Case Study
How PeakMojo Replaced Their AWS Stack and Unlocked Multi-Agent Workflows for Non-Coders
Company
Client
Bary Huang
Cofounder, PeakMojo
The Problem
Before Bubble Lab, PeakMojo’s data pipelines lived inside an AWS-based workflow stack. The engineering team was responsible for stitching together job queues, background workers, and a growing tangle of configuration — and any new pipeline meant another sprint of plumbing before a single document could be processed.
It worked, but it didn’t scale across roles. The data scientists who actually understood the healthcare scenarios they wanted to evaluate couldn’t ship anything end-to-end on their own. Every workflow had to go through engineering, every time.
The 0→1 Moment
Bubble Lab was incredibly easy to set up and immediately saved 40+ hours that PeakMojo would have spent standing up and wiring together the original AWS pipeline. But the bigger unlock came from who could now build.
“It enabled a 0→1 moment for us: my non-coder data scientists can now build complicated multi-agent workflows to evaluate and create healthcare scenarios by inputting thousands of documents — something that used to be a complete no-go unless engineers coded everything end-to-end.”
That shift — from “engineer-only” to “anyone on the team can ship a multi-agent workflow” — changed how PeakMojo actually does data science.
How PeakMojo Uses Bubble Lab
Bubble Lab now sits in three different layers of the organization — and each one gets a different kind of leverage from it.
Multi-Agent Healthcare Scenario Evaluation
Documents → Multi-Agent Pipeline → Scored Scenarios
PeakMojo’s data scientists feed thousands of documents into multi-agent workflows that evaluate and create healthcare scenarios. What used to require an engineer to wire up end-to-end is now a self-serve flow they author and iterate on themselves.
For: Data Science (non-coders)
Replacing the AWS Pipeline Infrastructure
Job queues + background workers + config → BubbleFlow class
Instead of maintaining a sprawl of queues, background jobs, and configuration across AWS, PeakMojo wraps each pipeline in a single, well-defined Python class with a built-in deployment framework. Bubble Lab is now their workflow engine.
For: Engineering
Explainable Pipelines the CEO Can Actually Read
Pipeline run → Step-by-step trace → Stakeholder visibility
Tian, PeakMojo’s CEO, gets visibility into exactly what the pipeline is doing at each step — turning what used to be a black-box data process into an explainable story she can take to customers. That clarity has translated directly into revenue.
For: Leadership
Engineering: From AWS Sprawl to a Single Class
For PeakMojo’s engineers, Bubble Lab didn’t just save time on pipeline setup — it replaced the pipeline infrastructure outright.
“Bubble Lab wraps the complicated job queues, background jobs, and configuration sprawl into a robust, well-defined Python class + deployment framework. It didn’t just save time — it effectively replaced our previous pipeline infrastructure, and we migrated to Bubble Lab as our workflow engine instead of maintaining our old AWS stack week after week.”
Instead of paying ongoing infrastructure tax on AWS week after week, the team now ships a class, gets a deployment, and moves on.
Leadership: Explainability That Converts
For Tian, PeakMojo’s CEO, the explainability of Bubble Lab workflows is a strategic asset — not a developer convenience.
“She’s happy to see what’s actually happening behind the scenes, and we can clearly explain our advanced data processing pipeline. That visibility directly helped convert into revenue.”
When PeakMojo can walk a customer through exactly what their pipeline is doing — step by step, with no black boxes — trust follows, and so do deals.
Final Thoughts by Bary
“Bubble Lab didn’t just save us time — it changed who on our team can build. Our data scientists ship multi-agent workflows on their own, our engineers stopped babysitting an AWS stack, and our CEO can finally see and explain what our pipeline does. It’s now the workflow engine PeakMojo runs on.”