How to build your first application using Enable Banking, without knowing how to code
It starts with an idea. Maybe it's an Excel sheet that pulls in your business transactions automatically. Maybe it's a simple dashboard showing your balance across accounts. Maybe it's an automation that pings you when cash runs low. (All three are real examples below, ready for you to try.) But it feels out of reach in some way. You know what you want to build. You just don't know where to even start. Perhaps you’re assuming it requires you to be technical in some way or that you need to learn how to code first.
It doesn't. And more people are figuring that out every day. We keep hearing the same thing from people building on Enable Banking: they never even open our documentation. They paste the link straight into their favourite AI assistant (it's right here to make that even easier) and start building. It works, more often than you'd think, and much faster than expected. This guide follows that same approach, just with a bit more structure, so it works the first time instead of the fifth.
That's what this guide is for. You won't write a line of code. Instead, you'll learn how to describe what you want to an AI assistant, like Claude or ChatGPT, and give it the right context so it can build it for you, using Enable Banking's open banking API.
Enable Banking connects to 2,700+ banks across 30 European countries through a single API. It's built by developers, for developers. But with an AI assistant doing the coding for you, you don't have to be one yourself.
I ran this whole process myself, start to finish, and timed it. I'm not a developer. Every number in this guide is real, not an invented estimate to sound impressive.
You Don't Need to Learn to Code. You Need the Right Prompt.
This isn't new anymore. Half my friends are vibe coding, building their own tools without a technical cofounder in sight. I've watched someone on Instagram rebuild the core of an app he used to pay a hefty monthly fee for, using AI, just to prove it could be done.
And still, surrounded by all of that, founders as friends, a feed full of people clearly doing this, it feels slightly out of reach to me. Not hard. Out of my grasp. I didn't know where to begin. I don't think that's unusual. It's just not talked about as openly as the wins are.
Open banking APIs make that feeling worse before they make it better. They involve security tokens, redirect flows, technical steps that make "just prompt it" sound like something other people get to do.
What's changed is that you no longer need to understand any of it technically. A modern AI assistant can read the documentation, write the code, run it, and explain what happened in plain English, as long as you give it the right information up front. The gap was never the API. It was knowing where to start. That's the part this guide actually solves.
Before you prompt anything: five minutes of setup
These steps happen on the Enable Banking website, not with your AI assistant. They only need to be done once.
Step1: Create your account. Go to enablebanking.com/sign-in and sign in with your email. An account is created automatically the first time.
Step 2: Register an application. In the Control Panel, go to API applications and select "Add a new application." The form has a few more fields than you'd expect.
Here's what to actually do with each one:
Environment: leave Sandbox selected. This is a safe test mode with fake bank data, so nothing about your real accounts or money is at risk.
Private key generation: leave the first option selected, "Generate in the browser and export private key." It's already the default and the simplest choice.
Application name: anything you like, e.g. "My First Project." This name will appear at the top of your "API applications" list, and a test bank will show it to you later, when you "log in."
Allowed redirect URLs: this is the field that tripped me up. It's the web address you want your browser to land on after you authorise access to a test bank. You don't need a real website here. Enter something simple like https://example.com/callback. When you're redirected there later, nothing needs to actually load; you'll just copy a short code out of the browser's address bar.
Everything else (description, data protection email, privacy/terms URLs) is optional. Leave them all blank for now.
Click Register.
Step 3: Save your private key. When you click Register, your browser generates the key and downloads it as a file named after your application, something like 7c3a91f0-5e2d-4b8a-9f16-c4d0a8e372b5.pem — a long string of letters and numbers, unique to you.
Note: Depending on your browser, you may see a pop-up first, asking permission to download from enablebanking.com. Click Allow (or Save). It's just your browser's standard first-download check, nothing to worry about. That long string of letters and numbers in the file name is also your application ID. Keep this file safe. You'll tell your AI assistant where to find it in a later step. Never paste its contents into a public chat or share it outside your own AI conversation.
That's it. You now have everything your AI assistant needs to start building. This step took me five minutes total, including the time spent reading it.
The context prompt: paste this first
Before asking your AI assistant to build anything, paste this message so it has the full picture. You don't need to understand every line below, just fill in the blanks and paste the whole thing.
You're going to help me use the Enable Banking API, even though I'm not a developer.
Here's my situation:
- I have an Enable Banking account with a sandbox application registered.
- My application ID is: [paste your application ID]
- My private key file is saved at: [paste the file path, e.g. /Users/me/Downloads/aaaaaaaa-....pem]
-
The official quick start guide is here, please follow its authentication method exactly:
[https://enablebanking.com/docs/api/quick-start/] - I want to test everything against the sandbox first, using the "Mock ASPSP" test bank, which needs no login credentials at all.
-
I don't know how to code. Please:
- Write complete, working code (Python is fine unless you recommend otherwise).
- Tell me exactly what to install and which commands to run. I'm using [Windows/Mac].
- Explain what each step does in plain English before running it.
- Never hardcode my private key or application ID directly into a script that you'd save or share — read them from a file or ask me for the path instead.
My goal for this project is:
Below are three examples for that last line, though you can describe whatever you actually want built. Example 1 is the one I ran myself end to end, so it's the one with real numbers attached. Examples 2 and 3 follow the same pattern: a dashboard and an automation instead of a spreadsheet.
Example 1: Get your bank transactions into Excel
Who this is for: someone who wants to stop manually copying transactions from online banking into a spreadsheet.
Fill in the goal line with:
Pull my last 30 days of transactions from the sandbox test bank into an Excel file, with one row per transaction and columns for date, description, amount, and currency.
What your AI assistant should do: connect to the test bank following the quick start guide, pull your transactions, and write them into a .xlsx file it hands back to you. (Under the hood, this means calling a few of Enable Banking's endpoints in order, your AI assistant handles that part.)
What you'll need to do manually: open a link in your browser to "log in" to the test bank (there's nothing to enter: Mock ASPSP requires no credentials) and paste back a short code from the redirect page. Your AI assistant will tell you exactly when and how.
Once this works in the sandbox, the same script works on a real bank account too, once you're approved for production access.
How this actually went for me: filling in the context prompt with my real application ID and private key file path took around a minute, mostly hunting down exactly where the download had been saved. From there to a working connection took nineteen minutes, because Claude hit a real wall partway through: my sandbox app had no test transactions loaded yet, so there was nothing to authorise against. That's a one-time setup step (loading Enable Banking's sample transaction data into the Mock ASPSP), not a flaw in the process, and it was the one point where I had to make a call myself instead of just clicking through. A further thirteen minutes got the actual export running cleanly.
Call it thirty-two minutes, start to finish, for a real spreadsheet of transactions. Since it had gone faster than planned, I pushed further: six more minutes for basic categorisation on top (incredibly rudimentary; just under half the transactions came back uncategorised), and four more for a chart of spending by category. Under an hour, total, for an export, a categorisation pass, and a chart I now check more than I anticipated.
Thirty-two minutes, start to finish, for a real spreadsheet of transactions. It had gone so well, I kept going. Six more minutes got rough categorisation (incredibly rudimentary; just under half the transactions came back uncategorised). Four more got a chart of spending by category. Under an hour, total, for an export, a categorisation pass, and a chart of spending.
If you hit the same wall: Enable Banking provides ready-made sample transaction data for exactly this situation, so you don't need to build a test dataset yourself. Ask your AI assistant to import Enable Banking's synthetic sample dataset into the Mock ASPSP setup in your control panel, and it can walk you through the same import step I did.
Example 2: Build a simple dashboard
Who this is for: someone who wants a single screen showing balance and recent activity, instead of logging into their bank app.
Fill in the goal line with:
Build a simple, private webpage (only I can see it, running on my own computer) that shows my current balance and a table of my 20 most recent transactions from the sandbox test bank.
What your AI assistant should do: build a small local web page (it may suggest a lightweight framework; let it choose, and ask it to explain why), pull the same information as Example 1, and display it as a balance figure and a table.
A good follow-up prompt once it works: "Now add a simple bar chart of spending by week" or "add a second bank account and show both side by side." AI assistants are good at iterating. You don't need to plan the whole thing in the first prompt.
Example 3: Automate something
Who this is for: someone who wants to be notified automatically instead of checking manually.
Fill in the goal line with:
Check my balance from the sandbox test bank once a day and send me an email if it drops below an amount I choose.
What your AI assistant should do: write a script that checks your balance on a schedule, compares it to your threshold, and sends an email (it'll likely ask which email service you want to use).
Other automation ideas to try the same way:
Every morning, send me a Slack message listing yesterday's transactions over 100 EUR.
When a payment comes in with "invoice" in the description, add a row to a Google Sheet automatically.
A quick word on security
Good documentation isn't just convenience; it's a security feature. When your AI assistant follows Enable Banking's own quick start guide instead of improvising its own authentication logic, it's using a pattern that's already been built, reviewed, and tested, rather than inventing one from scratch. That matters more than it sounds like it should. A lot of security problems in home-built integrations come from someone, human or AI, solving an already-solved problem slightly differently, and getting one detail wrong.
That said, none of this makes using an AI assistant risk-free, and I don't want to wave that away. Claude, ChatGPT, or any other LLM can still write code with mistakes in it, mishandle a file it shouldn't, or behave unpredictably in an edge case. This guide doesn't erase that risk, it just shows how much of the process is now genuinely within reach for someone with no coding background. Treat your AI assistant's output the way you'd treat a first draft from a junior teammate: often very good, and still worth a second look before it touches anything real.
Tips for prompting AI about a banking API
Give it the docs link every time you start a new conversation. Open banking APIs have specific security requirements that a general-purpose AI won't know unless you point it to the quick start guide.
Paste errors back exactly as they appear. Don't summarise the error message yourself. Copy and paste it. AI assistants are far better at diagnosing a real error than a description of one.
Ask it to explain before it runs anything. A one-line request, "explain what this does before running it," keeps you in control and helps you learn as you go.
Start in the sandbox, always. It uses fake data from test banks, so there's no risk while you're getting the AI's code right. Move to a real account only once everything works end to end.
Never paste your private key's contents into the chat. Give your AI assistant the file path instead, and let it read the file locally if it has that capability. Treat the key like you would a banking password.
Good to know
Enable Banking only ever uses banks' official, regulated PSD2 connections, and it never monetises any of your data. It's ISO 27001 certified. That matters here for a simple reason: you're about to connect an AI assistant to your real financial data, and it's worth knowing the layer underneath it is built to the same standard banks themselves are held to.
From sandbox to your real account
Once your AI-built script, dashboard, or automation works against the sandbox, moving to your real bank account involves two things, both covered in Enable Banking's docs:
Getting access to the production environment so you can connect your own, real accounts.
Re-running through your AI assistant, but pointed at production instead of sandbox: a change of one setting your assistant can make for you.
Quick reference: what these terms mean
| Term | In plain English |
|---|---|
| ASPSP | The bank or financial institution you're connecting to. |
| Sandbox | A test mode with fake bank data — safe to experiment in. |
| Private key / JWT | The security credential your app uses to prove it's really you. Like a password, but for your application rather than for you personally. |
| Session | Enable Banking's record that you've been given permission to see a specific account's data. |
| PSD2 | The EU regulation that requires banks to allow this kind of secure, authorised data access in the first place. |
This guide is a starting point, not a substitute for Enable Banking's full documentation. If your AI assistant gets stuck, the docs and Enable Banking's support team are the next place to look.
I built the version above in under an hour, most of it spent reading. If an idea's been sitting on your list because you assumed otherwise, this is your sign to actually time it.
Explore More:
You don't need to learn to code to build on Enable Banking's API. You need the right prompt. I tested this myself: a real transaction export to Excel, a balance dashboard, and an email alert for when cash runs low, all built by an AI assistant, all timed. Here's exactly how it went, and how long it actually took.