Product

Role-based access controls in Snorkel Flow secure enterprise data

May 14, 2024
4 min read

Our customers have made clear that they need role-based access controls, and we’ve listened. As a product lead at Snorkel, I’m thrilled to share that we’ve released a new feature in our 2024.R1 product update specifically focused on data upload access controls.

This development is a direct response to feedback from our enterprise customers who need greater control over what data enters and leaves their systems. We’re confident this new capability will significantly enhance our enterprise readiness.

Let’s take a look at how it works.

Data access controls: an enterprise necessity

Enterprises (especially banks and firms in other highly regulated industries) need to control what data leaves their organization and how. Administrators must gate and approve the egress of proprietary data to prevent conflicts with the company’s security guidelines and federal regulations. If they allow employees and teams to arbitrarily access data and bring it onto an external platform, they could quickly run into a problem.

This, of course, includes moving data from the enterprise environment into the Snorkel Flow AI data development platform.

The top-level dashboard for role-based access controls in Snorkel Flow

Snorkel’s response: role-based access controls

To address our customers’ data access control needs, we have begun developing and deploying new features that allow admins to restrict what types of data can be imported into Snorkel Flow. Admins can now disable or enable different data connectors, effectively locking down what data your users can import into the platform.

Administrators can configure ingress controls both by role and by workspace. For example, they can enable access for SQL for data annotators in one workspace, Snowflake for data scientists in another workspace, and lock down local file upload for all roles across all workspaces.

Editing a rule using Snorkel Flow's Role Based Access Control utilities.

Control down to the configuration level

These features allow administrators to further refine access controls. For each role, each workplace, and each data upload method, administrators can decide which roles can interact with connection configurations and how. For example, they can choose to only allow administrators and super administrators to add or edit configurations for the company’s SQL database. They can allow developers to view and use existing configurations without modifying them. Still others, like annotators, could be locked out of all data ingress entirely.

This multi-level configuration control allows specificity in managing data access, ensuring that only the right people access the right data, enhancing both security and operational efficiency.

These granular controls are a testament to Snorkel’s commitment to providing robust and flexible solutions for enterprise data management. We understand that every enterprise has unique needs and challenges, and we are dedicated to providing tools that can be tailored to meet these specific requirements.

Showing the results of controlling upload settings using Snorkel Flow's role-based access control tools.

Role-based access control impact on enterprise AI

These new features provide several benefits. First, it significantly improves data security by ensuring that only authorized users can import data into Snorkel Flow. Second, it ensures data integrity by controlling what data can come in and out of the system. Lastly, it can improve efficiency in workflow, by preventing Snorkel Flow users from stumbling down the wrong path with the wrong data.

This release is just the start. Over the next year or two, we plan to offer the ability to gate both data ingress and egress. We also plan to investigate improvements to feature access control. This will lead us to a world where admins can define their own custom roles that align closely with their active directory representation of their entitlements.

RBAC in Snorkel Flow: a step forward

Snorkel Flow’s new data upload access control features represent a significant step forward in enhancing our enterprise readiness. It gives our customers much-needed control over what data comes in and out of their systems, improving data security and integrity while also enhancing workflow efficiency.

We encourage all our customers to explore this new feature and see how it can benefit their operations. As always, if you have any questions or feedback, feel free to reach out to us. We’re excited to hear from you and look forward to continually improving Snorkel Flow to better serve your needs.

Ready to accelerate AI development?

Deploy production AI and ML applications 10-100x faster with Snorkel’s experts, using our proprietary technology.

Request a demo

Share this article
Image
Daniel Xu
Product Lead

Daniel Xu is Snorkel AI’s infrastructure product leader with extensive experience in building enterprise and security platforms. 

Previously at Facebook, he led two teams in Privacy infrastructure, focusing on protecting user data through identifying and implementing anti-scraping mitigations on both 1st party apps and 3P developer platforms. Daniel has also served as Tech Lead and Product Manager at DataVisor focused on deploying anti-fraud / anti-bot solutions for both US F500 and Chinese companies, and held a Product Manager role at Addepar specializing in building financial analytics platforms for wealth managers and family offices.

He holds a Bachelor of Applied Science from the University of California, Berkeley, providing a strong academic foundation. Daniel excels in driving highly technical product strategy, leading teams, and delivering innovative solutions in data-driven environments.

Recommended articles

View all articles
Stylized illustration of a glowing neural network hub rising above a grid of abstract city blocks, representing an AI agent operating inside a structured enterprise environment.
Enterprise environments and training AI agents for real-world workflows
Most agent benchmarks still evaluate a thin slice of the job. The agent receives a task, produces an answer, gets scored, and the episode ends. Enterprise workflows work differently. An underwriting agent may need to read policy documents, inspect customer records, call internal tools, ask a simulated user for missing information, update state, and follow approval rules. A correct final
July 14, 2026
Ramya Ramakrishnan
Image
Milestone-Based Evaluation and Training for Long-Horizon AI Agents
Long-horizon agents are outgrowing pass/fail evaluation. A short task can be scored as one outcome. A longer workflow is different. The agent may plan correctly, navigate to the right place, call the right tool, and still fail during execution. A final failure score does not show which phase broke. Training has the same shape of problem. Reinforcement learning for long-horizon
July 9, 2026
Zhengyang (Jason) Qi
Image
Agents’ Last Exam: AI Benchmarking for Real Work
At our latest Snorkel AI Reading Group, Yiyou Sun and David (Xinyang) Han (UC Berkeley, Center for Responsible and Decentralized Intelligence) presented Agents’ Last Exam (ALE) — a benchmark designed to evaluate AI agents on long-horizon, economically valuable, real-world tasks with verifiable outcomes. ALE is a collaboration between Berkeley RDI, Snorkel AI, and 300+ expert contributors across 55 professional subfields. ALE asks a deceptively simple question: can
June 29, 2026
Snorkel Team
Image
Image

Join our newsletter

For expert advice, the latest research, and exclusive events.
By submitting this form, I acknowledge I will receive email updates from Snorkel AI, and I agree to the Terms of Use and acknowledge that my information will be used in accordance with the Privacy Policy.