Microsoft Considers DeepSeek for Copilot
ATP Insights discusses the most important tech news in Asia.
ATP is presented by:
Welcome back to The Asia Tech Podcast. Today we covered Microsoft’s move to consider a Chinese open-source AI for Copilot, what it takes to get enterprise AI out of prototype mode, autonomous payments and the stablecoin connection, how smart devices are eliminating blind spots in global shipping, and why the old model of cyber defense has finally broken down.
We were joined by:
• 22:48 Carlos Herrera - Founder & CEO at Nimble
• 46:44 Harm-Julian Schumacher - Co-Founder & CEO at OneLot
• 01:02:58 Kiran Sharma - Co-Founder & Director at Vasu International Payment Solutions
• 01:24:37 Ben Lawson - CEO at Sxored
• 01:43:25 Ittay Hayut - CEO at Hoopo
• 02:03:46 Damian Bierman - Co-Founder at Blackwired
Here is an overview of the topics we discussed today:
Microsoft Eyes DeepSeek for Copilot
Why would one of OpenAI’s biggest investors consider swapping it out for a Chinese open-source model? Microsoft is reportedly exploring a hosted version of DeepSeek V4 on Azure as a cheaper option inside Copilot Cowork, its agentic productivity system. The reason is cost. When an AI agent runs continuously across your calendar, email, and documents — calling tools, retrieving context, completing long tasks — the tab on a subscription model becomes unsustainable even for Microsoft.
DeepSeek’s advantage isn’t just price. The company built its edge around compute efficiency — doing more with less. It just raised $7.4 billion at a $50 billion valuation, with a structure where most investors get no voting rights and a five-year lockup. China’s national AI fund invested directly and kept its voting rights. Could that detail change how Western companies think about integrating this model into their core productivity tools?
Carlos Herrera on Getting Enterprise AI Into Production
Carlos runs Nimble, a Bangkok-based tech consulting firm that helps enterprises move from AI pilots to real production systems. His observation: most AI-native SaaS is failing in the enterprise because it shows up without anyone to deploy it. The model that works is Forward Deployed Engineers — technical people who work side-by-side with client teams to build, customize, and get internal buy-in.
What’s changing inside those client teams? Carlos sees engineers, designers, and product managers converging into a single profile — the “product builder.” Is that a new kind of role, or just a higher bar for everyone? And if every company now needs these hybrid technical-communicator types on-site, where do they all come from?
Kiran Sharma on Agentic Payments
Kiran runs Vasu, a payment gateway operating across 55 countries and serving high-risk, high-volume merchants in Forex, crypto, and online gaming. His focus is infrastructure for a world where AI agents transact on your behalf.
Here’s the scenario that makes it concrete: an AI agent is authorized to find the best flight upgrade. A bidding window opens for 30 seconds. The AI can’t wait for banking settlement windows or business hours. That pressure — autonomous agents acting on time-sensitive opportunities — is why real-time agentic payments almost inevitably point toward stablecoins. Does every major payment gateway need to rebuild its rails before autonomous agents can really work?
Harm-Julian Schumacher on Financing the Invisible Dealer
Harm-Julian is building OneLot in the Philippines, where 95% of used car dealers have no access to bank loans. The average dealer has 20 cars and no credit history that traditional lenders recognize. OneLot uses AI-powered underwriting — machine learning on inventory data, app-based inspections, and cash flow modeling — to unlock capital for dealers who are otherwise invisible to the financial system. What happens to a fragmented market when the financing gap finally closes?
Ben Lawson on AI and Credit Analysis in Indonesia
A $50,000 loan in Indonesia currently takes two to three months to process and requires 40 to 45 documents — many still handled manually. Ben’s company, Sxored, uses AI to extract, analyze, and flag anomalies across those documents, including fraud signals like inflated revenues, duplicate transactions, and fake invoices. The biggest challenge isn’t the technology. It’s getting institutions that have operated the same way for 20 years to trust a process where AI sees the data. How long before the pressure to scale forces the issue?
Ittay Hayut on Container Intelligence
Ittay’s company Hoopo makes solar-powered tracking devices that attach to shipping containers in two minutes, with up to 15 years of battery life. Once installed, the container reports location, door status, temperature, and shock events in real time. The problem they’re solving: once a container leaves the port, it enters a complete data blind spot — handled by truckers, drayage companies, and third-party logistics firms who all interact with your asset but report nothing back.
The implications go further than efficiency. According to EU data, 80% of cocaine entering Europe moves inside shipping containers. Hoopo is working with the EU to help customs agents focus on the right ones. What else is hiding in those blind spots?
Damian Bierman on Proactive Cyber Defense
Damian’s company Blackwired operates at what he calls the bleeding edge — where threat actors actually live. His core argument: “detect and respond” has been cybersecurity’s default posture for decades, but agentic AI has finally broken that model.
AI systems can now discover zero-day vulnerabilities faster than security researchers can publish patches. Worse, attackers increasingly use legitimate, authenticated protocols to slip through undetected — no alarm triggers because nothing looks wrong. Damian’s answer is to cut the kill chain upstream: identify the infrastructure attackers are hiding behind, understand their tools before they deploy them, and sever the path. Is there any other approach that actually keeps pace with autonomous offense?
Watch the full episode here:
ATP Is Presented by Twilio:
Twilio is the secure and scalable platform to build amazing customer engagement without the complexity, with the AI-driven workflows you need to connect.
Visit Twilio.com to start building today
ATP Is Made Possible by:
Boomi, the data activation company for AI, powers the agentic enterprise by bringing data to life across the business. Trusted by over 30,000 customers and supported by a network of 800+ partners, Boomi helps organizations of all sizes achieve agility, efficiency, and innovation at scale.
Visit www.boomi.com to learn more