Build an n8n AI Agent That Logs Certificate of Insurance Requests

Step 1: Go to n8n.io

Certificate requests never stop. Most are routine. The risky ones look routine too, like a contractor who wants to be added as additional insured by Friday. Answer that with a quick sure, and your agency has promised something the policy may not have. We are going to build an AI agent that logs every request and flags the risky ones. First, you need an n8n account.

Step 2: Enter your email

This is just practice, so anything works — it won't actually send anything.

Step 3: Sign up

In the real n8n, it sends you a six digit code, then asks for a name, a password, and an account name. Go do that in your own browser later, then come back and we will build.

Step 4: Add credit first

Your agent needs a brain, and that means an API key from OpenAI. Go to platform dot openai dot com — that is the developer side, not ChatGPT, and it bills completely separately. Before you make a key, put some money on the account. A brand new key with a zero balance just returns errors. If you would rather not spend anything yet, n8n gives every new account free credits. In your own n8n, you can pick gateway credits on the model step and skip the key.

Step 5: Ten dollars is plenty

A little credit goes a very long way. You are charged per request, not per month, and logging a day of certificate requests costs fractions of a cent.

Step 6: Now the key

Balance sorted. Now go to API keys in the left menu.

Step 7: Create a secret key

Step 8: Name the key

Name it after whatever is going to use it. Future you, staring at a list of forty keys, will thank you. I am calling this one n8n agent.

Step 9: Create it

Default project, all permissions — that is fine for this. Create the key.

Step 10: Copy it now

Now pay attention, because this is the only time you will ever see this key. Copy it and paste it into n8n, or into your password manager. If you lose it, you do not get it back — you delete it and make another.

Step 11: Back to n8n

One last warning — this key can spend your money. Never commit it to code, never paste it into a chat, and if it ever shows up on screen while you are recording, revoke it before you publish.

Step 12: Start a new workflow

Now you are inside n8n. This overview page lists every workflow you have. Top right, click create workflow, and you land on an empty canvas.

Step 13: Add the first step

Every workflow needs a trigger — the thing that kicks it off. We want a chat trigger so you can paste in the day's certificate requests.

Step 14: Trigger on a chat message

A panel slides in with every way a workflow can start — manually, on a schedule, on a webhook. We want on chat message, so we can have a conversation with the agent.

Step 15: Back to the canvas

n8n opens the node so you can configure it. There is nothing we need to change here, so close it with the X in the top right and get back to the canvas.

Step 16: Add the next node

Now we decide what happens when a message comes in.

Step 17: Open the AI nodes

A panel opens and asks what happens next. The agent lives under AI, right at the top.

Step 18: Add the AI Agent

There it is — AI Agent. This is the piece that can plan and then carry out actions using whatever tools you give it. You give it the requests, and it can sort each one and save it to a spreadsheet — all in one step.

Step 19: Open the options

Right now this agent has no personality and no orders. It does not know it works for Harborview Insurance, or which requests need a second look.

Step 20: Add a system message

Pick system message. This is the standing instruction the agent reads before every request. It is where we tell it who Harborview is, how to log a certificate request, and what it must never do.

Step 21: Write the system prompt

Here is the actual prompt. It tells the agent what each entry needs: the date, the client, the certificate holder, the request in the client's words, and a check. Anything that asks for additional insured, a waiver, or special wording goes to the account manager. And the rule that matters most: never say a policy covers something.

You are the certificate request assistant for Harborview Insurance Group.

Harborview Insurance Group is an independent insurance agency. Its commercial clients email for certificates of insurance every day, usually because a landlord, a property manager or a general contractor asked for one.

The audience is the customer service rep (CSR), who reads every entry before any certificate is prepared.

The tone should be plain, brief and factual.

Your job is to turn each certificate request the user gives you into one entry.

For each request, return:
- Date
- Client (the business asking)
- Certificate holder (who the certificate is for, or "Not given")
- Request: what the client asked for, in their words, including any deadline
- Check: "Standard request", "Account manager review" with one sentence on why, or "Unclear" with what to ask the client

Use "Account manager review" whenever a request asks to add anyone as additional insured, a waiver of subrogation, primary and non-contributory wording, special limits, or anything else that may need a change to the policy.

When the user asks you to save the requests, use the Google Sheets tool to store each one.

Never say that a policy covers something, and never promise that a certificate will show wording or coverage. A certificate shows what is on the policy; the account manager decides what the policy can support. Never include policy numbers, driver's license numbers or other personal identifiers.

Step 22: Back to the canvas

That gives the agent everything it needs to know about logging certificate requests for Harborview. Close the node.

Step 23: Give the agent a brain

See that red triangle on the agent? It is complaining because it has no model. An agent can plan, but it cannot think without a language model underneath it.

Step 24: Pick OpenAI

OpenAI chat model. This is where that API key finally gets used.

Step 25: Open the credentials

And here is where that key you copied finally goes.

Step 26: Create a new credential

You want create new credential, right at the bottom. In your own n8n, if you skipped the key, you would pick gateway credits here instead.

Step 27: Paste the key

This is the one moment it matters that you copied it — remember, OpenAI never shows it to you again.

Step 28: Save the credential

n8n tests the connection straight away, and from now on this credential is available to every node in every workflow — you only do this once.

Step 29: Open the model list

Now the model dropdown.

Step 30: Choose the model

GPT five mini. Fast, cheap, and plenty for sorting certificate requests. Do not reach for the expensive model until you have a reason.

Step 31: Back to the canvas

That is the brain wired in. Close it up.

Step 32: Add memory

Next, memory. Without this, every message starts from nothing — you could not say save those, because it would not remember which requests you meant.

Step 33: Simple Memory

Simple memory is fine here — it keeps the recent conversation inside n8n, no database, no extra credentials.

Step 34: Back to the canvas

n8n opens simple memory so you can set it up. The defaults are fine. It keeps the last five messages. Close it with the X.

Step 35: Add a tool

And now the good part — tools. This is what separates an agent from a chatbot. A chatbot just talks. An agent does things. We are going to give it Google Sheets so it can save every request into a log your service team can work from.

Step 36: Search for Sheets

Google Sheets is not in the recommended tools, so search for it.

Step 37: Add Google Sheets

Google Sheets tool. Now the agent can write rows into a real spreadsheet, with columns matching what the system prompt tells it to return. When you say save those, each request becomes a row.

Step 38: Back to the canvas

Before we set this up, we need to actually create the spreadsheet. Close this node for now — we are going to hop over to Google Sheets and make one.

Step 39: Open Google Sheets

Switch to Google Sheets. You need to create the spreadsheet that this node will write to.

Step 40: Name the spreadsheet

First, name it so you can find it later.

Step 41: Type the first column header

Now set up the column headers. These need to match what the system prompt tells the agent to return. In A1, type Date.

Step 42: Client

B1 — Client.

Step 43: Certificate holder

C1 — Certificate holder.

Step 44: Request

D1 — Request.

Step 45: Check

And E1 — Check. Five columns, matching exactly what the system prompt tells the agent to return. Each request becomes a row in this sheet.

Step 46: Back to n8n

The spreadsheet is ready. Switch back to n8n and connect the Google Sheets node to it.

Step 47: Open the Sheets node

Double click the Google Sheets node to open it back up. Now we can connect it to the spreadsheet we just made.

Step 48: Add a credential

Now we need to connect this to a real Google account.

Step 49: Create a new credential

Create new credential — this one uses Google OAuth instead of an API key, so it is going to open a sign-in window.

Step 50: Sign in with Google

This opens the Google account chooser so you can pick which account to connect.

Step 51: Choose your account

Pick the Google account you want n8n to use. This is the account that owns the spreadsheet.

Step 52: Allow Sheets access

Google is asking what n8n is allowed to do. Check the first box — this lets n8n read and write your spreadsheets.

Step 53: Allow Drive access

Second box — n8n needs Drive access so it can find your spreadsheets by name.

Step 54: Allow external service

Last one — this lets Google connect to n8n as an external service. Check it.

Step 55: Sign in

All three checked.

Step 56: Pick the operation

Now we tell the node what to do. By default it reads rows, but we want to add rows.

Step 57: Append Row

Append row. Every request the agent logs gets added as a new row at the bottom of the sheet.

Step 58: Choose the spreadsheet

Now pick the spreadsheet.

Step 59: Certificate Request Log

Certificate Request Log — the sheet we set up with columns for the date, the client, the certificate holder, the request, and the check.

Step 60: Pick the sheet tab

Now pick which tab inside that spreadsheet.

Step 61: Sheet1

Sheet one — it is the only tab, so this is simple.

Step 62: Map the Date column

Now the columns. Each column has a purple button — click it and the AI agent figures out what data goes there.

Step 63: Map the Client column

Client. Same thing — click the purple button.

Step 64: Map the Certificate holder column

Certificate holder.

Step 65: Map the Request column

Request.

Step 66: Map the Check column

And check. Each one is now set to defined automatically by the model — the agent reads your column headers and figures out what goes where.

Step 67: Back to the canvas

The sheets tool is fully wired up — credential, operation, spreadsheet, sheet, and column mapping. Close it and let us test this thing.

Step 68: Open the chat

Everything is connected. Time to see if it works.

Step 69: Paste today's requests

Give it today's requests the way they actually arrive: one routine, one from a contractor with a deadline, and one that names nobody at all.

Log today's certificate requests. Oct 20, Delgado Landscaping: needs a certificate for Ridgeway Property Management for a new mowing contract, nothing special. Oct 20, Northpoint Electric: the general contractor, Keller Builders, wants to be listed as additional insured with a waiver of subrogation, and needs it by Friday. Oct 20, Summit Café: the email just says "the landlord needs our insurance thing again."

Step 70: Send the message

Send it off. The agent reads your system prompt and turns each request into one entry, with the certificate holder, the request in the client's words, and a check.

Step 71: Save to the sheet

Three requests. One is standard. The additional insured and waiver request goes to the account manager, because it may need a change to the policy. And for the café, it didn't guess the landlord. It says what to ask. Now tell it to save them. This is where the Google Sheets tool kicks in.

Step 72: Send it

Send it. The agent uses the Google Sheets tool to write each request as a row — date, client, certificate holder, request, and check.