Tulsa Hackathon
Musa Capital Tulsa Hackathon
Submit your innovative multi-agent AI solutions.
Event Details
Submission Guidelines
Ensure your project aligns with the hackathon themes.
Provide clear and concise descriptions.
Working demos are highly encouraged.
Originality and innovation will be key judging criteria.
Cast Your Vote!
Support your favorite projects by voting.
Vote Now
Voting closes on July 20, 2025.
Project Submission
Fill out the form below to submit your project for the Tulsa Hackathon.
Submissions
All Submitted Projects
Browse the projects teams have entered so far.
Factory Flow AI
Team: Mighty One
Members: Solei W. (Virtual attendee)
Problem: Factories lose a ton of money when bad products slip through or machines suddenly stop working. A single breakdown can throw off the whole schedule, causing wasted time, wasted materials, and unhappy customers. Right now, most systems catch problems too late instead of preventing them.
Solution: The solution: a multi-agent quality assurance system that acts like three smart assistants on the factory floor.
- A Vision Agent watches products in real time and spots defects.
- A Prediction Agent monitors machine data and warns when equipment is about to fail.
- A Workflow Agent takes that info and adjusts schedules so production keeps moving.
Together, they cut down on waste, reduce downtime, and save manufacturers serious money. And it’s built it with responsible AI in mind… humans stay in the loop, and data stays private.
Police Operator Calls
Team: AITechies
Members: Zakie, Marzook, Daisha, Desiree, Steve
Problem: A Police caller co-pilot
Solution: It’s a 911 “co-pilot” that helps call-takers make faster, safer decisions: it reads the caller’s words, pulls out key facts (what, where, how severe), searches your local SOPs, and suggests a dispatch plan (EMS/Fire/Police and priority) with a brief why. Humans stay in charge—agents only recommend, and the system flags unclear or high-risk cases for supervisor review. Centers buy it because it cuts decision time, reduces over/under-response, improves consistency and compliance, and creates clean audit trails and training data. The business model is simple SaaS (per seat or per incident) with optional integration and analytics add-ons. In short, it makes the first 30–90 seconds of every emergency call smarter, faster, and more consistent without replacing people.
Integral Logistics
Team: DBJ
Members: Bre'Shawna Chambers, Jasmine Royal, Dalmo Mendonca, Marques Chambers, Jamie Johnson
We lost access to the labs at 4:55 PM before we could get it to github! :((
Problem: Traditional logistics systems focus almost exclusively on external metrics—loads per day, cost per mile, route efficiency—while ignoring the inner human and cultural factors that drive those outcomes. Pre-AI dashboards and rule-based BI tools can’t parse free-text driver journals, synthesize qualitative with quantitative inputs, or integrate across organizational culture and individual behavior. The result is systemic blind spots: driver fatigue, unsafe cadence patterns, and vendor misalignment remain invisible, bleeding millions in detention, idle time, and missed deliveries.
Solution: Our team built a prototype for a 5-agent system in Azure AI Foundry: four specialist agents aligned to each AQAL quadrant (Individual-Internal, Individual-External, Collective-Internal, Collective-External) and an Integral Orchestrator that fuses their insights. One agent embeds driver sleep narratives into fatigue vectors; another detects unsafe cadence and prescribes chronotype-aligned breaks; a third agent evaluates vendor culture feedback to predict dock friction; yet another agent forecasts throughput and detects capacity cliffs. We used fake sample data generated from ChatGPT across all 4 domains. The orchestrator merges these signals into traceable, explainable 90-day recommendations with PII scrubbing, bias checks, and audit logs. This solution directly hits the judging criteria: measurable impact ($1.7M/year savings assuming a fleet of 250 truck drivers, –18% idle, +7.5% OTD), responsible AI safeguards. We didn't have a chance to present a working demo, but we were very proud of how close we came in just a few hours. This could be a serious weekend project to create something that would be extremely valuable to local Tulsa companies like Tenstreet. This has even wider applications, including the military, government contractors, and road accident prevention.
Warehouse ShiftCast
Team: A Tribe Called Knowledge
Members: Gjarred Baker, Ryan Anderson, Stephen Dennis, Draper Sturdivant, Nicole Halper
Problem: Warehouse shift changes are high-risk moments for miscommunication. Critical updates—such as machinery malfunctions, task progress, staffing changes, and safety incidents—are often relayed verbally or in unstructured notes. This leads to lost tribal knowledge, repeated mistakes, and unsafe working conditions.
Our project, ShiftCast, solves this by using Microsoft Azure Foundry to orchestrate a multi-agent system that captures, processes, and delivers key operational knowledge at shift handover. It converts spoken updates into structured data, flags safety risks, and automatically generates a 2-minute audio “shift podcast” for the incoming team. This ensures continuity, accountability, and safety—without relying on memory or manual reporting.
Solution: Our solution, ShiftCast, is a multi-agent AI system built with Microsoft Azure Foundry that automates and standardizes shift handovers in warehouse operations. It addresses the communication gap between outgoing and incoming teams by capturing unstructured verbal updates and transforming them into structured, actionable summaries.
The system tracks and processes updates across four key operational domains:
Inventory Management – Tracks order volume spikes, low stock alerts, and restocking priorities.
Equipment Repair Status – Monitors machine faults, repair progress, and outstanding maintenance tickets.
Employee Attendance & Task Delegation – Logs call-outs (e.g., illness, PTO) and dynamically reassigns critical tasks.
Shift Checklist Completion – Captures daily operational checklists (e.g., zone cleanups, order packing status) as structured progress indicators.
The system is orchestrated through four specialized agents:
Outgoing-Shift-Agent: Captures voice notes using Azure Speech-to-Text, parses them into structured data (inventory, repair, HR, checklist items).
Machine-Agent: Cross-references equipment-related data to validate and enrich repair updates (via simulated or live CMMS/WMS integration).
Safety-Auditor-Agent: Performs NLP-based safety audits by scanning notes for high-risk keywords (e.g., “jam,” “leak,” “injury”) and flags near-miss events.
Incoming-Shift-Agent: Aggregates all data, generates a summarized task handover, and creates a scripted 2-minute audio podcast using Azure Text-to-Speech.
This agent collaboration ensures accurate, timely communication across operational systems—reducing human error, improving traceability, and boosting shift readiness.