AI: State of Industry and Investment | Aravind Kandiah, Jun Wakabayashi, John Homer Alvero - E717

"If you can own a space, both ideologically and also be in all the right rooms, sure, someone else may have the right model, but they will pick you. I think brand is very, very under-indexed, and doing that work of go-to-market is so hard, especially in a low signal, high noise environment."

"We want to move fast and we want to adopt AI, but we're also concerned about our data. We want to protect our customers' data. We want to make sure that the data of our customers is not used for training the next generation of models. So we're balancing. We want to move fast, and we're also careful at the same time."

"Startups of course are pioneering the frontier here, but we're actually starting to see a lot of enterprise teams basically using tools like Cursor, Copilot, and OpenAI. If you go into an engineering bullpen now, it actually is starting to mimic what seems like a call center. Engineers are no longer typing much anymore; they're just speaking, using voice to code now."

Everyone has the same access to the same frontier models. So what's actually left to compete on?

Jeremy Au moderates a panel at the Crosscurrents Summit in Parañaque on where enterprise AI adoption really stands, with Aravind Kandiah of Bifrost, Jun Wakabayashi of AppWorks, and John Homer Alvero of Converge ICT.

Aravind Kandiah is the CTO and Co-founder of Bifrost, a robotics infrastructure company that simulates the world to evaluate robots. Bifrost's simulations power some of the world's largest robotics companies, spanning autonomous Mars exploration with NASA through to automating high-risk industrial work. He spent over five years in AI and robotics research before co-founding the company with Charles Wong, and Bifrost is backed by Sequoia Capital, Lux Capital and Airbus Ventures. On the panel, he argues enterprises split cleanly into those running pilots to hit an R&D spend target and those facing a genuine forcing function, like Korea's birth rate leaving factories unstaffed. Only the second group ships. His view on defensibility: the moat is everything except the model.

Jun Wakabayashi is a Venture Principal at AppWorks, one of Asia's leading accelerators and VC firms. He joined as an Analyst in 2017 and rose to Principal by 2023, leading the firm's Beacon Funds arm, a fund-of-funds backing emerging venture managers across Southeast Asia and web3. Immersing himself in founder communities across the region built the network that made AppWorks a first stop for startups seeking capital. He holds a B.S. in Finance from NYU Stern and previously worked at Focus Reports and PwC. He puts engineering at 60 to 80% of enterprise token spend, describes bullpens that now sound like call centres with engineers dictating to coding agents instead of typing, and tracks the teams accidentally burning $5 million a month on tokens. His counter to Aravind: the real moat is distribution, the one thing the frontier labs lack.

John Homer Alvero is Head of AI Engineering at Converge ICT, one of the Philippines' largest fiber broadband and digital infrastructure providers. An experienced AWS architect, he has worked across e-commerce, fintech, telco and gaming, with deep expertise in cloud architecture, operations and security. He previously held cloud and service engineering roles at Voyager Innovations and was Cloud Solutions Architect at SM Investments, giving him a practitioner's view of moving enterprise AI from pilot to production. His blocker isn't the technology: governance and cybersecurity teams can't write credible guardrails until they're AI-literate themselves. His sharpest point is that with model access equalised, the only variables left are your proprietary data and your orchestration.

Crosscurrents Summit, presented by Clouted

Watch on YouTube: https://www.youtube.com/watch?v=akjg47d6s6c&list=PLl9u6ECOP8_7scb97PE3whKu4yJVizIOd

Listen on Spotify: https://open.spotify.com/episode/0yULzJRdUltAQDlbp1JhJl

Keywords: Enterprise AI Adoption, AI ROI (Return on Investment), Token Spend Management, Build vs. Buy AI, AI Startups, Cybersecurity and AI Governance, Autonomous Agents, AI Distribution and Moats, Voice-to-Code Engineering