Unlocking the Power of Unstructured Data: Deasie's Journey in Data Governance for Language Models
In a world driven by data, the rise of language models powered by artificial intelligence has been nothing short of revolutionary. Enterprises are increasingly turning to these models for various applications, but they face a common challenge: ensuring the quality and compliance of the vast amounts of unstructured data they feed into these models. In 2023, a startup named Deasie emerged in New York with a mission to address this challenge head-on. In this article, we will explore the story of Deasie, the team behind it, the problems it aims to solve, and the solutions it offers to enterprises.
You might also be interested in: CreatorML and How It Works
Who Are the Visionaries Behind Deasie?
Every great startup begins with a visionary team, and Deasie is no exception. Let's meet the co-founders who have embarked on this mission to revolutionize data governance for language models:
Background: Reece Griffiths brings a wealth of experience to Deasie. With a background in information engineering from the University of Cambridge, he has worked as a Management Consultant at McKinsey & Company and served as a Product Manager for AI for Data Governance at QuantumBlack. Reece is also the founder of X26, a global entrepreneurs community.
Prior Experience: QuantumBlack (Product Manager, AI for Data Governance), McKinsey & Company (Management Consultant), X26 (Founder), University of Cambridge (Masters in information engineering).
Personal Blog: Reece Griffiths' Blog
Background: Mikko Peiponen is a technologist with a deep-rooted passion for data and machine learning. With over five years of experience in this space, he has previously worked at McKinsey and QuantumBlack, where he implemented and deployed models for large enterprises. Before co-founding Deasie, Mikko served as the Lead Data Scientist building AI data quality tools at QuantumBlack and holds a graduate degree from MIT.
Prior Experience: McKinsey and QuantumBlack (Data Scientist), MIT (Graduate Degree).
Background: Leonard Platzer's expertise lies in software engineering and machine learning. He has worked at QuantumBlack as a Software Engineer, focusing on AI for Data Quality products. Leonard's previous engineering roles at Amazon, Mercedes-Benz, and a ChatBot Startup showcase his diverse skill set. He earned his bachelor's degree from the National University of Singapore and graduated from UWC Mostar.
Prior Experience: QuantumBlack (Software Engineer, Machine Learning), Amazon, Mercedes-Benz, ChatBot Startup (E-Bot 7), National University of Singapore (Bachelor), UWC Mostar (Graduate).
With a team boasting such diverse experiences and a deep understanding of data, Deasie was well-equipped to tackle the challenges ahead.
The Birth of Deasie: Data Governance for Language Model Applications
Is Your Enterprise Data Ready for Language Models?
Enter Deasie, a platform founded in 2023 with a singular purpose: to ensure that enterprises feed only relevant, high-quality, and safe data into their language models. The team recognized a critical issue – the inability of enterprises to identify the right set of documents for their AI applications. This realization led to the birth of Deasie.
What Does Deasie Do?
Deasie's core mission is to filter thousands of documents based on key data quality dimensions. These dimensions include relevance, timelines, consistency, and bias. Additionally, Deasie runs checks to identify sensitive information within the data, ensuring that enterprises maintain data compliance. Ultimately, Deasie delivers the optimal set of data to the target Generative AI (GenAI) application, enhancing the reliability and effectiveness of language models.
Leveraging a Proven Track Record
Deasie's founders come with a track record of success. Prior to founding Deasie, they built an ML data governance tool from scratch at McKinsey and deployed it with 11 Fortune 500 companies. This experience served as a crucial foundation for their current venture, highlighting the demand for robust data governance solutions in the industry.
Problem: Unstructured Data's Challenge in Language Model Applications
The Uncharted Territory of Unstructured Data
For the first time, enterprises are venturing into the realm of unstructured data. This includes documents, reports, emails, and a myriad of other text-based resources. Unstructured data has become a goldmine for Generative AI use cases, but it comes with unique challenges.
Critical Questions Left Unanswered
As enterprises delve into this unstructured data, they are confronted with critical questions that need answers:
Does this data contain sensitive information?
Is this the most relevant data for this problem?
Are there inconsistencies in the data that could skew my results?
These questions are paramount to ensuring the success and ethical usage of language models, but without the right tools, they remain unanswered.
Solution: Deasie's Data Governance Platform
Automated Checks for Compliance and Quality
Deasie offers a comprehensive solution to the challenges posed by unstructured data. The platform provides automated checks for compliance, including Personally Identifiable Information (PII) and proprietary data. It also assesses the quality of the data, identifying irrelevant, untimely, or inconsistent information.
Reliable Governance for Language Model Use Cases
By implementing Deasie, enterprises gain the ability to reliably govern which data is used for specific language model use cases. This not only ensures compliance with data regulations but also enhances the effectiveness of language models, resulting in more accurate and valuable AI-driven insights.
The Team Behind Deasie
A Team of Innovators
The Deasie team is a group of seasoned innovators who have already made significant contributions to the fields of data governance and machine learning. Here's how their backgrounds align with the mission of Deasie:
Reece Griffiths: With a background in information engineering and experience as a Product Manager for AI for Data Governance at QuantumBlack, Reece brings a deep understanding of data quality and governance to the team.
Mikko Peiponen: Mikko's expertise as a Data Scientist and his experience building AI data quality tools at QuantumBlack make him a valuable asset in ensuring the quality of data fed into language models.
Leonard Platzer: Leonard's role as a Software Engineer, Machine Learning at QuantumBlack equips him with the technical skills required to develop and implement the data governance platform.
Deasie's Vision for the Future
Unleashing the Power of Unstructured Data
Deasie's journey is just beginning, but its vision is clear: to unlock the power of unstructured data for the upcoming wave of language model applications. By providing enterprises with the tools they need to govern their data effectively, Deasie is poised to play a pivotal role in the AI-driven future.
Deasie, with its founders' combined expertise and a laser focus on data governance for language models, has emerged as a beacon of hope for enterprises seeking to harness the full potential of their unstructured data. As the world of AI and language models continues to evolve, Deasie stands ready to lead the way in ensuring data quality, compliance, and, ultimately, success for its clients. With investors like General Catalyst and Y Combinator backing their mission, Deasie is poised for a bright future in the world of enterprise data governance.