Latest AI News
OpenAI Teases GPT-5.4 AI Model Launch Just a Day After Releasing GPT-5.3 Instant
OpenAI teased the release of the GPT-5.4 artificial intelligence (AI) model on Wednesday. The San Francisco-based AI giant said that the major update to the existing model could arrive soon, making it the fastest launch teaser after the launch of a major model. Interestingly, on Tuesday, the company released the GPT-5.3 Instant to all ChatGPT users. The model focuses on improved conversations and better writing quality. Notably, the developments arrived after a website claimed that about 2.5 million ChatGPT users intend to quit the platform.
View
OpenAI’s Codex App Is Now Available on Windows, Can Be Downloaded via Microsoft Store
A month after launching a dedicated Codex app for macOS, OpenAI has now launched the app to Windows devices as well. On Wednesday, the San Francisco-based artificial intelligence (AI) giant announced that the app is now available in the Microsoft Store, and developers using the Windows operating system can now use the app for agentic coding as well. At launch, users will be able to access all the existing features of the platform, including skills and the ability to run multiple AI agents in parallel.
View
Lio raises $30M from Andreessen Horowitz and others to automate enterprise procurement
Lio’s co-founders know firsthand that procurement — the process enterprises use to purchase services from vendors — is often a bottleneck. Vladimir Keil, the company’s co-founder and CEO, had experienced this problem as an employee inside a large company and then again while building his first startup. “When we were selling enterprise software, we had to go through procurement ourselves and saw how manual and fragmented the process still is,” he told TechCrunch. Kiel and his team have built an automated platform of AI agents — software that can complete tasks on behalf of humans — to help fix some of those fragmented processes. On Thursday, Lio announced a $30 million Series A in a round led by Andreessen Horowitz. SV Angels, Harry Stebbings, and YC also partook in the round (Lio was part of the Spring’23 batch). The company has raised $33 million in funding to date. Keil said the fresh capital will be used to expand the company throughout the U.S. and increase the capabilities of Lio’s AI agents, which aim to complete the entire procurement process for enterprise customers. Procurement is at the heart of enterprise spending, where companies look to buy everything from raw materials to professional services. Each purchase order requires focus and commitment: One usually has to open some type of Enterprise Resource Planning, or ERP, software, check contract management systems, search the supplier database, run compliance checks, cross-reference budgets, dig through emails, and so on. “Even with modern eProcurement software, most of the real work is still done manually,” Keil told TechCrunch. Companies are left to build large internal teams or outsource this work, resulting in a slow, expensive process. Keil had an idea — if the procurement process is largely unstructured data and repetitive workflows, then surely this is the type of task an AI agent is well-equipped to handle. He teamed up with friends Lukas Heinzman and Till Wagner and in 2023, the trio launched Lio, a virtual procurement workforce. Lio operates an AI-native platform with agentic infrastructure that completes the entire procurement process “Every previous generation of procurement technology was built on the same assumption, that humans will do the work and technology will help them do it faster,” Keil said. “We take a fundamentally different approach. Instead of building software to help humans do procurement work faster, Lio deploys AI agents that execute the workflow themselves.” These Lio agents operate across and on top of enterprise systems to read documents, evaluate suppliers, negotiate terms, and complete transactions. “Processes that once took weeks can now be completed in minutes,” Keil said, adding that the startup is already helping companies manage billions in enterprise spend. “In one case, a global manufacturer was able to automate 75% of its previously outsourced procurement operations within six months.” Lio is among the many companies that have popped up tocompletely redefine enterprise software, aided by agentic AI’s ability to fundamentallyshift how enterprise application software operates. Keil considers Lio’s competitors to be legacy procurement software vendors (such as SAP Ariba and Oracle), Business Process Outsourcing (BPO) providers, and consulting firms that help companies with these operations. “Instead of spending most of their time processing requests and paperwork, teams can run more negotiations, analyze more suppliers, and capture savings opportunities that would otherwise be missed,” Keil said. “In the long run, we think this changes procurement from a back-office function into a much more powerful lever for enterprise performance.”
View
How 1,000+ customer calls shaped a breakout enterprise AI startup
This isn’t David Park’s first rodeo. The veteran founder and TechCrunch Startup Battlefield alumnus has certainly been battle-tested in the enterprise arena. On this episode of Build Mode, Park joins Isabelle Johannessen to discuss how he and his team are intentionally iterating, fundraising, and scaling Narada. This enterprise AI solution uses large action models to automate complex, multistep workflows across enterprise systems. At face value, Narada has everything that would likely have investors banging down its door: a dream founding team of experienced researchers and operators from Stanford and Berkeley, big name enterprise customers, and a product that works. So in 2024, when Narada applied for Startup Battlefield, it surprised the team how little fundraising they’d done. That choice was by design. “We wanted to not waste too much money,” said Park when asked about why they waited to fundraise. “Because I do believe that when, again, when you have too much money in the bank and you are not near product-market fit, you’re tempted to just spend money on things that actually don’t help you evolve the company in the right way. It removes the friction to do a lot of wrong things.” Park previously founded and exited Coverity. That founding experience taught him one crucial lesson that he’s taken with him to Narada: Take the time to talk to your customers before you do anything else. Park said that in the early days, he and his co-founders were not focused on reaching out to VCs, but instead the three of them made over 1,000 customer calls to deeply understand what the pain points were. Once the problem was extremely clear, the solution came into focus. These teams needed an AI product that they could speak to like a person and trust to take on multiple steps at once. “If you want to build a real business, ask the hard questions, right? Spend time with customers, and not just in selling, because when you have that contract and that purchase order, that’s just the beginning, right?” Park advises viewing those early conversations as more than sales calls: “And some of those customers that we bootstrapped with ultimately turned into multimillion-dollar deals, right? And it’s always easier to sell more to a company that has already chosen you and has some level of trust in you.” As a veteran founder, Park has a foundational belief that to build a company the correct way, the customer must be centered in every decision. Because at the end of the day, no matter how trendy, interesting, or well-received your product is by the industry, if people won’t pay for it, it won’t be a winner. Loading the player… Apply to Startup Battlefield:We are looking for early-stage companies that have an MVP. So nominate a founder (or yourself). Be sure to say you heard about Startup Battlefield from the Build Mode podcast.Apply here. Founder Summit 2026:If you want to take these conversations beyond the podcast, then join us in person at the TechCrunch Founder Summit 2026 event on June 9 in Boston. This is essentially Build Mode in real life. It’s a full day focused entirely on founders, builders, and the conversations that actually move startups forward. It’s also a great way to sharpen your story.Get your tickets. TechCrunch Disrupt 2026:We’re back for TechCrunch Disrupt on October 13 to 15 in San Francisco, where the Startup Battlefield 200 takes the stage. So if you want to cheer them on, or just network with thousands of founders, VCs, and tech enthusiasts, thengrab your tickets. Isabelle Johannessen is our host.Build Modeis produced and edited by Maggie Nye. Audience Development is led by Morgan Little. And a special thanks to the Foundry and Cheddar video teams.
View
Lonza to Establish GCC in Hyderabad to Strengthen Telangana’s Life Sciences Ecosystem
The company’s decision follows an extensive evaluation process, during which Hyderabad emerged as the preferred destination.
View
Can AI Make India’s Roads Safer?
ThinnAI targets aspiring drivers aged 16 and above, aiming to reduce road fatalities by training first-time drivers and encouraging safe driving habits.
View
Amazon Cuts Jobs in Robotics Division, at Least 100 Roles Affected
The layoffs affect robotics unit roles as Amazon restructures warehouse automation and shelves the Blue Jay project.
View
Predictive Diagnostics Is Scaling in India Even When Foundations Are Uneven.
Predictive AI healthcare systems are only as strong as the diagnostic infrastructure they sit on.
View
MWC 2026: Oppo, MediaTek Join Hands to Showcase New On-Device AI Capabilities for Future Smartphones
Oppo and MediaTek jointly announced new on-device artificial intelligence (AI) capabilities for future smartphones at the Mobile World Congress (MWC) 2026. The two companies are collaborating to develop and deploy new AI features for Oppo phones that do not require cloud connectivity to function. The announcements were made at MediaTek's AI for Life keynote on Wednesday. The main highlight was a preview of the jointly developed Omni AI model, which can be deployed and run entirely on-device, powered by MediaTek's chipsets.
View
Google Introduces Gemini 3.1 Flash-Lite as Its Fastest and Most Cost-Efficient AI Model
Google introduced the Gemini 3.1 Flash-Lite artificial intelligence (AI) model on Thursday. Calling it the fastest and the most cost-efficient AI model in the Gemini 3 series, the Mountain View-based tech giant said it is designed for high-volume developer workloads. The model is currently not available to end users and has been reserved for developers and enterprises via specific channels. The company also claimed that the model's output speed is higher than that of the 2.5 series. Notably, the Gemini 3.1 Flash-Lite is currently only available in preview.
View
Pocket FM, Open AI Join Hands to Build AI-Native Entertainment Ecosystem
The Pocket FM-OpenAI partnership will enhance storytelling, localisation and production speed for thousands of creators worldwide.
View
Meta’s AI Strategy is Bigger Than Just Llama
Meta’s LLM drought may finally be coming to an end, with Chief AI Officer Alexandr Wang teasing new AI models.
View
