In today’s enterprise, speed is everything. Ambient agents keep work flowing. They are able to spot issues before they escalate, respond in real time, and power a more proactive organization.
This October, Moveworks is bringing the power of Ambient agents to developers and builders around the world. The Moveworks AI Agent Hackathons are spread across three cities: San Francisco (October 7), London (October 14), and Bengaluru (October 28).
These events are bringing together builders ready to explore a new class of Ambient agents: intelligent systems that listen, understand, and act on enterprise signals – reshaping how work gets done across organizations.
CEO and Co-Founder Bhavin Shah shared, '‘Moveworks was built on one idea: that AI should work for everyone, everywhere. These hackathons are your chance to build the future with the next generation of AI agents that listen, act, and transform how we work today.”
October 7th marked the first AI Agent Hackathon, located in the iconic Chase Center in San Francisco. Hosted together with Deloitte, 89 Solutions, and Nymbl, participants gained hands-on experience building AI agents in Agent Studio across business systems. Winning solutions included a Project Management Agent, Website Generation Agent for Sales Demos, and Real-Time Outage Detection Agent.
Now we want to share all the best of the San Francisco AI Agent Hackathon 2025 with you: the insights, key takeaways, and how builders are redefining the future of work with Ambient agents.
Welcome keynote: Moveworks CEO Bhavin Shah on building the future of work
When Moveworks CEO and Co-Founder Bhavin Shah took the stage to kick off the Hackathon, you could feel the energy.
Shah painted a picture of the modern enterprise, a landscape of powerful apps that each do their job brilliantly, but don’t always connect. “We built the skyscrapers,” he said, “but forgot to build the roads, the bridges, and the connecting infrastructure.”
The result of this disconnect? Economic gridlock. Context switching. Missed momentum.
And that’s where agentic AI comes in — not as another app, but working as the connective tissue for work itself. Bhavin described the next wave of AI as "AI that acts'" It’s capable of reasoning across systems, handling complex workflows, and adapting to how employees actually work.
“The next decade will belong to those who master the flow of work,” Bhavin said. “You’re the ones who will build it — right here, today.”
The keynote set the tone for the day and built excitement within the community to explore what’s next for AI in the organization.
Customer round robin: Real stories, real impact
The customer lightning round was all about showing what’s possible when AI agents move from concept to real-world impact.
Siemens’ Head of IT Automation Excellence, Pavithra Babu, shared her experience with how their myIT Digital Assistant is transforming employee support. A quick-action agent for smart card issues alone has already cut tickets by 30%, saving thousands of hours. For Babu’s team, it’s more than automation — it’s momentum at scale.
At Mantech, Brian Shrift, Executive Director in the Office of the CIO, showcased how Ask MANTECH became their “global front door for AI assistants.” With Agent Studio, their team built an Employment Verification Letter (EVL) Agent in under five hours — this agent frees up over 1,000 hours of manual work each year and allows employees to self-serve EVLs.
Finally Micron, represented by JungAh Choi, VP of Enterprise Applications, who introduced Mia, the company’s intelligent AI assistant supporting 40,000+ employees. What started as a simple Q&A chatbot has evolved into a GenAI-powered Assistant that acts across systems like SAP and Workday.
Their Purchasing Intake Agent — winner of the Most Powerful AI Agent at our AI Agent Hackathon earlier this year— now speeds up the procurement process by 68% and saves 5,000 hours every month.
These stories made one thing clear: AI agents aren’t just helping employees work smarter — they’re transforming how work flows across entire organizations.
Hackathon kickoff: Cofounder and President Varun Singh on the power of developers
Next up, Cofounder and President Varun Singh took the mic to get everyone ready to build and share the inspiring work achieved by the Moveworks developer community.
He shared empathy for the challenging environments that developers are building in: fast, complex, constantly shifting. All while needing to give thousands of employees an agentic AI experience across hundreds of systems.
Singh urged listeners to, “Make your Moveworks Assistant a true front door for your business systems. Let AI agents do the heavy lifting in a matter of minutes instead of having complex processes take days and weeks.”
He also stated how our developer community continually inspires us in overcoming these obstacles with their creativity:
- Speed: Over 10,000 agents built since the beginning of the year.
- Connectivity: Agents are able to now execute actions across hundreds of business systems, going beyond just search and synthesis.
- Creativity: They're transforming complex, 32-step procurement processes into seamless, policy-compliant workflows.
Agent Studio: Major product innovations unveiled
Our participants were the first to not only learn about, but also experience, the latest innovations and advancements to Moveworks Agent Studio, the platform to build, customize, and deploy enterprise-grade AI agents, together with the AI Agent Marketplace library of prebuilt agents.
Both Varun Singh and Ajay Merchia, Director of Product Management at Moveworks, shared about the power of Ambient agents and the Agent Architect.
These product innovations make it even easier to use Agent Studio and AI Agent Marketplace to extend the AI Assistant’s functionality, while rapidly innovating and scaling countless automation across business functions.
Discover how to build Intelligent AI agents, faster in Moveworks Agent Studio
Deploying Ambient agents
Unlike conversational AI agents, Ambient agents don’t sit idle waiting for prompts. They can be configured to respond to triggers and perform actions automatically, an event like a Zoom call ending or a scheduled signal every Monday at 9 a.m., and are able to take action automatically.
At Moveworks, this “listen, understand, act” loop is the foundation of every Ambient agent. It’s how we turn system noise into system intelligence. Ambient agents start in the same way:
- Listen: It detects a trigger: a webhook, a schedule, a data event.
- Understand: It applies reasoning using an LLM to determine context and intent.
- Act: It executes through our agentic automation engine to deliver the result.
Developers learned about a set of example Ambient agent patterns:
- Pattern one: An LLM-powered agent that transforms noisy alerts into concise, meaningful summaries, giving teams focus and context instantly.
- Pattern two: A scheduled agent that automates brand monitoring, scanning and summarizing conversations so insights arrive before your first coffee.
- Pattern three: A smart routing agent that filters out alert noise, engaging engineers only for true emergencies and auto-ticketing the rest.
- Pattern four: An unmoderated agent listens to sales calls, summarizes, qualifies, and updates the CRM automatically, so reps can focus on selling.
- Pattern five: A moderated agent drafts thoughtful replies to sensitive posts and waits for human approval, blending AI speed with human judgment.
Moveworks customers can sign up to build Ambient agents through the Moveworks Community. The Community allows developers to stay up-to-date on the latest events, get inspired by what others are building, get involved with special interest groups, and participate in Limited Preview programs.
Introducing the Agent Architect
Moveworks also revealed the Agent Architect, your dedicated AI partner for the entire agent development lifecycle, created with the goal to make building agents as intelligent as the agent itself.
Building intelligent agents should feel smart from the very beginning. That’s the idea behind Agent Architect, which will be your AI partner for designing, building, and debugging Ambient and conversational AI agents.
Agent Architect isn’t just built to help you code faster. It also will be able to help you build more powerful AI agents by supporting actions like:
- Collaborative ideation: Start with “What if I could…?” Agent Architect explores possibilities, proposing triggers, patterns, and outcomes before you build.
- Blueprint generation: Instantly generates a complete development blueprint, to define triggers, reasoning, and automation flows for faster execution.
- Context-aware debugging: Goes beyond error logs to explain failures in plain language, mapping issues across dependencies and logic chains.
- Continuous improvement: Learn from every build, anticipating your patterns, optimizing workflows, and preventing inefficiencies before launch.
In these ways, Agent Architect will help to turn the complexity of agentic AI development into a guided, intelligent process – from concept to deployment.
Winning hackathon solutions
With these innovations unveiled, participants moved from concept to action.
Participants shared how easy it was to install, configure, and deploy AI agents and Ambient agents. Many developers arrived with ambitious ideas and clear visions for the agents they wanted to build, and by the end of the event, those ideas turned into reality and these moments were celebratated right on center court.
During the showcase, we honored the winners of the AI Agent Hackathon, spotlighting those built during the event.
Congratulations to our AI Agent Hackathon winners!
1. Project Management Ambient Agent
Grand Prize: Molina Healthcare | : Harmandeep Brar, Tarak Patel, Asif Hussain
Stand-ups might just be a thing of the past. This team built an agent that keeps track of all projects and automatically reaches out to task owners to get updates and missing details, helping developers ship faster and PMs stay in sync.
2. Website Generation Agent for Sales Demos
Honorable Mention: Twilio | Ram Padte, Sarmad Syed, William Hoopes, Ben Ringerman, Karl Bigley-Bodlak
This one was incredibly creative. To help sellers demo more effectively, Team Twilio built an agent that generates personalized websites showing what their product could look like for each prospect. It does this by pulling customer info from the CRM, researching using LLMs, generating HTML, CSS, and JS, pushing everything to GitHub, and deploying a live site, all within seconds.
3. Real-Time Outage Detection & Root Cause Analysis Agent
Honorable Mention: Amadeus | Sylvain Baeriswyl
Solving one of the toughest problems in IT operations, this agent detects app outages in real time, identifies recent code changes, combs through network logs, and provides clear recommendations to restore service faster and minimize downtime across critical systems.
Stay ahead with Ambient agents
Imagine AI that is able to anticipate exactly what your organization needs. That's the reality Ambient agents are creating, laying the groundwork for a truly proactive operating organization.
In San Francisco, we watched this come to life: Participants swiftly built, tested, and deployed agents that could listen for system signals, intelligently reason within enterprise permissions, and take decisive action within minutes.
Now across the world, from San Francisco to London to Bangalore, developers are proving how fast these agents are able to move from idea to impact – reacting instantly, running quietly in the background, solving before escalation, and working before you even know there’s work to do.
Moveworks customers can sign up to access Ambient agents now in Moveworks Community, our customer and builder hub where you can get information, get inspired, and get involved with special groups and Limited Preview programs.
Table of contents
