DecodeFM

How do you scale high-performing video ads when creators are expensive, slow to hire, and hard to test at volume?Today, we look at ArcAds, a startup using AI actors to generate realistic, UGC-style video ads in minutes instead of weeks. By lowering the cost and friction of testing creative, ArcAds is reshaping how marketers approach performance advertising. It has attracted $16 million in recent seed funding.

Corporate legal teams face rising costs and slow turnaround times, driven by hourly billing models that reward inefficiency. Today, we take a look at Eudia, which is trying to address these issues by combining AI systems with an in-house legal practice. Enabled by regulatory changes in Arizona, Eudia offers a case study in how AI and new business structures may reshape high-volume legal work. 

Happy New Year! Welcome to our first episode of the year.As we kick off a new year, is the future of AI about who trains the biggest models, or who can run them most efficiently?Let's take a look at Together AI, a company focused on making open-source models fast, affordable, and usable at scale. By pairing open models with deep systems optimization and flexible deployment, Together AI is arguing that the next phase of AI competition is really about infrastructure.

As 2025 comes to a close, we’re ending the year with a slight shift. We’ve spent most of the year focused on AI applications and infrastructure - today, we’re looking at how AI is starting to change security.This episode looks at how AI-driven automation is reshaping security operations, helping teams move faster with less manual work, and what that could mean as we head into 2026.

What would it take to challenge NVIDIA’s dominance in AI compute? Today, we’re looking at TensorWave, a startup betting big on an all-AMD cloud, a one-gigawatt infrastructure deal, and a fast, debt-fueled expansion. We’ll break down how this model could reshape competition in AI computing, and the risks that come with it.

Nscale is a fast-growing infrastructure company focused on large-scale AI data centers. Fresh off a $433 million Series B that valued the company at $1.1 billion, nscale has announced a massive $14 billion data center deal with Microsoft. We’ll break down what nscale is building, why this partnership matters, and what it says about where AI infrastructure is headed.

Today, we’re looking at Retell AI, a company building ultra-realistic voice agents that are already replacing real call-center workflows. Retell has grown from $8 million to $36 million in ARR in just eight months, all with a team of only 21 full-time employees. We’ll unpack what makes their tech so compelling, how businesses are using it today, and why voice automation is becoming one of the most competitive corners of AI.

What if one of the most interesting challenges to today’s AI giants isn’t coming from Silicon Valley, but from a young team in Paris? Today, we’re looking at Mistral, the fast-rising European startup taking an open-weight approach to the generative-AI race. With the recent release of their Mistral 3 model family, they’re positioning themselves as a cost-efficient, highly customizable alternative through Mixture-of-Experts models that deliver strong performance at lower operational cost. Backed by major partnerships and a multibillion-dollar valuation, Mistral is betting that efficiency and openness can carve out meaningful foothold in the enterprise AI market.

What if the best way to power AI isn’t expanding the grid, but moving compute to where energy is cheapest? Today, we’re looking at Crusoe - the company turning wasted and stranded energy into an edge for high-performance computing. With a fresh $1.375 billion Series-E round and a valuation topping $10 billion, they’re quickly becoming a key player in the infrastructure behind the AI boom. 

Today, we’re looking at Nuro and its major strategic shift in the autonomous vehicle space. Once known for its cute delivery robots, the company has now pivoted toward licensing its full AV stack to automakers and mobility partners. It has recently raised Series-E at $6 billion valuation.

What if the real bottleneck in AI isn’t the model, but the data itself? In this episode, we’re talking about Snorkel AI, a Stanford spinout helping enterprises turn expert knowledge into high-quality training and evaluation data, far faster than manual methods. The company recently raised $100 million in Series D funding at a $1.3 billion valuation. We’ll explore how Snorkel’s approach works, its shift toward AI evaluation, and what that means for the next wave of enterprise AI.

Today, we’re turning to the networking side of AI - the part that keeps all those GPUs talking to each other. Nexthop AI recently came out of stealth with $110 million in funding at a $500 million valuation. The company rethinking how massive AI clusters are connected - not just building faster switches, but co-developing custom network hardware and software with hyperscalers. We’ll look at how Nexthop’s approach could reshape the AI infrastructure stack and why investors are betting big on it.

Today, we’re talking about Reflection AI — the new startup everyone’s watching. Fresh with a $2 billion raise at an $8 billion valuation, it’s founded by ex-DeepMind researchers behind AlphaGo and Gemini. Their goal? To build open, frontier-scale intelligence that rivals the closed labs like OpenAI and Anthropic. We’ll dig into their technology and discuss the future of open-weight models.

Today, we’re looking at Hippocratic AI, a startup building generative AI agents for patient-facing, non-diagnostic healthcare tasks. From post-discharge check-ins to chronic care management, Hippocratic’s goal is to ease staff shortages and create what it calls “healthcare abundance.” The company recently raised a $141 million Series B at a $1.6 billion valuation, and we’ll explore how its safety-focused approach and deep clinician involvement could reshape the way patients and providers experience care.

Today, we’re looking at Fyxer AI, a startup building AI executive assistants for email and meetings. Fyxer plugs directly into Gmail and Outlook to sort messages, draft replies, and even handle scheduling and notes — all to save users an hour a day. The company has grown ARR from $1 million to $17 million in just eight months, and we’ll explore how its rapid rise is reshaping the future of workplace productivity.

Today’s episode is a little different — we’re talking about defense AI. Shield AI builds autonomous systems designed to protect soldiers and civilians by enabling aircraft and drones to operate without GPS or communication links. We’ll explore the company’s origin story, its core software platform HiveMind, and how its AI-powered drones are changing modern warfare. We’ll also look at its $5.3 billion valuation, key defense partnerships, and the ethical questions around AI on the battlefield.

Today, we’re looking at Groq, a startup aiming to redefine AI inference speed. Groq recently raised $750 million, bringing its valuation to $6.9 billion. We’ll explore Groq’s founding story, its specialized LPU chip, and how its focus on real-time AI performance positions it against Nvidia and other giants in the GPU cloud market.

Today, we’re discussing Decagon and how they’re rethinking customer support with human-like AI agents. We’ll look at how their generative AI engine differs from traditional chatbots, the impact of integrations like Stripe, and what it means for enterprises facing rising support costs. Plus, we’ll explore where this technology might go next, from personalization to new applications beyond customer service.

In this episode we explore the differences between U.S. and Chinese venture capital, from software-driven models to hardware and supply chain innovation. It highlights how biases shape investment, the role of trust networks, and why overcapacity can fuel new opportunities.

In this episode, we talk about CalAI, a fast-growing nutrition app that tracks daily calories through photos. Its founders are teenagers. While some online reports speculate its ARR could be over $30 million, we’ll focus on how the app works and its challenges.