It’s been nine months since I arrived in Beijing through the NUS Overseas College program. Between classes, startups, and coffee chats, I’ve been quietly watching how AI reshapes the way people work and build across different cultures — from Singapore to China to the U.S.
During the recent China's National Day break, I spent two weeks in Silicon Valley — just in time for San Francisco Tech Week, where over 1,100+ events unfolded across the Bay Area, from Presidio to Palo Alto. Cafes, coworking spaces, and rooftop bars turned into a gathering spot for founders, investors, and developers — a living map of how innovation actually happens.
This blog is part travelogue, part reflection — capturing what I saw, heard, and learned during those intense days of conversations about Agentic AI, startups, and the future of AI visibility.
Day 1: Agentic AI — What’s Hot & What’s Not
The first event of the week, “Agentic AI: What’s Hot & What’s Not,” felt like a concentrated shot of Silicon Valley realism. The hype is everywhere — but so is the growing self-awareness.
“AI lets you start from 90% complete. The future is about agents that actually get work done.” — Sang Wen, Co-Founder & COO of Genspark
The Speakers agreed: the hype is real, but so are the opportunities — if you’re willing to look past the noise.
Here’s what stood out:
- Hype vs. Reality: Many boards and founders are still chasing the “AI narrative” for funding, often overpromising and whitewashing products. Valuations are inflating, but so are expectations.
- Real Value: Tasks that once cost thousands now run for $25/month. The future belongs to agents that execute, not just assist.
- Enterprise Friction: Integrating with legacy systems and turning data into decisions remain the hardest problems. Open source mitigates cost and privacy, but legal questions around build vs. buy are escalating fast.
- People & Culture: The real shift is human — from outsourcing to retraining. Companies are upskilling employees to use AI effectively. Growth hiring now taps content creators as creative generalists. Picking cofounders with complementary values is non-negotiable.
- Investor Dynamics: VCs are increasingly allergic to buzzwords. Real traction, authenticity, and resilience matter more than hype. “VC money,” one speaker said, “is personal debt you repay with multipliers.”
- Building to Last: Don’t just surf the wave. Ask yourself — will you still care about this problem in four years? Write “a day in the life of a customer.” AI can code 90% of it, but you still need taste, systems, and process to make it matter.
- Risks & Outlook: Velocity is insane — small ideas can explode overnight — but implosions will follow. Corporate silos and internal knowledge bases remain cold spots. In a ruthless capitalist loop: retrain, adapt, or get left behind.
Bottom line: Agentic AI is maturing from hype to utility. The winners will integrate deeply into existing systems, empower human talent, and solve real problems with authenticity and adaptability.
Day 2: Zero to Unicorn — Fireside with Eric Simons (Bolt.new)
The next day, I joined a fireside chat with Eric Simons, founder of Bolt.new — one of the most talked-about AI-native productivity startups of the year.
Within just five months, Bolt hit $40M ARR, built its MVP in 90 days with a 12-person team, and became a case study in speed without chaos.
“It’s you vs. you. The biggest enemy of any founder is avoiding reality.” — Eric Simons
Eric’s message wasn’t about hype or virality. It was about discipline — moving fast, but staying grounded in customer truth.
Key takeaways:
- Rapid Build, Rational Growth: Speed matters, but only when aligned with clarity. Their inbox overflowed, but focus remained on customer outcomes.
- Customer Obsession: Success = how much time you spend with users and how deeply you help them succeed. “Get customers to post about using your product” — that’s the new marketing.
- Startup Reality Check: Founders often escape into talking instead of building. Staying honest with yourself is the hardest — and most critical — part.
- Product Focus: Bolt’s goal is to automate PM workflows — turning specs and tickets into agent-driven processes. Engineers fix bugs; PMs define meaning.
- Distribution > Code: Anyone can ship an MVP. Scaling still depends on people, processes, and community. Community is distribution — your next product will launch into an existing trust network.
- Fundraising & Scale: Building is cheap; distribution isn’t. Enterprise sales mean hundreds of conversations, not a few lucky intros. Product-market fit still lives in the hands of real users.
- Outlook: A billion-dollar “vibe-coded” startup may not be far away. But lasting companies will combine fast iteration with authenticity, user obsession, and repeatable distribution.
Bottom line: Launching has never been easier, but scaling and staying authentic are what separate toys from unicorns.
Day 4: AI Search Visibility — Being Found in an AI-First World
By midweek, the events I joined shifted from building products with AI to being seen by AI.
I joined Liliana Pertsanova, CEO of Auxiliary, who ran a sharp session on AI Search Visibility — how startups can remain discoverable when AI replaces traditional search.
The Shift: From Links to Answers
Traditional search gave you a list of links; AI search gives you the answer.
Nearly 60% of searches now end without a single click. If AI doesn’t know you, the world doesn’t either.
“In the era of AI, if AI doesn’t know you, the world doesn’t either.” — Liliana Pertsanova
A Three-Step Framework for AI Visibility
- Create Content AI Loves
- Write structured, referenceable content: FAQs, comparisons, implementation guides.
- Use schema markup to help AI parse and trust your site.
- Prioritize depth and clarity over keyword volume.
- Let Others Build Your Authority
- Leverage user reviews and community platforms like G2, Trustpilot, Reddit, and specialized forums.
- Encourage organic mentions — AI treats these as trust signals.
- Strategic PR
- Appear in outlets AI engines already trust (TechCrunch, Forbes, niche media).
- Position executives as industry experts with consistent narratives across touchpoints.
Case Study: Exponent 21
A boutique agency that focused entirely on optimizing for AI visibility. Within a year, they achieved 4,000% traffic growth by aligning content, authority, and presence around how AI systems learn and retrieve information.
Five-Step Action Plan
- Audit your AI presence — see how ChatGPT, Perplexity, or Gemini describe you.
- Define 3–5 areas of expertise.
- Build a content hub around each.
- Invest in PR across trusted publications.
- Monitor and iterate as AI search evolves.
Day 4 (Evening): AI Search Writes the Story
Later that day, a fireside hosted by limy.ai with Google veterans Jitendra Kumar and Michael Carney explored how AI search and marketing are collapsing the old funnel — from keywords to intent.
1. Search Is Becoming Multimodal
AI search isn’t just text anymore — it’s video, audio, and images all fused into “AI Overviews.”
Every query will soon return multimodal answers.
“There’s a very good chance that every search, every AI overview, will be multimodal.”
2. Hyper-Personalization Is the Default
Users now expect instant, context-aware answers — no more scrolling or clicking.
They want responses that understand who they are and what they need, in real time.
3. Marketing Must Focus on Outcomes
Clicks and impressions are vanity metrics.
“It’s not about impressions, it’s not about views — it’s about outcomes.”
Marketers now optimize for measurable ROI strong enough for CFOs to justify continued investment.
First-party data becomes the new creative engine:
Imagine a pizza ad that automatically changes its background based on your location and the weather — sunny patio in LA, cozy fireplace in Seattle.
Reflections: Beyond the Hype
AI is not just about what’s possible, but also about what’s useful and sustainable.
Technology is accelerating faster than human adaptation — but culture, systems, and mindset remain the true bottlenecks.
As I return to Beijing for the final months of my program, I carry a quiet clarity — that the next phase of AI won’t be defined by who builds the most powerful models, but by who learns to integrate them gracefully into how humans think, create, and collaborate.