Connecting Trengo to AI: one inbox for WhatsApp, email, chat and phone
Trengo can be connected to AI. The assistant surveys the running conversations across all channels, looks up a contact, opens a new conversation and assigns a conversation to a colleague or a team. It can also send a message, and that goes through the channel the conversation runs on. It genuinely reaches the customer, so it only happens on explicit request and after the text has been shown.
What the assistant does in the Trengo inbox
Survey conversations across channels
The assistant retrieves the running conversations with their channel, their status and their label. One answer tells you what is still open through WhatsApp, email and chat, instead of checking the same thing three times.
Look up a contact
Before a conversation is opened, the assistant checks whether this customer already exists. That saves duplicate contacts and keeps the history in one place.
Open a new conversation
When a question comes from somewhere else, the assistant can turn it into a conversation on the right channel and with the right contact. The inbox then stays the complete picture of what is going on.
Assign to a colleague or team
A conversation can be assigned to a person or to a whole team. Handy under pressure, during holidays or when a question clearly belongs with the technical service or the finance department.
Send a message on the right channel
A message follows the channel the conversation runs on. For WhatsApp that means the rules from Meta: within twenty-four hours of the customer's last message you may write freely, outside that window only with an approved template.
How teams use Trengo with AI
E-commerce
- Situation
- A customer sends a WhatsApp message about an order, having emailed about the same thing yesterday.
- What the agent does
- The assistant looks up the contact, pulls in both conversations and drafts one answer with the current status from the connected order system.
- Result
- The customer gets one clear answer instead of two half ones, and the colleague sees at once that it was the same question.
Retail
- Situation
- Around a promotion day the inbox fills up and it is no longer visible which conversations belong to nobody.
- What the agent does
- The assistant lines up the open conversations per channel and assigns the ownerless ones to the team on duty that day.
- Result
- Nothing is left behind because nobody felt responsible for it.
Installation and maintenance
- Situation
- A customer asks in the chat when the engineer is coming, and that appointment lives in another system.
- What the agent does
- The assistant looks up the appointment, drafts the answer and shows which channel it will be sent through.
- Result
- An answer with the right day and time, without anyone switching between two screens.
Customer service
- Situation
- At the end of the day a team lead wants to know what is left and with whom.
- What the agent does
- The assistant groups the open conversations per colleague and per channel and names the ones where the customer spoke last.
- Result
- The handover to the next day takes minutes instead of half an hour.
Connecting the shared inbox
-
1
Connecting through the platform
An administrator records access to your Trengo environment in IntraGPT once. That access can be withdrawn at any moment and the connector stops right away.
-
2
Choosing what the assistant may do
Looking, assigning and sending messages are separate choices. Many teams first let it only build overviews and assign, and add sending later.
-
3
Using the assistant
After that you ask what is running for a customer, what is still open today or who is working on what. The assistant pulls in the conversations and says which channel something came from.
-
4
Visible in the log
Every retrieval, assignment and send lands in the log, with the moment and the question that preceded it. For a sent message you can therefore trace at whose request it went.
Messages arrive on the customer's phone
A message from Trengo often lands on a private phone, which calls for restraint about what you include and clarity about who may have something sent.
Where the data lives: Nederland of de EU
Sending is final
A sent message cannot be pulled back, on any channel. The assistant first shows the contact, the channel and the full text, and does not send without an explicit request.
Within the rules of the channel
For WhatsApp the Meta rules around the twenty-four hour window and approved templates apply. Those rules are part of how the assistant works, so you do not cross them by accident.
Roles and teams
You decide per role who may use the assistant and what it may do. Assigning is a different authority than answering on behalf of the company.
The model runs on your own server
Conversation content is processed on our own server in the Netherlands with open-weight models. That content does not go to an external model provider and is not used to train models.
Questions about the Trengo connector
Can I connect Trengo to ChatGPT?
Not in a way that suits a customer service team. You would give a public service access to every conversation with your customers, without roles, without a log and without control over what gets sent. IntraGPT connects your own Trengo environment, with permissions you set, and processes the contents on our own server.
Does IntraGPT work with Trengo?
Yes. Trengo is in the connector catalogue and in use at customers. The assistant can survey conversations, look up contacts, open a conversation, assign a conversation and send a message. Which of those are switched on is up to you.
Which channel does a message go out on?
The channel the conversation runs on. If the customer started on WhatsApp the answer goes over WhatsApp; if the question came by email it goes by email. That is pleasant for the customer, but it also means the rules of that channel apply, and for WhatsApp those are the strictest.
What does the twenty-four hour rule on WhatsApp mean?
Meta only allows free-form messages within twenty-four hours of the customer's last message. After that you may only send a pre-approved template. That is Meta policy and applies to everyone using WhatsApp for business. The assistant sticks to it instead of discovering it through an error.
Can the assistant assign a conversation to a team?
Yes, to a colleague or to a whole team. That is often the most useful first step: the assistant makes sure every conversation has an owner, without sending anything to customers itself. Who is available on a given day remains the team's own call.
How does this differ from the WhatsApp connector?
With the WhatsApp connector, IntraGPT talks directly to the business WhatsApp environment. With Trengo everything runs through the shared inbox, so email, chat and the rest come along with WhatsApp. If you already work in Trengo, this connector is usually the logical choice, because that is where your team already sits.
What does a Trengo connector cost?
Connecting is part of the project we run together. What drives the work is how many channels are involved, whether the assistant may also assign and send, and which roles may use it.
How long does setup take?
The connector stands quickly and can be tested the same day. The lead time sits in the agreements: which channels the assistant may see, who may have a message sent and which templates you have ready for WhatsApp outside the window.