics
Plan equipment replacement and keep track of installation inspections.
“What equipment replacement budget should we plan each year through 2030, by equipment type?”
Real estate funds
Consolidate works, debt and cash for a portfolio held through several companies.
“What's the outstanding principal on loans and lease financing, by company, at the end of 2026?”
Trust and security
AI with guardrails,working within your rules.
AI doesn't bypass anything: it goes through the same controls as the application.
The user's rights, nothing more
Every question and every action runs under the signed-in user's identity, verified server-side. AI only sees what that user can see.
Property scope enforced
Figures and lists cover the companies and properties the user can access, and are presented as such.
Your data stays with you
Data stays in your hosting environment. Only the information needed for a question is sent to the model, at the moment it's asked.
The AI provider of your choice
GET17 works with several model providers, including Anthropic, OpenAI, Google and AWS. You choose the one that fits your policy.
Conversations on record
Conversations are kept in GET17 for the retention period you set. Calls to the MCP server are logged: user, tool and parameters, never the data returned.
Humans make the call
Extracted findings, summarized comments, support tickets: AI suggests, the user reviews and confirms. The MCP server changes nothing.
- No source, no answer: the assistant says when information is missing instead of making it up.
- Record content is treated as data, never as an instruction to the AI.
- Partial results are flagged as partial, never presented as totals.
Frequently asked questions
AI in GET17,your questions.
What's the difference between the GET17 assistant and the MCP server?
The assistant is built into the application: you open it from any page. The MCP server lets an external assistant, such as Claude, ChatGPT or Copilot, read GET17 data, read-only. Both apply exactly the same access rights.
Can AI see a property I don't have access to?
No. Every request runs under your identity and is limited to your rights and your property scope. If nothing matches within that scope, the AI tells you.
Where is our data processed?
Your data stays in your GET17 hosting environment. To answer, only the information needed for the question is sent to the chosen model provider. That choice is made with you, based on your security policy. With the MCP server, results are sent to the assistant you connect, under its provider's terms.
Can AI change our data?
The MCP server is read-only: it reads your data and never changes it. In the application, the assistant suggests (findings extracted from a report, a support ticket) and nothing is saved without your confirmation.
Where do the assistant's answers come from?
From three sources only: your records, the calculation engines behind your dashboards, and the GET17 documentation. The assistant links to the relevant records and doesn't fill gaps with general knowledge.
Do we have to turn on AI to use GET17?
No. AI features are enabled as you need them, and GET17 is complete without them.
Let's see what AI
can do with your data.
A demo of the assistant and the MCP server on real questions from your organization.