https://analyse.asia/, https://open.spotify.com/show/1kkRwzRZa4JCICr2vm0vGl?si=pEVLr9c1SKaMStLHP4dlmQ, https://www.youtu

Analyse Podcast, Singapore (2026)

13/09/2026

Episode Recap – How to Actually Read China’s AI Ecosystem with Jing Yang

In this conversation Jing Yang, Asia Bureau Chief at The Information, broke down why China’s AI landscape is fragmented into 7‑8 large‑language‑model players that stubbornly resist any kind of merger. The core reasons she highlighted: each model is tied to a different tech giant or regional government, and every backer pursues a distinct strategic agenda that aligns with local economic plans rather than a unified national AI front.

She also contrasted China’s “state capital” model with the Western venture‑capital approach. State funding is deployed like a policy tool—prioritising national goals, talent retention and regional bragging rights—while Western VC follows pure market incentives and rapid exits. That divergence means Chinese AI firms iterate under different pressures, often emphasizing long‑term control and ecosystem building over quick exits or consolidation.

Finally, Jing flagged three indicators to watch through 2027: (1) the rollout speed of proprietary data‑centers tied to regional budgets, (2) policy shifts that could open or close cross‑provincial collaborations, and (3) the emergence of “government‑backed AI platforms” that might start to serve as de‑facto standards despite the lack of formal mergers.

🤔 What do you think will be the most decisive factor shaping China’s AI race by 2027? Share your thoughts below—let’s unpack the future together.

Listen on Apple Podcasts: https://podcasts.apple.com/us/podcast/how-to-actually-read-the-chinas-ai-ecosystem-with-jing-yang/id914868245?i=1000776960324

Podcast Episode · Analyse Podcast · July 15 · 58m

13/09/2026

What does it take to move from impressive AI demos to truly autonomous work?

In our latest episode, we speak with Ang Li, CEO and co-founder of Simular AI and former research scientist at Google DeepMind, about why computer use may be the final frontier in AI — and the last mile to AGI.

Ang explains why AI models often break in production: the data distribution never stops moving. Real-world environments change, edge cases appear, and systems must continuously learn from what actually happens. In his view, the only way to close that loop is through computer use in the real world.

We explore:

• The Power Law of Practice for AI agents
• Why capability and reliability are two very different things
• Why deterministic work belongs in code, not in models
• Ang’s personal test for AGI
• What it will take to bring autonomous work down to the cost of water
• Why computer use could transform how software is built and how work gets done

Will the next major leap in AI come from bigger models, or from agents that can reliably operate computers and learn from the world around them?

Listen on Spotify:
https://open.spotify.com/episode/2qJ5rVHDxoRtKeTu4Fov3j

Analyse Podcast · Episode

13/09/2026

AI in the enterprise is not failing because the technology is “not ready.”

According to Sophie Dionnet, Senior Vice President of Product and Business Solutions at Dataiku, the bigger trap is change management.

Machine learning and LLMs have already proven what they can do. But moving from a promising demo to real business impact is where many organizations struggle. As Sophie puts it: “Taking a decision is one hour, implementation is two years.”

That gap is where the real work begins: changing workflows, aligning teams, managing expectations, integrating systems, and helping people trust and adopt new ways of working.

For leaders thinking about AI transformation, the question may not be “Can the technology do it?” but “Are we prepared to change how the organization works?”

What do you think is the hardest part of bringing AI into the enterprise: the technology, the people, or the process?

Watch on YouTube

"I think the biggest failure for me today is on the change manageme...

13/09/2026

What happens when a legal team—not a dedicated data science department—takes on AI?

In this key moment from Analyse Podcast, Sophie Dionnet shares how Roche’s legal team used Dataiku to automate patent analysis. One lawyer built the system himself, completing around 90% of the work without being a data scientist.

The result was more than a technical experiment. It completely changed how the team approached patentability decisions and handled requests from law firms, creating a remarkable return on investment from an unexpected part of the organisation.

This story is a powerful reminder that successful AI adoption is not always driven by large transformation teams. Sometimes, it starts with a domain expert who understands the problem deeply and has the right tools to solve it.

Could AI give more legal and business teams the ability to redesign their own workflows? What processes in your organisation could be transformed by the people closest to the work?

Watch on YouTube: https://youtube.com/shorts/exnqUHYnwFA

"Good question. I think what's quite interesting is we end up findi...

12/09/2026

Most companies are trying to "do AI." But where is the actual value? Why do some projects take off while most quietly stall or fail?

Our latest episode of Analyse Podcast features Sophie Dionnet, Senior Vice President of Product and Business Solutions at Dataiku. She cuts through the hype with a powerful, practical framework.

Sophie argues that real enterprise AI value comes down to three essential ingredients:

1. The Right People: It’s not just about data scientists. It’s the crucial bridge between technical expertise and deep business knowledge. We discuss the story of a Roche patent lawyer who, armed with the right tools, built a working AI agent system largely on his own.

2. Orchestration: The magic isn't in a single model, but in stitching multiple technologies and data sources together to solve a real business process.

3. Governance as a Scaling Mechanism: Too often, governance is seen as a brake. Sophie reframes it as the essential system that allows you to scale AI responsibly and reliably across the organization.

The most striking insight? Most AI failures are not technology problems—they are change management problems. The hard part isn’t the model; it’s the transformation.

We dive into:
- Why focusing on the "flashy" model is often a distraction.
- How to build a culture where business experts can leverage AI tools.
- Practical steps to move from isolated pilots to organization-wide value.
- The evolving role of the Chief AI Officer.

If you're involved in strategy, digital transformation, or technology leadership, this conversation offers a vital reality check and a clear path forward.

Where have you seen the biggest gap in turning AI potential into real business value? Is it the tech, the people, or the processes?

Listen on Spotify: https://open.spotify.com/episode/5VHvVxpF3IeM0bNLpuDYFN?si=YPFY_zumTT-Z8h7zO-l8_g&ref=analysepodcast.com

Analyse Podcast · Episode

12/09/2026

What if the final frontier in AI isn’t another benchmark, but the ability to use a computer and do real work in the real world?

In our latest episode, we speak with Ang Li, CEO and co-founder of Simular AI and a former research scientist at Google DeepMind, about why computer use may be the last mile to AGI.

Ang explains why AI models often break in production: the data distribution never stops moving. A model can perform impressively in a controlled environment, but real-world work is full of changing interfaces, unexpected edge cases and feedback loops. Closing that gap requires agents to interact with the world, learn from practice and become more reliable over time.

We discuss the Power Law of Practice for AI agents, the crucial difference between capability and reliability, and why deterministic work should belong in code—not in models. Ang also shares his personal test for AGI and his vision for bringing autonomous work down to the cost of water.

What will it take for AI agents to move from impressive demonstrations to dependable digital workers? And is computer use the bridge between today’s models and truly autonomous systems?

Watch on YouTube: https://youtu.be/DJRQB5sMgbg

Fresh out of the studio, Ang Li, CEO and co-founder of Simular and ...

12/09/2026

What does it really mean to use AI responsibly — not just in theory, but when lives are on the line?

In our latest episode, we sit down with Nur Hafiza Mutalif, Assistant Head of International Affairs at the Singapore Red Cross, for one of the most grounding conversations we've had about technology and its role in humanitarian work.

Here's the thing: the humanitarian sector is often described as "slow" to adopt AI. But Nur Hafiza reframes that entirely. The caution isn't a lag — it's a feature. Before any efficiency gain, before any business case, humanitarian organisations must clear a do-no-harm standard. That changes everything about how you approach a tool like AI.

So how did the Singapore Red Cross actually get started? Through a partnership with Dataiku's pro bono experts who didn't just hand over a model and walk away. They taught the Red Cross team how the models worked, and — crucially — let the Red Cross set the boundaries. That kind of trust-building is rare, and it made all the difference.

The results were real and meaningful. Four staff members were freed from daily data collation — time that could be redirected toward actual humanitarian impact. A leptospirosis forecasting model was developed for Thailand, helping communities prepare before a crisis hits rather than scrambling after.

But perhaps the most important insight from this conversation is a warning about big data itself. When you compress the world's humanitarian crises into datasets, you risk flattening the very human complexity that makes this work matter. Numbers can simplify value. They can obscure who is being helped and how. The Red Cross is thinking hard about that tension.

AI in the humanitarian space is not about doing more with less. It is about doing better — with trust, with communities at the centre, and with impact as the only metric that truly counts.

This episode will make you think differently about what "responsible AI" actually demands — especially when the stakes go beyond profit margins.

🎧 Listen on Spotify:
https://open.spotify.com/episode/2EGuyDkytbgP3YunnXYsjM?si=uyeMDw4NSRmq1RWslsOiJw&ref=analysepodcast.com

Analyse Podcast · Episode

12/09/2026

Episode Recap: The Three Ingredients That Turn AI Into Value

In our latest conversation Sophie Dionnet, Senior Vice President of Product and Business Solutions at Dataiku, reminded us that the real catalyst for turning AI projects into measurable business impact isn’t a flashier model – it’s the people behind the technology.

Sophie explained that deep domain expertise is the first of the three core ingredients she’s identified. When a specialist knows the problem inside‑out, they can shape data, models and outcomes in ways a generic data scientist simply can’t. She illustrated this with a striking story from Roche: a patent lawyer took his own professional knowledge and turned it into a working system of AI agents. The result wasn’t just a clever demo – it became a scalable solution that delivered real value because the “right people” fed the AI the right context.

The other two ingredients – seamless orchestration of tools and robust governance – only amplify what skilled experts bring to the table. Without knowledgeable users driving the orchestration and setting guardrails, even the best‑engineered pipelines stall.

What does this mean for your organisation? Who are the domain experts that could become your AI champions, and how are you empowering them to shape the technology? Drop your thoughts below – we’d love to hear how you’re building the talent foundation for AI success.

Ready for the full conversation, deeper insights on the LLM explosion, and practical change‑management advice that cuts through the hype? Listen now on Spotify 👉 https://open.spotify.com/episode/5VHvVxpF3IeM0bNLpuDYFN?si=YPFY_zumTT-Z8h7zO-l8_g&ref=analysepodcast.com

Analyse Podcast · Episode

11/09/2026

What happens when AI moves beyond generating text and starts using computers to do real work?

In our latest episode, we speak with Ang Li, CEO and co-founder of Simular AI and former research scientist at Google DeepMind, about why computer use may be the final frontier on the path to AGI.

Ang argues that models often break in production because the world never stops changing. Data distributions shift, workflows evolve, and real-world edge cases keep appearing. The only way to close that loop is to place AI agents directly in the real world, where they can learn from experience and improve through practice.

We explore the Power Law of Practice for AI agents, the crucial difference between capability and reliability, and why deterministic work should belong in code rather than in models. Ang also shares his personal test for AGI and his vision for bringing autonomous work down to the cost of water.

What will it take for AI agents to become genuinely reliable? And will computer use be the bridge between impressive demos and useful, autonomous systems?

Read/listen at:
https://www.analysepodcast.com/simular-ai-and-computer-use-is-the-final-frontier-in-ai-with-ang-li/

Fresh out of the studio, Ang Li, CEO and co-founder of Simular and previously a research scientist at Google DeepMind, joins us to explore why computer use is the last mile to AGI. Ang traces his path from studying catastrophic forgetting and continual learning at DeepMind to founding a

11/09/2026

Imagine stepping into a bustling Shanghai market, wallet in hand—only to realize no one wants your cash. Not today, not ever. Shops, street vendors, even the elderly hawking roasted sweet potatoes? All flashing QR codes, swiping phones, no coins in sight. This isn’t the future. This is China today: a society that vaulted from cash to cashless in what feels like a single breath. And it happened overnight.

How? Why? And can the rest of Asia—and the world—learn from this lightning-fast transformation? We asked Rebecca Fannin, author and keen observer of China’s tech revolution, to break it down for us. In our latest episode, we dive into the story behind China’s mobile-first leap. No gradual shift, no baby steps. Just a full sprint intoQR codes, WeChat Pay, Alipay, and a financial ecosystem so seamless it’s invisible.

What’s fascinating isn’t just the speed, but the scale. From luxury malls to rural noodle stalls, cashless isn’t a luxury—it's the norm. And the ripple effects? Huge. Businesses adapted at breakneck speed. Consumers embraced it without missing a beat. And now, ideas born in China are cascading across Southeast Asia, shaping how millions transact every day.

So here’s the big question: Is this a blueprint for the future, or a uniquely Chinese phenomenon? Could Singapore, Jakarta, or Bangkok ever replicate this kind of leap? And what happens when cash disappears—who wins, who loses, and who gets left behind?

Rebecca breaks it all down in this sharp, insightful conversation. No jargon, no hype—just real stories from the ground and hard-won lessons for anyone watching Asia’s digital transformation unfold.

Ready to see how China did the impossible? Watch on YouTube now: https://youtube.com/shorts/oPWszRgiEx4

Bernard Leong: "What can the rest of Asia now learn from how China ...

Analyse Podcast

13/09/2026

13/09/2026

13/09/2026

13/09/2026

12/09/2026

12/09/2026

12/09/2026

12/09/2026

11/09/2026

11/09/2026

Address

Website

Alerts

Shortcuts

Share

Category

Listing takedown request