Introduction

The wait is finally over. We have now launched DinoAI, Paradime's LLM-powered AI co-pilot, into public preview 🚀

At Paradime, our goal is to empower analytics engineers so that they can devote:

• More time to creating actual business value, and...
• Less time to fighting mundane tasks.

Over the last few months, we have been running loads of experiments to see how we can augment analytics work. We want to help organizations boost their operational efficiency and employee productivity, while making analytics engineering more fun and exciting. We want to help data practitioners harness the transformative powers of Generative AI to get stuff done.

We all know how much boiler-plate work analytics teams do today, that costs organizations time, money, and reduced business outcomes. With that in mind, we are now launching a whole set of new features to help accelerate analytics development.

It's time to remove boilerplate and invest in delivering business impact! Let's accelerate analytics.

We are shipping AI capabilities inside our Code IDE to deliver a best-in-class co-pilot experience. This makes the Paradime Code IDE the most advanced AI-enabled IDE for dbt™ development.

Under the hood

Our engineering team has been working hard on building a safe and compliant infrastructure and reimagining the editor UI to create a smoother overall experience. We wanted to build a user interface that keeps users in the flow. No context switching and no time wasted. We have:

• Looked into the vector database market and chose Pinecone for our embedding storage layer.
• Built our LLM-pipeline so that all suggestions are contextual to the dbt™ repo and warehouse connected to the Paradime workspace. This ensures users receive relevant output from the LLMs.
• Designed a highly-available infrastructure using multiple LLMs, including both open-source and proprietary LLM models available.
• Put in a robust logging infrastructure so that we can provide inference logs for security-conscious organizations. We can also use inference logs to improve the user experience over time.
• And finally, we have put commercial agreements in place with our partners so that no data is used for model training purposes. This is to protect our users' data and IP at all times.

Now, let's take a look at a few product highlights.

Code-IDE

The development environment is where analytics work happens. It's where hundreds of our users build, model, and analyze their data models. However, as we all know, this process can be very time-consuming. Something as simple as date manipulations or figuring out the SQL for moving averages or calculating a metric for the last 28 days can take way too long. It requires plenty of trial and error to get it right and involves doing this through documentation, too.

In current macro trends, data teams are becoming smaller, so the workload is constantly increasing. It's therefore imperative for teams to find operational efficiency by reducing non-value-add work. This is where automation through AI can give resource-starved teams a much-needed productivity boost. For rightly sized teams, it frees up time that can be invested in more complex and transformative projects.

Write commits in seconds

Who likes to write commit messages? When we asked our users, almost nobody 😂.

In the Paradime Code IDE, writing well-written commits is now only one button-click away. This is one of my favorite features because I don't have to write good commits anymore. Now, I get it done in seconds, auto-magically.

Inline Co-pilot

How many times have you been writing code and all of a sudden you face a block? You need to know the syntax of an SQL statement. Particularly notorious is finding the correct syntax for dates. In Paradime, just:

ask dino;

Access your Dino co-pilot chat in line with your code, ask any questions you have, and with a click, you'll get the code. The prompts are automatically aware of what dialect of warehouse you are using, so you will always get the correct recommendation. Not happy the first time around? Regenerate to get a new suggestion.

And you get to do that in the flow while you develop. This is another personal favorite of mine because I could never get date functions right. Now, I have that available at my fingertips—literally. No more scouring through endless docs to write my SQL.

Auto-complete

Sometimes what you want is plain auto-complete while you are writing code. You know what to write, but you want to get there faster.

/dino/

This makes dbt™ development so much faster while you are in the flow of your work.

Chat panel

Another favorite of ours is having access to a co-pilot chat panel to get answers to questions on the fly. Users can get answers to all their ChatGPT-like questions from DinoAI. The interface is code-aware and has memory. You can ask any question you might have, from "How to book an Uber ride?" to "Give me code for three files for dbt™ models for source, staging and mart for Stripe data extracted using Fivetran". So, all the important questions in life. You are in complete control of what you want to ask.

Pricing

During the public preview, DinoAI will be completely free to use. We have already been receiving tons of feedback and improving the system as we go along. When the public preview is over, customers will be able to purchase DinoAI as an add-on, just like on GitHub. We will be pricing this pretty competitively in line with the market, and it will not be usage-based (don't worry) 😀.

Migration

DinoAI will be available out-of-the-box on Paradime's Code IDE.

If you are currently using dbt Cloud™, then you can migrate to Paradime in less than 5 mins. Just sign up for a Paradime account, connect your git repo and data warehouse to Paradime, and you're all set. There is no data or processes to migrate, and no information loss.

If you are currently using a local editor like VSCode, or PyCharm - the process is the same. Sign up for a Paradime account, connect your repo and data warehouse, and that's it.

On Paradime, you will be developing 50-83% faster with all the power-ups and AI boosters 🚀 🤩.

What's next?

This is one of the most exciting evolutions of Paradime. We don't want analytics engineers to become old dinosaurs belonging to the past; let DinoAI do that work for you.

This is just the start. There is so much left to do to realize the full potential of removing time wasted on boilerplate work from analytics engineering. The team is actively working on shipping features such as one-click SQL fixes, and even more automation and suggestions around the development experience. If you are an organization not afraid of change, get in touch with me.

We would love to get you on the path to accelerated analytics.

‍