Sliq
Automated data cleaning, in minutes, not hours or days.
103 followers
Automated data cleaning, in minutes, not hours or days.
103 followers
Fast and accurate AI data cleaning for engineers and analysts. Sliq auto-fixes formats, missing values, and schema issues. Get analysis-ready data, in minutes — not hours or days.
Sliq
Hi Product Hunt! 👋
I’m Daniel, the founder of Sliq!
What if Your Data Cleaned Itself?
This is the question we asked.
Our answer: Sliq.
The Problem: The "Janitor Work" of Data Science
If you work with data, you know the painful reality: Data Analysts and Scientists often spend 70-80% of their time just cleaning data.
Instead of training models, testing hypotheses, or uncovering the narrative hidden in the numbers, we are stuck fixing date formats, hunting down typos, and patching missing values.
It’s the "janitor work" of data science - it kills our momentum, and everyone hates it.
The Solution: Data cleaning that understands context
Sliq is an AI-powered tool that automates this process.
Unlike standard cleaning scripts that just apply rigid rules, Sliq understands the context and the domain of your data.
When you upload a dataset, Sliq analyzes the semantic context (whether it's financial records, e-commerce logs, or medical data) to understand what the data should look like.
It diagnoses the specific quality issues and builds a tailor-made cleaning plan specifically for your dataset.
Why Sliq?
Context-Aware: It detects errors based on the context and the domain of the data, not just rigid rules.
Tailored Pipelines: No generic fixes. It builds a cleaning plan tailored for your specific dataset.
Speed: Go from messy raw CSV to analysis-ready data in minutes, not hours or days.
We are in Beta!
We are launching the website in Public Beta today, and it is completely FREE to try right now. More advanced features and higher cleaning limits are coming in the full version soon.
I’d love for you to try it at https://sliqdata.com.
We need your feedback to make the AI even smarter!
I’ll be here all day answering questions.
Cheers, Daniel
@daniel_d7
hey Daniel — congrats on the launch. two questions i’m genuinely curious about:
my first question: how do you decide what’s “bad data” vs “signal”?
a lot of “messy” things are actually meaningful (e.g., missing values can be informative; outliers might be fraud/incidents; rare categories might matter). considering i don't want to look through the dataset prior to feeding the dataset, would AI pick these things up?
my second question: what’s the advantage vs just uploading to ChatGPT (or similar) and asking it to clean the dataset?
specifically:
- reproducibility (same output every run?)
- privacy / data retention (especially for clients analysing health records)
- handling larger files / multiple tables (data scientists/analysts may have to handle big data)
Sliq
@ravenasjournal Hi Ravena,
Sorry for the late response.
Didn't see that you left a comment.
About your questions:
Sliq is not just a one-size-fits-all algorithm that performs the same set of data cleaning steps on every dataset.
Sliq leverages the power of AI, learns each dataset separately, recognizes the specific errors and problems that each dataset faces, and designs a tailor-made data cleaning plan for this dataset, while always keeping in mind the specific domain knowledge needed to clean this dataset.
For example, if we have a dataset of salaries in a company, Sliq sees that one of the salaries is much higher than the others, so Sliq can make the assumption that this person is the CEO or some other, high-paying specialist, and Sliq will keep this "outlier" data point, as it is valuable.
On the other hand, if Sliq sees a negative or empty salary, this is probably some kind of error, and Sliq will try to fix this error by finding out what the most probable salary is for this specific employee.
Sliq won't just fill this cell with the mean salary.
Sliq will go to the lengths of doing statistical analysis of other employees with similar attributes (e.g, job title, experience, time in the company, etc.), and will use statistical methods to derive the most accurate salary for this specific employee, based on the salaries of other similar employees.
About your second question, I believe that my first answer explains well enough why Sliq is better than just giving ChatGPT your dataset and hoping for the best.
Sliq can also handle files of up to 10 GB in size (this is huge!).
If you have any more questions or would like to email me, you are more than welcome to email me at daniel@sliqdata.com
@daniel_d7 thanks for the informative answer - very interesting, sounds like big potential for analyst and data scientist workflows !
DeepTagger
Congratulations on the launch!
This is indeed a very complex task, data cleaning. From my experience it requires a lot of domain expertise. Are there any ways to give additional context to "cleaning gnomes"?
Also, are there any ways to set this up as a part of pipeline? I would imagine "one off" cleaning jobs would be rare.
Sliq
@avloss Hi Anton!
Thanks for checking out Sliq.
The answer to your questions is yes and yes!
Question 1: Are there any ways to give additional context to "cleaning gnomes"
Answer 1: On the "Start Cleaning" page, located at the bottom of the page, you will find a "User Instructions" field.
There, you can add instructions to steer the cleaning gnomes in the right direction.
If you want to give them more context about the dataset or the field, you can use the Dataset description, Dataset purpose, and the Column Guide fields.
Question 2: Are there any ways to set this up as a part of the pipeline?
Answer 2: Yes, you can use our Python library. Download it using:
The code to use this library is very simple. Scroll to the bottom of the landing page, and you can see examples of code for all the file types and datasets we support.
If you need any other file types or integration that we currently don't support, write to me, and I'll do my best.
To use the library, you first need to get your API key.
You can find it in the account page of the dashboard.
To get registered for free, sign up here
Happy Cleaning!
Daniel.
minimalist phone: creating folders
Looking forward to see the results, wishing you good luck after the extensive preparation!
Sliq
@busmark_w_nika Thanks, Nika!
You know it like no other how much I worked towards this launch, and I can say that people are already using it!
minimalist phone: creating folders
@daniel_d7 Here is only the beginning. Product Hunt can help with the visibility, but the rest of the work needs to be done outside.
Congratulations on the launch! But I have a question: could the service accidentally delete something it shouldn’t?
Sliq
@mykyta_semenov_
Hi Mykyta!
Sliq might unexpectedly delete a column because it thinks that the value there is not valuable for the purposes for which this dataset will be used, but for this exact reason, there is a "user instructions" field.
In the "user instructions" field, you can specify that you don't want Sliq to drop any of the columns in the dataset, and by doing so, you will guarantee that the data won't be lost.
love this. data cleaning is one of those things everyone hates but has to do every single day. would be amazing to see things like confidence levels on auto-fixes or a quick “review and approve” mode. congrats on the launch and best of luck today!
Sliq
@samet_sezer Hi Samet,
Thanks for the comment.
Indeed, data cleaning is a big problem we all face, and it takes so so much valuable time.
This is why we build Sliq.
About the things you've suggested, I will add them to the to-do list.
I would love to talk to you and see what other things and features can be helpful when doing data cleaning.
Hey @daniel_d7 congrats on the launch!
How does Sliq handle messy or unstructured datasets?
Sliq
@byalexai
Hey Aleksandar
Thanks a lot for your support!!!
You asked a great question.
When it comes to messy or unstructured data, Sliq doesn’t apply one generic set of rules. Instead, it looks at the context of each dataset, things like column meanings, value patterns, relationships between fields, and obvious inconsistencies, and then builds a tailor-made cleaning plan specifically for that dataset.
So rather than forcing your data to fit a predefined template, it adapts the cleaning steps to what the data is actually trying to represent. That’s what lets it handle real-world messiness instead of just “textbook” datasets.
Wow, Sliq looks amazing! Auto-fixing formats is a game changer. How does it handle inconsistencies in categorical data like state abbreviations, is there a fuzzy matching component?
Sliq