Management Students Put Agentic AI to Work at UCLA's Innovate Tech Conference

Agentic AI Moves From Hype to How-To at UCLA's Innovate Tech Conference

About 300 attendees packed into UCLA Anderson to answer a simple question: how do you put autonomous AI to work in real businesses? Hosted by the Easton Technology Management Center and the Technology Business Association, the conference brought operators from Electronic Arts and Ancestry together with founders and alumni to share what's working now.

Agentic AI-an autonomous system that can pursue goals with minimal supervision-took center stage. For a clear definition, see IBM's overview of agentic AI here.

Why it matters for executives, strategy, and product

• Shift from prompts to processes: Agents don't just answer-they execute multi-step workflows end-to-end.
• Leverage without headcount: Automate routine analysis, outreach, and QA so teams focus on higher-leverage work.
• Faster cycle times: Move from idea to prototype in days, not quarters, using "vibe coding" and agent frameworks.
• Governance up front: Clear guardrails, human review points, and KPI tracking turn AI from a demo into a dependable system.

Inside the conference

Speakers with real operating experience set the tone, including Mihir Vaidya (Chief Strategy Officer, Electronic Arts) and Gene Alston (Ancestry board member). A closing conversation between Adam Cheyer (co-founder of Siri) and moderator Prema Sampath (former product leader at Meta) dug into what it takes to deploy AI at scale inside large orgs.

Student coordinator Raman Vedula put it bluntly: "We are moving from generative AI models that started off in early 2020 to Agentic AI, because we are automating everything. Humans can focus more on productivity and creativity."

Cephas Sund, a UCLA alumnus, left ready to lean into "vibe coding"-letting AI generate much of the code while developers steer architecture and constraints. "This conference had a really good mix," Sund said. "Not just high-level thought leadership-there was real talk on applications and usage."

Yvonne Wassenaar, former CEO of Puppet, shared practical ways to ask better questions so AI produces stronger outputs-useful for PMs, analysts, and execs who want higher-quality answers without extra cycles. For many attendees, the clear message was preparedness: AI is about to be part of every job. "I don't have any tech background… I know that you can use AI to help you do that," said MBA student Alfie Twum-Ampofo.

What leaders can do this quarter

• Identify 2-3 repeatable workflows ripe for agents (e.g., lead research, QA triage, monthly ops reporting, vendor outreach).
• Start with a human-in-the-loop design. Define where humans approve, edit, or escalate.
• Pilot "vibe coding" for internal tools. Ship small: one agent, one workflow, one clear owner.
• Write prompts like specs. Include role, data sources, success criteria, constraints, and output format.
• Set guardrails early: data access rules, logging, versioning, and off-switches for risky actions.
• Measure what matters: cycle time, cost per task, error rate, and percent of steps fully automated.
• Upskill the team. Train PMs and analysts to design agent workflows, not just write prompts.
• Audit vendors. Ask about evaluation datasets, fallback behaviors, observability, and SOC2/ISO posture.

Signals from the field

Speaker selection mirrored where agents are already delivering value-gaming, consumer platforms, and data-rich products. Planning began in November, and the event closed with a focused networking lunch that connected students, alumni, and operators who are shipping real systems.

The theme across sessions was confidence with accountability: build momentum with small wins, then formalize the operating model around agents as they prove themselves.

Next steps and resources

• Share a 1-page brief with your leadership team outlining two candidate workflows, expected ROI, and a 4-week pilot plan.
• Stand up a lightweight "AI Ops" review: security, legal, and analytics meet weekly to track pilots and clear blockers.
• If you're building your own skill stack, explore focused learning paths and certifications: Courses by Job and Popular AI Certifications.

Bottom line

Agentic AI isn't a future bet anymore. It's a practical way to turn manual, multi-step work into reliable, auditable systems-so your teams can spend less time on process and more time on product, customers, and growth.