Every guide to this was written by somebody who sells one of these tools. So is this one. I build Bionic Forms, and it appears near the bottom, where a product with no customers belongs. What I can offer instead of a shortlist is the decision itself, made in the right order, plus one number out of my own logs that nobody selling a platform is going to show you.
Search for how to build an internal tool without a developer and you get one undifferentiated list. It is not one category. There are three, they solve different problems, and picking a vendor before picking a category is where the money goes.
Connector platforms. Retool is the clearest example. Its own pitch is that it builds interfaces over a database, an API or a model you already run. It is middleware. It does not hold your data, and it expects somebody who knows where the data lives and can point the tool at it. If you have a production database and an engineering team that is too busy for internal screens, this is exactly right. If you do not have a database at all, this category is not a cheaper option for you. It is a non-starter, and the trial will feel like your fault when it is not.
Spreadsheet-shaped databases. Airtable and its many relatives. The data lives with the vendor, the grid is familiar, and you can start in ten minutes. This is the correct answer far more often than people who dislike spreadsheets admit. The friction shows up later, in per-editor pricing and in the moment you want the thing to behave like an application rather than a nicer grid.
File-shaped databases. The Microsoft Access tradition, where the database and the application are the same object and the whole thing is a file you own. This is what a surprising number of small organisations actually run on, which is why the 13 October 2026 end of support for Access 2021 is generating so much traffic. Its virtue is that nobody can take it away from you. Its cost is that it is one person's Windows machine.
So the first question is not which vendor. It is: do I already have a database somebody else maintains, do I want a vendor to hold my records, or do I want to own the file?
Here is the number I promised.
I have built nine applications on my own software while developing it. Every single one that holds a serious quantity of records was filled by importing a file or loading a prepared set. The three that could only have been filled by somebody sitting down and typing hold 3, 7 and 10 records.
Three, seven and ten. Those are not abandoned experiments. They are tools I built deliberately, for myself, on software I wrote, and I still did not type the data in.
That is worth more than any feature comparison, because building the structure stopped being the hard part some time ago. An assistant can produce a workable set of tables and screens from a sentence, and several products now do it well. The part nobody finishes is the records.
So before you shortlist anything, go and find the file. A spreadsheet, an export, a report, a list somebody keeps in a document. If it exists, your real evaluation question is how well each candidate reads it. Does it guess types sensibly, does it tell you what it guessed, and does it stop and ask when a column is ambiguous instead of quietly ruining a date column.
And if there is no file anywhere, be honest that you are asking a group of people to start typing. That is a change in how they work, not a software purchase, and no platform on any list makes it easier.
If the valuable part of your tool is a procedure rather than the records, hire the developer.
Pricing logic with real edge cases, a scheduling algorithm, parsing an awkward supplier document, talking to a system whose integration is a fax number in spirit if not in fact. In all of those the records are incidental and the code is the product. Every no-code platform, mine included, offers automation as a fixed catalogue: a list of things that can trigger, a list of things that can happen. Configuration inside the catalogue is genuinely powerful. Anything outside it is refused. There is no escape hatch, and a platform that pretends otherwise is selling you a worse programming language.
Work out which half you are in before you look at a single price.
| If this is true | Look at |
|---|---|
| You already run a database or API, and have someone who knows it | Connector platforms (Retool and similar) |
| You want to start today and are content for a vendor to hold the records | Spreadsheet-shaped databases (Airtable and similar) |
| The records are the whole point, and you want to own the file | File-shaped databases (the Access tradition, and what I build) |
| The value is in a calculation or a procedure | A developer. Genuinely. |
| There is no file and nobody has ever kept this list | Nothing yet. Get the list existing first, anywhere. |
Bionic Forms is the third row. You describe what you keep track of, and you get a working application: records that link to each other, screens that follow those links, and a full history of every change. The desktop version keeps an encrypted copy of the data on your own machine and syncs when it can; there is a browser version that needs no download. Every plan has unlimited users, because charging per editor is what makes people share one login and stop using the thing properly.
Now the part that belongs in the same breath. It is pre-launch and has no paying customers, so anyone using it is early, and that is a cost rather than a badge. There is no scripting and no calculated fields. It imports plain CSV files, and it refuses an Excel workbook rather than guess what is inside it. There is no iOS or Android app, only the browser. The Windows build is not code-signed yet, so Windows warns you the first time you run it. If any of those matter to you, one of the other two rows is your answer, and I would rather say so here than after you have spent a weekend.
Try the decision, not the pitch. The demo builds a real application from a sentence, with no account and nothing to install. If you have that file, bring it. Watching a product read your own columns tells you more in two minutes than any comparison table.
Written 6 September 2026. The Retool description is from their own site; the Access 2021 end-of-support date is Microsoft's. The nine-application figure is my own usage of my own software and is offered as one person's evidence, not a study.