Customer Story: Upstart

ABOUT UPSTART

AI-powered lending at scale.

Upstart Network, Inc. is the leading AI lending marketplace, connecting millions of consumers to more than 100 banks and credit unions. Upstart’s AI models and cloud applications help lending partners deliver better credit products to more people — approving more applicants at lower rates than traditional credit models allow.

Operating at that scale means a complex technology stack and a security team responsible for monitoring it all. Upstart’s security engineers needed consistent, reliable log data from dozens of SaaS applications to detect and respond to critical incidents. Getting that data in a usable form was eating far more engineering time than it should have.

‍

THE CHALLENGE

Building connectors was a full-time job in itself.

Upstart started by writing in-house connectors to their SaaS applications. The problems stacked up quickly: custom alerts for API failures, manual recovery for certain failure categories, ongoing maintenance every time a log format changed, and poor documentation from vendors on how to retrieve audit logs in the first place. There was no standardization or enrichment across sources, so even when the data arrived it required additional work before it was usable.

When Upstart ran the numbers, the math was clear. Building and maintaining 25 connectors in-house would require over a year of engineering time upfront, with ongoing maintenance costs every time an API changed. With 20 to 30 sources still on the backlog, that approach wasn’t going to scale. They needed a better path.

‍

THE SOLUTION

Onboard any log source in seconds, not weeks.

Upstart adopted Monad to bring SaaS audit logs online with zero overhead, eliminating both the upfront integration cost and the ongoing maintenance burden.

• Rapid onboarding: log source integration dropped from 1.5 weeks to 15 seconds
• Zero maintenance: all connectors maintained automatically, including API changes and version upgrades
• Collaborative development: new connectors shipped on Upstart’s timeline
• Automatic normalization: indicators of compromise normalized for easy correlation across sources

‍

“50% of the work in writing a new detection is in onboarding and normalizing that log source. Tarsal (now Monad) takes away that overhead, allowing us to focus on protecting our company, instead of wrangling API documentation. With Monad, we onboard logs in minutes, not months.”

Chris Schafer
Senior Engineering Manager, SecOps, Upstart

‍

THE RESULTS

More integrations. Less overhead. Better detections.

01      1.5 weeks became minutes for every new log source onboarded to the platform.

02      One new integration per month shipped continuously from Upstart’s connector backlog since contract start.

03      High-priority connectors in under two weeks, demonstrated during the proof-of-concept phase.

04      75% less engineering effort to onboard a log source compared to building in-house.

05      Detection quality improved because engineers now focus on threat logic, not API documentation.

‍