Eddie Blomkvist
Yes, monday.com can be connected to ChatGPT. With the integration, what is written in monday.com can be sent to OpenAI's models as a step in a flow: a support ticket is summarized, a new lead is classified, a draft reply is added to the ticket before the agent opens it. It is usually built in an integration platform like Make, with the OpenAI API behind it, and should not be confused with monday AI or with a custom AI agent. This article covers what the integration does, how it is built, how it differs from the alternatives and what to consider before you get started.
- A new lead is created in monday.com
- The description and company details are sent to ChatGPT
- Router · Does the lead fit your target audience?YesSegment and priority are set, a rep is assignedNoThe lead is flagged for manual review
- The rep gets a three-line summary
The model gets your definition of a good lead as its instruction and replies with segment, priority and reasoning, which are written to columns in monday.com.
- A new ticket comes in through a form or email
- ChatGPT summarizes the ticket and suggests a category
- Category, summary and urgency are written to the ticket
Long email threads become a short summary and a category that decides which team gets the ticket.
- The ticket is assigned to an agent
- ChatGPT writes a draft based on the ticket and your tone of voice
- The draft is added as an update on the ticket
- The agent reviews, adjusts and sends
The draft is a suggestion. The agent always decides what goes out to the customer.
- Every Friday, the week's updates are pulled from the project board
- ChatGPT summarizes status, risks and decisions made
- The summary is sent to the project owner
The status report is based on what was actually written in the project.
The flows above are only examples. Every integration is built around your process: which steps to include, what instructions the model gets and who reviews the result.
Short answer
There is no ready-made ChatGPT button in the monday.com Integrations Center for AI steps in your flows. What exists is three things. monday AI is monday.com's own AI blocks that can summarize, categorize and extract details from text directly in automations. Make has an OpenAI module that sends text from monday.com to OpenAI's models and writes the response back, with full control over instructions and format. And monday.com can be connected to ChatGPT through monday MCP, but that is for asking questions about your boards, not for automated steps. If you need your own instructions, several steps or responses in a fixed format, Make is the right choice. That is where we build most integrations.
What the integration can do
Here are four common examples. Your integration can do more, less or something entirely different.
Email threads, form responses and meeting notes are summarized into a few lines in a column, so whoever takes over sees the core right away.
Leads, tickets and requests get a category, segment and priority based on their content and your rules. The result decides who gets assigned and how quickly it should be handled.
Replies to tickets, follow-up emails and short status reports are written as drafts based on what is on the item. A person always reviews before anything is sent.
Company registration numbers, amounts, dates and products are pulled out of unstructured text and placed in the right columns.
Why connect monday.com and ChatGPT?
- Less reading before the work starts. Whoever picks up a ticket or a lead gets a summary and a category instead of a long thread.
- More consistent quality. Classification and prioritization follow the same instruction every time, not whoever happens to be watching the inbox.
- Faster first response. A draft is ready when the agent opens the ticket, and what remains is to review and adjust.
- AI where the work already happens. The result lands in columns that drive the rest of the flow, not in a separate chat window.
How the integration is built
There are three ways to build it, and which one fits depends on how much control you need.
| Option | Best when | Keep in mind |
|---|---|---|
| monday AI | Simple steps in existing automations, for example summarizing an update, categorizing an item or translating a text | Quick to turn on and needs no external service, but the instructions are limited and usage counts against AI credits |
| Make with the OpenAI module | Flows with your own instructions, several steps and responses in a fixed format, for example classifying a lead by your definition and writing back segment and reasoning | Requires an OpenAI API account billed by usage, and a well-designed flow: how the instruction is written, how the response is interpreted and what happens when the model gets it wrong |
| Custom code against the OpenAI API | Large volumes, your own knowledge sources or agents that work through several steps without a person starting each run | The most flexible, but requires someone to own and maintain the code and the instructions |
For most companies, Make is the right choice, but the platform is just the tool. What determines whether the integration holds up is how it is built: how the instruction is written, how the response is interpreted and written to the right columns, how the flow handles the model occasionally answering unexpectedly, and what happens when something goes wrong. An integration that gives good answers in a test but silently sets the wrong category in production costs more than it saves. That is why we build with logging, review steps and alerts from the start, and manage the integration as the process or the models change.
What to decide before you build
An integration is never better than the process it automates. These questions determine whether it works in practice:
- What should the model get to see? Everything sent to OpenAI leaves monday.com. Decide which fields can be included, especially when it comes to personal data. Read more in Is monday.com GDPR compliant?.
- What is a good answer? A hot lead or an urgent ticket has to be defined in words before the model can make the call. Write down the rules the way you would explain them to a new colleague.
- Who reviews? Should drafts be sent directly, or added as suggestions that a person approves? We almost always recommend a review step for anything that reaches customers.
- What format should the response have? Responses that will be written to columns must come in a fixed format, for example one of your segments and not free text. This is controlled in the instruction and checked in the flow.
- What happens when the model gets it wrong? A good flow logs every run, catches responses that do not follow the format and sends them to manual review instead of silently writing them in.
ChatGPT, monday AI or a custom agent?
The three options solve different problems, and many companies use more than one. monday AI fits when you want an AI step in an existing automation without connecting anything external. We describe how it works on monday.com and AI. The integration through Make fits when you need your own instructions, several steps or responses in a fixed format, and when the flow pulls data from more systems than monday.com. A custom AI agent is the next step: it works through several steps on its own, looks things up in your knowledge sources and makes decisions within limits you set. When that step is worth it is covered on AI agents. The integration is often combined with Gravity Forms for what comes in through the website. All the integrations we build are listed under all integrations.
Straviont builds the integration
We are an official monday.com partner and certified CRM specialists with monday.com, and we build AI steps between monday.com and OpenAI's models in Make. We start with the process: what should be sent to the model, what a good answer is and who reviews it. Only then is the flow built, with format checks, logging and error handling, and we manage it as your business or the models change.
Frequently asked questions
Can monday.com be connected to ChatGPT?
Yes. monday.com is connected to OpenAI's models through an integration platform like Make, which has an OpenAI module, or with custom code against the OpenAI API. monday.com also has its own AI blocks in monday AI and can be connected to ChatGPT through monday MCP.
What is the difference between the ChatGPT integration and monday AI?
monday AI is monday.com's built-in AI blocks that can summarize, categorize, extract details from text and translate directly in automations, using AI credits. The integration through Make gives you your own instructions, several steps and responses in a fixed format, using an OpenAI API account.
Can ChatGPT reply to customers automatically from monday.com?
Technically yes, but we almost always recommend that the reply is added as a draft that an agent reviews before it is sent. Responsibility for what goes out to customers should sit with a person.
What data is sent to OpenAI?
Only what you decide the flow should send. Each step in Make chooses which fields are included, and you can leave out personal data and sensitive customer information.
What do we need to connect the systems?
A monday.com account on a plan with integrations, an OpenAI API account and a Make account. If you use monday AI instead, no external account is needed.
When is a custom AI agent better than an AI step in a flow?
When the work requires several steps in sequence, lookups in your own knowledge sources and decisions within limits you set. An AI step does one thing per run, an agent keeps working until the task is done.
Next steps
Start by finding the steps where someone today reads, summarizes or sorts before the work can begin, and describe what a good result is and who reviews it. From there, building it goes quickly. If you want help, you can book a free mapping session and we will go through it together.