Hacker News new | past | comments | ask | show | jobs | submit login
Show HN: I built a database GUI with ChatGPT integration (dbpilot.io)
89 points by Dennizz on April 30, 2023 | hide | past | favorite | 61 comments
Hey there! I’ve been working on DB Pilot for the last couple of months, and I recently added an AI assistant powered by GPT 3.5 to help you write SQL queries tailored to your DB schema.

Simply ask what data you are looking for - GPT will figure out which tables to use, how to join them, and then write a query for you.

The AI assistant knows which tables and columns exist in your database, meaning it can write queries specific to your schema.

Besides that, it doesn't have access to any actual data from your database though, meaning your data doesn't get exposed to OpenAI.




Very cool.

But nearly every time I've asked ChatGPT for code, it gives me either plausible-but-wrong code, or total hallucinations (flags and APIs that could exist, but don't).


I think there are a few tricks I use to get correct code:

1) GPT4 is a must, the code quality is so much better than GPT3

2) You can use a longer explanation than you think. Be detailed in exactly what you want. Clearly describe function inputs and outputs.

3) Provide GPT3 with the APIs to use, and break these down. If you want it to call an API and do some logic, you can prompt it twice - once to build a function to get data from the API and once to build a function that requests from the function in the first request. Do both of these as separate chat GPT sessions.

4) Give GPT a chance to fix errors - pipe the exact errors back in after you have run them so GPT can amend its code.

5) if you want the code to operate/be written in a particular way then tell GPT and even provide pseudo code if worthwhile. If you want error handling, request it. If you want the SQL statements to be run in a transaction, request it. It probably won’t do these things by default.

6) don’t use even slightly obscure languages.


Not my experience, nearly every time GPT 4 has produced code that compiles / runs first time without errors.


I've had mostly the same experience as calmoo.

And when it does go wrong, I've had good success telling it what it did do wrong. It will usually correct itself then.


Considering it is a one-time payment, it would be nice if I could use my own API key.

Although you call it a "no-tricks pricing" model, a 'one time lifetime license' for a product that I can't use the key feature of without a subscription after a year isn't a 'no-tricks pricing model'. Unless I am misunderstanding?

Although the product looks great - just maybe needs the words changing and making it clear what capabilities don't require the subscription.

Might even be better removing the term "no-tricks pricing model" too, because if anything the license model is a bit more tricksy and more ambiguous compared to most other applications imho.


That's fair. The "no-tricks pricing" wording was taken 1-to-1 from the UI template I used :)

I'll update the copy, and try to explain the pricing model more clearly.

Also a good idea to make it possible to use one's own API key. I'll consider that.


This is brilliant, but please do MongoDB. It's less intuitive to query and could benefit greatly from GPT


There are numerous libraries which can make querying MongoDB a bit nicer. Here are a few which may be worth considering, depending on what language is used:

C-sharp: https://github.com/mongodb-labs/mongo-csharp-search

Java: https://github.com/streamx-co/FluentMongo

Scala: https://github.com/osxhacker/scala-bson-query

Typescript: https://www.npmjs.com/package/ts-mongodb-orm

HTH


There's also this https://huggingface.co/spaces/rueckstiess/english-to-mql


At Veezoo (https://veezoo.com) we're about to launch a MongoDB connector leveraging Trino, hit me up if you wanna test it.


Hey, I'd love to try this!


Noted!


There have been a few products like this announced recently and they all have intro level examples on the home page. Can yours deal with things like entity-attribute-value model tables?


No, currently it doesn't.

This is a case where it could help to give the AI some example values of the actual data. The same applies for e.g. JSON fields.

I've been thinking about ways to make it possible for users to easily provide examples to the AI. Essentially that would mean selectively giving access for specific rows/columns to the AI.


the concern i would have here is that if you don't know SQL then you have no way of knowing whether or not the results are correct and if the results are wrong you have no way to debug the query.

to construct an SQL query you need to know the database schema, you need to be able to specify the relationships between tables, and you need to specify what data you are trying to retrieve.

the only way using plain written language can work is if you can infer the correct table names and relationships from the input text, which can only work if the database follows very strict naming conventions and/or proper foreign key usage and is small enough that none of the table or column names are ambiguous.

plain written language is much more ambiguous than a formal language like SQL.

how do you resolve ambiguities in the naming of tables and columns in the schema and in the input text for queries? if the output is wrong, is there an iterative process that can help the user refine the query?


>plain written language is much more ambiguous than a formal language like SQL.

This, by the way, is why programmers won't be out of a job due to AI any time soon... using an AI for non trivial generation of code just moves the code-run-debug loop from the computer language to AI prompts in English... which can be much harder to debug than e.g. Python.

Programmers will work hand in hand with large language models, but they're not going to be replaced with AI powered low code solutions any time soon.


Correct, you probably should know SQL.

In my experience though it is often a significant time saver if I only have to review the output of the AI vs. coming up with everything myself.

Re: ambiguities: You are right, the AI won't always be able to infer the correct fields to use. In such cases though it is often enough to help the AI out a little. E.g. by saying "information X is stored in field Y of table Z".

Even if you have to do that, you can still save time and more importantly mental effort by letting AI help you, compared to writing all the SQL yourself.

Re: iterative process: You can simply send a follow up message saying "You did X wrong". It is usually happy to correct itself.


I 100% agree. This is why I think that LLM's will help encourage better coding practices, not worse. The more that you can write "clean code" (things which are named correctly, follow the single responsibility pricinciple, don't violate the law of demeter, build the correct data-models/data-marts), then the easier it will be for LLM's to assist senior engineers in writing code/sql faster.

And if there is an interface given to people who will struggle to evaluate the correct-ness of something, ie: a Business Intelligence tool for people who don't know/care about the data model, then the question becomes "How can the tool facilitate getting the people who can verify the correctness of the sql/code do so in an efficient manner?" I'm thinking like a short-term pull-request, where the "code committer" is the LLM and the reviewer is a human. Said in another way: the chatbot is going in reverse, where chatgpt is asking the questions and the human gives the answers...... haha


>> the concern i would have here is that if you don't know SQL then you have no way of knowing whether or not the results are correct and if the results are wrong you have no way to debug the query.

> Correct, you probably should know SQL.

If your customers need to know SQL, presumably they also need to know the data architecture in order to verify correctness and/or fitness of purpose.

Assuming both of these preconditions are true, why would someone not just write the requisite SQL themself?

> Even if you have to do that, you can still save time and more importantly mental effort by letting AI help you, compared to writing all the SQL yourself.

Not really, at least in my experience. Staying "in the flow" is easier when writing SQL queries instead of having to:

1. Take time to think of a "good" ChatGPT request.

2. Review/test what was generated.

3. Take more time to refine the ChatGPT request to make it "better."

4. Goto 2 until a satisfactory SQL query is generated.

Contrast the above steps with:

1. Take time to determine what SQL query is needed.

2. Write the query.

3. Test the query.


> iterative process: You can simply send a follow up message saying "You did X wrong". It is usually happy to correct itself.

cool. i am interested to try it out.


This is the problem with these models in general, right?

I think it can help you write the queries faster but you have to know these things already. For example, at work, I can recall I need to join 5 tables to perform a certain lookup. It'd be cool if I didn't have to type all that, but without that knowledge, it would be useless.


Thoughts on a view?


Does this use a common ChatGPT account, or do users have to sign up themselves?

I find it depressing that so many people are willing to give their phone number to OpenAI.

FYI, there are a couple of typos on your homepage:

"It allows you to easily run SQL queries on a variaty of file formats"

"Combine the two, and you get a straigtforward way"

Good luck with the project!


Do you worry about phone numbers in the sense that it allows OpenAI to match data input with a person? Because at least in my country most phone numbers are public info available online together with our social security numbers, so it's already "shared".


No. I worry about it being used to spam my cell phone. I don't give that number out to Web sites or any other suspect parties. For that, I have a Google Voice number... which OpenAI refuses to accept. So they can lump it, along with their bogus "open" name.


Thank you!

It does use one common OpenAI AI account.

Thanks for the hints, typos should be fixed now.


Welcome! Thanks for the info.


This is looking nice. Would definitely try it if I was still doing SQL daily.

Small suggestion: show that it's a one time payment closer to the price. I kept staring at the price and thinking if it was yearly, until I looked at the button.


It’sa lifetime license for the gui but the AI is annual $49


Correct. When you buy the lifetime license you get one year of the AI included, afterwards it is $49 per year for the AI then.

I should make that more clear on the website.


Thanks for the feedback, I'll try to improve that.


Looks good - typo on the front-end: "AI assitant for your SQL needs".

By the way, is this a template you used here? Looks modern and similar to a few other apps.


Thanks for the hint!

Yeah I started with a landing page template from TailwindUI, then mixed in a few other building blocks also from TailwindUI, plus some customization.


Also variaty -> variety, "can not" -> "cannot", and there's at least one more place on the page that says "assitant" :)


Have you looked into having the AI assistant parse through the explain output and recommend changes to database structure?


Not yet but I have it in my list of TODOs!


Out of curiosity - do you think users/companies would feel secure knowing that the app uses a common API key?


If you mean the common OpenAI API key, I'm not sure how that's different in regards to feeling secure compared to separate keys. Would you mind elaborating further, please?

To clarify, the common API key is known only by my server, not by the app!


When will SQLite support be added? Then I'm all in! :)


This is cool, could you expand this into an API/SDK to plug the chat feature into existing apps?


Interesting idea. Do you have some examples of how you would like to use that, or maybe even an app where you'd like to integrate it?


I'd be interested in integrating it into a record management app we use as a feature for users, so they can ask questions about the data, not just the developers.


Any plans to get this on windows?


If I can get some traction on this, I plan to support Windows and Linux eventually.


Quick question out of curiosity: how do you handle the token limit of ChatGPT?


Right now it's pretty simple: First I drop the table/column information if the token limit gets exceeded otherwise (this is shown as a warning in the UI). Then if the limit still gets exceeded, the I show an error to the user.

I plan to make this process more smart eventually.


Curious, what software did you use to make your videos on the landing page?


They are made with https://www.screen.studio/


How to run this on linux?


So far it's Mac only. If I can get some traction on this, I plan to support Windows and Linux eventually.


Doesn't seem to work with a Supabase Postgres DB.


fwiw, there is an OpenAI assistant built into the Supabase dashboard now -

https://supabase.com/blog/supabase-studio-2.0#supabase-ai--a...


I just pushed a new release. It should work now!

To install the new version, just restart the app. It should then prompt you to install the update.


Thanks for reporting. I'll have a look what's missing.


Which language did you use to code the application?


Typescript and Rust, with SvelteKit and Tauri.


Do you mind telling where and why are you using rust?


How come you didn't use ChatGPT 4?


I'd like to use it eventually but as of now API usage is too limited.


Cool can I use LLama or BERT with it?


It's GPT-3.5 only so far but I might add additional models at some point.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: