Case Study

How Parabola Powers Natural Language Workflow Automation With Daytona

100+

hours of engineering time saved on infrastructure build alone

120k

sandboxes created weekly with Daytona

15%

of workflows supported by Daytona

Parabola is a data automation platform that turns manual processes into AI workflows any team can build, run, and maintain. Teams at Lyft, WHOOP, and hundreds of other companies use it to automate recurring work like invoice audits and month-end close.

Headquarters

San Francisco, CA

Industry

Software Development AI & Automation

Department

Engineering Product

Key Features

Sandbox Creation Speed Sandbox Statefulness Infrastructure Scale

parabola.io

Learn how this AI-powered workflow automation platform partnered with Daytona to provision the isolated, stateful sandboxes needed to accelerate its product roadmap.

Daytona’s sandbox infrastructure cleared the bar on isolation, scalability, and support. For a team using AI-generated code across a growing user base, these capabilities are everything.

Sam Coxon

Engineering Manager at Parabola

01 -- CHALLENGE

Shipping AI-Powered Workflow Automation Required Secure Sandbox Infrastructure

After building a flexible no-code workflow builder, Sam Coxon, Engineering Manager at Parabola, and his team set out to develop Prowork, an AI feature that enables users to automate data-intensive processes via natural language. Instead of navigating Parabola’s step toolbar, users describe their workflow goals and let the AI generate code defining the steps to ingest, transform, and route data. Ensuring that AI-generated code runs safely and efficiently at scale made robust sandbox infrastructure a core requirement.

Every AI-generated step is unique to the user and generated at workflow-build time. Running it in an isolated sandbox would confine dynamic code to tightly scoped environments, extending Parabola’s established security controls and reinforcing its HIPAA- and SOC 2-compliant posture.

Isolated execution couldn’t stall under load. Users could trigger a flow at any moment, and as Parabola’s user base grew, so did the number of code runs the platform had to handle concurrently. To meet demand, Sam and his team needed isolated, reproducible environments that could scale with demand.

Because each workflow step in Prowork operates on the output of the one before it, the sandbox running the flow had to be stateful across the entire chain. With sandbox provisioning sitting between user requests and output, startup times had to be fast enough to maintain the user experience.

Building infrastructure with microVM configurations, Kubernetes orchestration, and networking would’ve taken weeks of engineering work upfront, with even more time required for ongoing maintenance. With Parabola’s integration layer written in TypeScript, Sam and his team would also have to create a native SDK layer from scratch. So to avoid pulling his engineers away from other roadmap initiatives, he set out to find a managed sandbox infrastructure provider.

After evaluating several providers, Sam narrowed his options to two solutions. One stood out with the security credentials, speed, and scalability that Parabola needed to launch its AI workflow feature with confidence. That solution was Daytona.

We needed a secure environment to execute code we didn't write and couldn't fully trust. We knew that managing sandbox infrastructure in-house would take significant effort, so we turned to dedicated providers. Daytona emerged as a top choice.

Sam Coxon

Engineering Manager at Parabola

02 -- SOLUTION

Isolated Sandboxes That Unlock Scalable AI Automation on Demand

From Daytona’s audited access controls to its SOC 2 Type I and HIPAA certifications, the platform met each of Parabola’s compliance requirements. Confident in Daytona's security posture, Sam and his team integrated Daytona into Parabola’s architecture using the platform’s TypeScript SDK. At every step, the Daytona team was readily available to answer any questions and provide hands-on support.

With Daytona, Parabola now has the sandbox infrastructure to efficiently bring Prowork to production. Daytona provisions an isolated sandbox for every workflow that Prowork’s AI generates, adding a dedicated layer of isolation on top of Parabola’s robust security architecture.

As the AI encodes each workflow step, the sandbox state persists. As a result, users can turn a plain-language description into a fully built, production-ready workflow that matches their needs.

That foundation extends across Parabola’s entire user base. Daytona’s concurrency enables Parabola to provision dozens of sandboxes simultaneously, so users build workflows instantly without contention. As a result, Prowork scales with demand without sacrificing speed.

Users experience minimal latency regardless of platform demand. Using Daytona’s Declarative Image Builder, Sam and his team pre-configure OS packages, runtimes, and dependencies for sandboxes supporting standard data operations. That way, Daytona always keeps a pool of pre-warmed sandboxes that launch within approximately 800ms of a workflow being triggered.

While the platform checked every box, Daytona’s team cemented the partnership. From answering platform questions to handling paperwork for Parabola’s data processing compliance, the Daytona team provides quick, hands-on assistance that keeps Sam and his team focused on their roadmap. When Parabola hit rate limits during testing, Daytona clarified exact error codes and failure behavior so Sam’s team could build reliable retry logic, hardening Prowork for production.

Sandboxes that start quickly and retain context between runs are critical for our AI-powered platform feature. Daytona provides us with that speed and statefulness regardless of how many sandboxes we’re running concurrently.

Sam Coxon

Engineering Manager at Parabola

03 -- RESULT

Parabola Provisions 120k Sandboxes Every Week With Daytona

With Daytona, Parabola runs its AI workflow feature on secure, scalable sandbox infrastructure, all without introducing engineering overhead. Sam’s team can stay focused on their broader product roadmap, while Parabola’s users benefit from an intuitive way to automate complex data processes.

100+ hours of engineering time saved on infrastructure build alone
120k sandboxes created weekly with Daytona
15% of workflows supported by Daytona

As Parabola expands its AI workflow step library, adds integrations, and reaches new customer segments, Daytona will ensure AI-generated workflows run reliably at scale. Sam is especially excited to use Daytona’s resource management capabilities to keep infrastructure costs optimized as demand increases.