Osmos AI Data Wrangler is now GA with self-configuration capability

The release of our AI Data Wrangler at Microsoft Ignite solidified Osmos as one of the first organizations to deliver a Public Preview Workload on Microsoft Fabric. Today, at FabCon 2025, we are proud to share that AI Data Wrangler is one of the first Microsoft Fabric Partners to reach General Availability (GA) along with our new, much-anticipated feature, “AI Wrangler Context.”

“With this release, Osmos’s AI Data Wrangler allows Microsoft Fabric customers to automate complex data cleanup. AI Data Wrangler now self-configures—eliminating complex setup. Osmos continues to redefine how businesses transform messy data into ready-to-use insights enabling customers to deliver greater impact.”
– Dipti Borkar, Vice President & GM Fabric, Microsoft. ‍

Implementing Data Transformation has always been hard, until now

Data transformation is a messy, error-prone process. Incoming data is often inconsistent, incomplete, and incorrect leading to countless hours spent on manual cleanup and/or on building and maintaining one-off brittle code. This is the problem we set out to tackle with Agentic AI. Read our launch blog to explore the full scope of these challenges and learn how Osmos AI Data Wrangler offers a transformative solution.

The problem

Cleaning and transforming data into a format that meets business needs, requires expensive engineering resources to do the following:

• Generate business requirements documentation that defines your target destination system, your source data and business logic for transformation.
• Design data schemas.
• Create destination tables.
• Write transformation code.
• Operationalize.

Osmos is on a mission to get customers from business requirements to a reliable data wrangling system fast by automating steps 2-5.

‍

Since the Public Preview launch of the Osmos AI Data Wrangler in November, we’ve seen overwhelming enthusiasm from enterprises across a broad range of industries. Some of the high impact use cases we encountered include:

• Retail/CPG: Ingestion of Product and Pricing data from hundreds of business partners into Fabric Lakehouse to optimize pricing and packaging strategy
• Manufacturing: Centralizing diverse data from 100s of ERPs into Fabric Lakehouse for better visibility into business operations
• Finance: Streamlining financial data ingestion to analyze spend for all suppliers across the organization
• Audit: Eliminating manual and expensive laborious processes from Accounts Payable workflow

Customers immediately saw the potential for tremendous value and savings - millions in savings and faster insights led to new market opportunities, and empowered business teams to see results quickly.

As part of onboarding, our customers shared a treasure trove of valuable documentation including business requirements, scripts, sample data and knowledge articles. These assets were accumulated over years to meet ever-changing business needs and customers wanted to train Wranglers using this knowledge.

We absolutely loved where this was headed!

What if the Wrangler could take advantage of all this knowledge?!

Better still, what if the Wrangler could train itself on all this information?!

Wouldn't that be game-changing? And wouldn't it accelerate the time to value?

We saw a powerful opportunity: leverage this existing material to configure Wranglers and better still, do that configuration automatically!

Introducing Wrangler Context – Better transformation with minimal configuration

Osmos generates guidance for Wranglers by analyzing information such as business documents and schema designs to generate guidance for AI Data Wranglers. We call this guidance, Wrangler Context. During Wrangler configuration, users point the Wrangler to documentation that describes business logic, code snippets used for data transformation, examples of good and bad data, and any other knowledge articles. The Wrangler analyzes the information and creates “Context” that it uses to guide itself when it is cleaning and transforming data.

Why is this game-changing?

Typically, translating business requirements into transformation/validation logic is a multi-month effort involving a cross-functional team of business analysts, data engineers, and business partners/customers. Wrangler Context converts that multi-month effort into a few minutes by self-learning from the provided information and setting guard-rails for itself in the form of validations. The user’s responsibility goes from coding this logic to refining and fine-tuning the context generated by the AI using natural language.

This combination of instructions, descriptors and validations ensures that the AI generates high quality output that meets business needs. Not to mention, the time to activate the use case goes from months to minutes. The result is high-quality data transformation that aligns with your company’s documented business practices from idea to implementation in minutes.

How It Works

Setting up an AI Data Wrangler is easy, and you already have everything you need. Create a Wrangler, upload your documentation, and review the auto-generated Wrangler Context.

1. Create a Wrangler and point it to your destination
2. Upload relevant documentation (business rules, sample files, etc.)
3. Review and refine the auto-generated descriptors, validators, and instructions

Wrangler Context

Wrangler Context is auto-generated using your uploaded documentation and includes instructions, column descriptors, and data validations.

✅ Descriptors – ensure that the schema level constraints are met.
✅ Validators – capture business logic that may be unique to different types of source data.
✅ Instructions – provides guardrails for AI to work within the constraints.

Your AI Data Wrangler is configured and ready to transform messy data into clean, structured insights. You went from requirements to a production data Wrangler with no code by using documentation you already have! As business needs change, simply point to updated documents and the Wrangler Context will be updated accordingly.

See Wrangler Context in Action

Want to see how it works? Watch our demo video to experience the power of Wrangler Context firsthand.

‍