What is Querri?
Querri is a new way of working with data. It looks like a spreadsheet with a prompt box, but under the covers is the very latest in AI, Large Language Models, and a robust data science toolkit. Basically, you load up some data and say what you want from it, from comparing the averages of different groups, to complex transformations, to rich graphs. There's a wide range of possibilities, and we can't wait to see what you'll do with it!
How do I get started?
If you don't have an account yet, fill out the "Request Early Access" form and we'll reach out to get you going. Once you're in, you'll be able to start uploading data and working with it using simple prompts!
How can I get help?
If you're in Querri and need some help, please send us an email at firstname.lastname@example.org and we'll be glad to assist.
What types of data does Querri work with?
Anything that can be downloaded as or converted to a CSV file can be loaded into Querri (and we're working hard at making other file types easy to bring in). This means basically any kind of tabular data.
What kinds of things can I ask Querri?
When you ask Querri to do something, we get to work trying to determine which of Querri's many "tools" can be used to answer your request. These tools include drawing graphs, writing formulas and instructions to manipulate the data or even the structure of your data table, replacing values, creating new columns based on formulas, or even running statistical analysis. Basically, if it's the kind of thing you feel like you might be able to do in Excel or some other data manipulation tool, give it a try and you may be pleasantly surprised. If you run into challenges, we'd love your feedback and can help you out at email@example.com
You can also check back here regularly for updated blogs and videos to walk through a wide range of use cases.
Are you going to use or share my data?
How does Querri keep my data secure?
Querri follows industry standards around data handling and storage, utilizing Amazon Web Service with data encrypted “in transit” and “at rest”, meaning we take this very seriously. We are currently working on our SOC2 certification as well to give you further peace of mind. It’s a bit of a process, but we’ll be sharing results as soon as we have this ready and are happy to talk through what we’ve done and how we’re addressing security in more detail any time you like.
If you're using AI, can I trust the results?
Most actions you take in Querri will use AI to understand your intent and to help process the instructions, but in the end the instructions are run as formulas or code which work reliably on every row of data or on the full set of data in the case of things like graphs or our statistical tools. Sometimes we write the actual values output by the LLM into a new column for you, and this will not always be completely accurate. After you run any prompt, it's a good idea to validate your result visually, or even with additional questions to Querri to ensure you got what you were looking for. If you're ever unhappy with the result of one of your instructions, just hit the undo button and try again.
What are large language models (LLMs)?
A large language model is an AI system that understands, generates, and manipulates human language. Developed through deep learning and trained on vast text data, it comprehends context, patterns, and nuances in text so it can answer questions, translate languages, and write content.
Does Querri use ChatGPT under the covers?
Querri performs actions based on your data and your requests in the interface. We use a variety of Open AI’s models to do things like infer which one of our tools you’d like to use, format the requests to those tools, and respond to your requests in simple language. In most cases, we are sending only a small sample of data and the column headings so that Open AI’s LLMs (which Chat GPT are based on) will be able to better understand your request. From there, we do most of our data processing internally using the same set of tools that a data science team might use to answer the question. The only time we would pass all of the data through Open AI is when you use our “row looper” tool which comes in when you ask Querri to do something like “loop through every row and..” you can create a new column to analyze sentiment analysis, categorize products, write sales pitches or all kinds of other things based on this data.
How is Querri different from ChatGPT?
In a way, you can think of Querri as ChatGPT for working with data. We use large language models like ChatGPT to understand your requests, and we’ve super charged the raw capability of LLM with a wide range of tools that a data scientist, data engineer or data analyst would use on a daily basis.
We then made it really easy to bring in your data and brought in familiar spreadsheet interfaces, an easy to follow history panel, and an easy way to draw graphs of all kinds. While you work, we also keep track of all the operations you take, building out an effective data pipeline that will allow you to automate your data workflows in a wide variety of ways, opening doors to a more automated business and saving considerable time on repeat tasks.
How is Querri different from extract, transform, and load (ETL) tools?
ETL tools are a powerful tool for automating data workflows. They allow users to visually program data pipelines and often provide a wide variety of integrations into other tools. They don’t usually provide easy ways to explore and analyze data in the process.
Querri can perform most if not all of the operations traditional ETL can using natural language, simplifying the “Transform” of ETL. It then has the ability to re-run these transformations as new data arrives. And Querri can do all of this without a lot of setup or technical knowledge. While we don’t have the breadth or integrations that many ETL tools have, we have a simple way to import and export data files which are compatible with most tools on the market. If you have a problem that can’t wait for an ETL project, Querri may be an ideal solution.
How is Querri different from querying my data lake or data warehouse?
Data lakes are incredibly powerful. However, they can easily require up to a year to set up, and still aren’t likely to have all the data you might want to work with in them. Querri does not require any setup. Simply upload your data file, and off you go!
We’re big believers in the concept of a Data Mesh over here, and Querri can help to enable your organization’s Data Mesh in a wide variety of ways by putting robust data capabilities in the hands of everyday business users. We have some exciting features coming around data governance, data libraries and data access both in Querri and through APIs. If you’re looking to build out your capabilities in this area, we’d love to hear from you and help you make it happen.
Got questions about Querri, AI or LLMs?
differently about how we use and benefit from data. We’re here to help you navigate this new world.