Connect Make to AI: a scenario that starts with what someone asks
Make can be connected to AI, purely to set something going. The assistant can fire off a scenario and supplies the details the scenario needs. What that scenario then carries out stays invisible to the assistant: all it learns is that the scenario has departed. Because the modules behind it do real work that sometimes cannot be undone, firing one off happens only when somebody explicitly asks for it.
What happens when the assistant starts a scenario
Fire off a scenario
The assistant starts the scenario you have pointed out and supplies the details the first module expects. In the conversation that is no more than a sentence; in Make it is the starting shot.
The modules do the work
Everything that follows is built in Make: module after module, possibly with a router choosing which branch fits. The assistant does not interfere with that and cannot change any of it.
Hand work to systems further down the line
A scenario can touch several systems in sequence, including systems with no connector of their own. The assistant does not have to know them; it merely sets the chain you built in motion.
No outcome in return
What returns is the notice that the scenario departed, without content. Questions about the result are answered by Make, not by the assistant.
The scenario stays under your control
If a step is added or a module replaced, nothing has to change in IntraGPT. The assistant keeps firing off the same scenario, even when it is built differently underneath.
Chains that begin in a conversation
Industry and manufacturing
- Situation
- A breakdown has to set off a fixed series of actions, spread across a maintenance system and a supplier portal.
- What the agent does
- The work planner describes the breakdown and the assistant fires off the scenario that works through that series.
- Result
- The chain runs while the engineer is still standing at the machine, instead of after a form has been filled in.
Customer service
- Situation
- An unusual case has to pass three departments, and which route it takes depends on the type of complaint.
- What the agent does
- The assistant starts the scenario with the type and the customer details; the router in Make picks the right branch.
- Result
- The right department sees it immediately and nobody has to know the route by heart.
Administration
- Situation
- Every month the same details from a conversation have to land in a series of systems, always in the same order.
- What the agent does
- The assistant fires off the monthly scenario as soon as the details are complete.
- Result
- The same order, every month, without anyone having to work through a checklist.
Facilities
- Situation
- A report about a building has to reach the right party, and that differs per location.
- What the agent does
- The assistant passes on the location and the report and starts the scenario that contacts the right party for that location.
- Result
- The report reaches the correct place, even when raised by colleagues who do not know the local arrangements.
From scenario to assistant
-
1
Preparing the scenario
You build the scenario in Make and hand over its secret address in IntraGPT. That is all the assistant needs to be able to fire it off.
-
2
Setting the roles
Who may set a scenario going is arranged with the roles in the platform. That is often a smaller group than the one allowed to search and read.
-
3
What the assistant gets to see
Only what it sends along itself. It cannot look into Make, not into the run and not into the systems the scenario visits along the way.
-
4
What gets recorded
The log shows which request was passed on, when and by whom. What the scenario made of it you read back in Make.
One address that can start a whole chain
Dozens of steps can hang behind a scenario, together setting quite a lot in motion. The address that starts such a chain therefore deserves the same care as a password.
Where the data lives: Nederland of de EU
Kept shielded, replaceable
The address sits shielded in the platform and never turns up in a conversation. If you have doubts, you create a new address in Make and the old one loses its effect.
Never unasked
The assistant first lists what it is going to send along and starts the scenario only after a clear confirmation. That matters most for scenarios that touch money, stock or customer contact further down the line.
Two logs side by side
On our side it is fixed when a start happened and at whose request. On your side Make shows per module what happened afterwards. Together that is the full picture.
On a dedicated server
The model runs on our own server in the Netherlands. The details travelling with a scenario then arrive at Make and at the systems the scenario visits, so decide deliberately what goes along.
What people want to know about Make and IntraGPT
Can I connect Make to ChatGPT?
A public chatbot has no safe place for the address that starts a scenario, and no roles to decide who may use it. In IntraGPT that address sits locked away, who may start something is arranged per role, and every start is kept.
Does IntraGPT work with Make?
Yes, for starting scenarios. This connector is deliberately narrow: it fires off and supplies details, nothing more. It is a working integration already in use at clients.
Do we see the result of the scenario back in the chat?
No. The assistant learns that the scenario has departed and nothing further. What each module did you read off in Make, where the run is kept. If you want the answer in the chat instead, this is not the route, and we will look together at a connector that can read.
Who decides what the scenario does?
Your own administrator in Make. We do not touch the scenario and do not need to know how it is built inside. If the modules change, the assistant simply keeps firing off the same scenario.
Where is our data once the scenario runs?
From the moment the scenario starts, the details sent along sit with Make and with the systems the scenario visits on the way. Make offers European hosting, which is worth checking with your administrator. In any case include only what the chain genuinely needs.
Can we pause a scenario temporarily?
Yes, in two ways. You switch the scenario off in Make, after which nothing happens when the assistant tries, or you remove the connector in IntraGPT, after which the assistant no longer offers it. For a temporary pause the first is usually more practical.
What does a Make connector cost?
Switching it on is a small part of the implementation. The time goes into deciding which scenarios may start from a conversation and which details belong with them. For Make itself you pay Make directly; that sits outside our project.
How long does setup take?
If the scenario is ready, it comes down to pointing it out and doing a test run. If it still has to be built, that is the real work and it happens in Make, where your administrator lines up the modules.