Interviews Love, Death + Robots – Volume 4: Tim Miller (Creator & Director) & Jennifer Yuh Nelson (Supervising Director) By Vincent Frei - 02/06/2025 Earlier this year, Tim Miller spoke to us about his animated anthology Secret Level. Now, he returns to discuss the latest season of Love, Death + Robots. Jennifer Yuh Nelson talked about season two of Love, Death + Robots in 2021. She later worked on The Sea Beast, before returning once again to the anthology universe. What was your overall vision for the fourth season of Love, Death and Robots and how did it evolve from previous seasons? Tim Miller (TM) // We have the same strategy as every volume – we try to pick the best stories we know of and provide a mix that is hopefully appealing to everyone. There are a lot of variables to consider. Including genre, tone, and style of animation, such as stop motion, CG, and 2D. We try not to have two stories that are too similar. For example, if there’s already a military sci-fi story, we avoid selecting another one. We like to mix humor, horror, sci-fi, fantasy, and anything else that we think might be interesting from either a story or animation perspective. How did you approach the balance between experimenting with new styles and maintaining the signature identity of the show? TM // Honestly, we just try and follow our gut. What we think is interesting as filmmakers, animators, and storytellers will also be interesting to the animation community and fans alike. So, we keep an eye out for new voices, filmmakers, and new ways of doing things to keep things interesting. I’m not sure we have an identity of the show. In fact, I think if we did have an identity, it would be that we don’t have an identity… but we try and do whatever we think is interesting. Jennifer Yuh Nelson (JYN) // The fortunate thing about LDR is that the signature itself is experimenting with new styles. The trick is finding new aggressively experimental styles that still communicate to a mass audience. The stories are key to that. If the story is engaging, even to an audience that doesn’t usually gravitate to animation, then you can make it looks as weird as you want. What are some of the key challenges you faced while overseeing this season and how do you tackle them? TM // This season, there was a lot going on in the animation community that created some challenges with getting work done, whether studios were too full or ceased to exist entirely. Everyone struggled with budgets. But I didn’t feel like it was a problem with our show but rather a problem with the entire industry. People were struggling. And then it’s just always difficult when your ambition is high, your budgets are reasonable but still challenging, and you have to wrangle hundreds of people to get on board with your vision. (JYN) // These shows take a long time to make. R&D for a look that doesn’t exist can take a lot of trial and error. For example, Emily Dean, who directed Very Pulse of the Machine last season, did For He Can Creep this season. She had a cool angle of making her episode look like lithography. That was very very hard, but somehow Polygon, the studio that made both shorts, came through with it. And I think it turned out very well. Can you talk about how you selected the different animation studios for this season? What made you decide to work with the studios involved? (JYN) // We’ve been very fortunate to have worked with amazing people and studios these last few seasons, so it made sense to float some stories by them again. But it really comes down to the stories, and how each leans towards a certain technique. For example, How Zeke Found Religion was holding a slot where we wanted something 2D. We went to Titmouse because they were great with pushing the boundaries of 2D animation, and they suggested Diego Porral as a director who could bring a modern edge. How do you ensure each studio’s unique visual style complements the story and tone of each episode? TM // I know this sounds a little mystical and I don’t mean it to be, but I think the story speaks to you about style. Some things just feels right, and you have an innate concept of what would be the best version of the story, whether it’s stop motion, CG, 2D animation, or even live action. When you start thinking about the story in a creative way, a style becomes apparent. Which is not to say there aren’t many ways to do things and tell stories, but we feel a best version becomes clear. (JYN) // We do a lot of research, not just into what the studios have done before, but also into what they wish to do but haven’t had the chance to do. Often it’s just a matter of getting to know them and seeing if they have a philosophy of pushing for experimentation and risk. Then we try to support them as much as possible in their creative R&D. You both directed episodes for this season, what was that experience like? How did it differ from your work as overseeing directors? TM // For me, it’s really just trying to create the best story and I love working with the artists and trying to be open to what everybody brings to the table because everybody wants to do the best possible episode they can. I try and be open to letting people help carry that load. The best thing about being a director is that you get to pick and choose between all the great ideas that everybody has and shape the narrative by getting the benefit of everyone’s expertise and talent. (JYN) // It’s a different mindset. As a Supervising Director, I help. As a Director, I do. On episodes I’m not directing, I am deciphering that director’s ambition and pushing for whatever is required to make that absolutely great. On an episode I’m directing, every choice and image has to go through my brain so it’s more a reflection of my personal taste. Plus I tend to storyboard a lot more on my own episodes since it’s a way for me to communicate to the crew. I storyboard a lot on other episodes, but mainly to help figure out problems here and there. It also doesn’t come out of those director’s budgets so the free storyboarding is often welcomed. How did you choose the episode you worked on yourself and what aspects of it made it resonate with you both? TM // In my case with “The Screaming of the Tyrannosaur,” it was really by default. I had written the episode for Zack Snyder but Zack was too busy, and by that time I’d already fallen in love with the story, so I figured, why not just do it myself? As for “Golgotha,” I always loved the story. It was very efficient and short, which is hard to find in a story – it felt like a full meal. It has a beginning, middle, and end and it resolves in a satisfying way. “Golgotha” had all of that, plus it was funny. (JYN) // Spider Rose was on the story wall since the beginning. It was one of the “special” ones- very hard, ambitious, uncomfortable. Over the seasons we offered it to different directors and they veered away from it for one reason or another. But it glowed with a complexity that’s rare in a short story. I think that’s because it was written as an exploration for a far larger world that Bruce Sterling was developing. For me, it was the raw emotionality that drew me in. It’s how I understand how to communicate any story. And I love the way Spider Rose draws you in with emotion then shivs you with it. Were there any episodes in this season that particularly pushed the boundaries of what you had done before? How did that push happen? TM // I think “How Zeke got Religion” pushed the boundaries of 2D animation. The amount of detail and action that the guys at Titmouse were able to pull off was truly astonishing. Once again, Robert Valley outdid himself with 400 boys. The action scene at the end was one of my favorite pieces of animation in all of Love, Death, and Robots. (JYN) // Golgotha, Tim’s episode is live action, which is a rarity for the show. There was one live action episode in season 1, but none since. It is primarily an animation series, but nowadays, the line is so fuzzy that it seemed to make sense. How do you balance creative freedom with the thematic unity required for a show like Love, Death, and Robots? TM // There isn’t really a thematic unity. We’re just trying to create the best version of each of the episodes. They don’t tie into each other, they don’t relate to each other, they aren’t supposed to be about either Love, Death or Robots – the title is a meant to be a “catchall” that could hold ANY story or visual art we thought might be cool. Hopefully, the overall assemblage feels like a balanced meal with a little bit of something for everybody. But thematically speaking, again, I think our theme is that there is no theme. (JYN) // We try to set the foundation with a good story, based off the many short stories Tim has read over the years. Then the HOW of what that story becomes is the wooly Wild West. The directors and studios are fully encouraged to push all the boundaries of how to make these as innovative, impractical, and beautiful as they want. And, since each short is under 15 minutes, the studios we choose can be as experimental and scrappy as each story demands. Looking at the overall direction of the season, was there any unexpected moments or surprises that stood out to you during production? TM // Yeah, I think the color palette for “Zeke” was a shock to me in a wonderful way because it was completely unexpected and nothing I would ever do as a director but boy did I love it. And I think that “Can’t Stop” was an interesting addition. We wanted to do a music video from volume 1 onward, and this was the moment that we took to do it. I think it’s the greatest concert video ever made. (JYN) // Why do we have so many cats and babies? I’ve no idea. But when we saw the first giant baby shots in 400 Boys, it was a rare joy. They walk like babies, real babies. And somehow that was both accurate and terrifying. Looking forward, where do you see the show heading in future seasons, are there any new themes or concepts you’d love to explore? TM // So many directors in the industry have asked if they could play in our sandbox, and I would like to expand our reach to get some established names. Not that we don’t want new talent – we will always want that – but it would be great to have some really fantastic directors who have accomplished big movies come and play with our stories. I also think there’s a version where we bring in some content that may have existed in other mediums like comic books and perhaps tell some larger stories that take more than one episode to tell. Truthfully, I’ve already got a some really interesting stories picked out for the next few seasons, — of course those will change as the show evolves, but they’re fascinating stories that explore the whole arc of history… past, present, and future and some of the big challenges that humanity is facing today. I’d be lying if I didn’t mention that many of them explore the future of what mankind will become with the advent of AI and how artificial intelligence and humanity’s future intersect. (JYN) // Often themes only show up afterwards. There is a bit of a “herding cats” energy to the show that promises surprises in the production process. But the point of a show like this is that it is surprising. It has its own energy, and sometimes we just have to listen to it rather than dictate. If you had the opportunity to create any kind of story for Love, Death, and Robots, what would your dream narrative and what type of animation style would you envision for it? TM // Well, I have to say that I love high-end 3D animation, and that’s what Blur does for a reason. And secondly, I’d like to do a kind of story that could be live action and has some vast scope to it, but we choose to do it in animation because we get more value from using the techniques that animation brings. We can tell a bigger story, with more scope, and more action than we would using any other methodology…. and it competes favorably with live action in terms of the kind of audience that comes to watch it. Not just fans of animation, but fans of good cinema. (JYN) // I’d love to see an anime episode, like a Tsutomu Nihei fight scene, or something by Katsuhiro Otomo. A big thanks for your time. WANT TO KNOW MORE?Blur Studio: Dedicated page about Love, Death + Robots: Volume 4 on Blur Studio website. © Vincent Frei – The Art of VFX – 2025

See why some game developers think their games are worth $80.

Director Danny Boyle dishes on how the Infected have evolved for '28 Years Later,' and making them more terrifying.

While updating, upgrading, and reimagining your home can transform it into your dream home, the logistics involved can be daunting, even if you have a general contractor managing the project for you. One of the biggest challenges is setting your budget—unless you have an unlimited budget (lucky you!), determining what a home renovation will involve based on what you can afford is a key aspect of your plan.It’s also one of the most confusing. When budgeting reality bumps up against your dream home fantasy, figuring out what to remove from the project can be difficult because of the emotional aspects involved—everything can seem equally necessary when you’re imagining your future life in the home. In order to pare things down in a coherent and rational way, take a page from business school experts and use the time-tested MoSCoW Method.What is the MoSCoW method?The MoSCoW method was innovated by software developer Dai Clegg in the 1990s as a way to prioritize components of a project in order to stay on schedule and within budget. Although it was initially envisioned as a software development tool (and more widely as a tool for managing business projects in general), it’s malleable enough that it can be ideal for varied circumstances—including getting control of a home renovation project.The method involves breaking all the aspects of your project into four buckets, represented by the letters M, S, C, and W (the Os are just there to make the name more readable). The categories are:Must-haves. These are aspects of the project that are non-negotiable and mandatory.Should-haves. These are parts of the project that aren’t absolutely necessary, but are relatively important.Could-haves. These are smaller details that can easily be removed from the project or added in later if budget allows.Won’t haves. These are aspects of the project that aren’t under consideration at all.The simplicity of the MoSCoW method is its main strength—it’s easy to whip your home reno budget into shape in a short time by plopping everything into the relevant bucket.Using MoSCoW to plan a home renovation projectWhen you’re planning out your home renovation, start dropping each aspect into a bucket as you go:Things that absolutely have to be done are Must Haves. For example, if part of your renovation is replacing a leaking roof, that’s a Must Have—you have no choice, so that cost is baked into your budget from the get-go.Major components of the project that aren’t absolutely necessary fall into the Should Have bucket. If your old hardwood floors are worn but serviceable, replacing them is a big part of what you want to get out of the renovation—but you could leave them in place, or try to refinish them instead of replacing them. These would be the last parts of your project that you remove or downgrade.Grace notes and luxuries go into the Could Have bucket and held there pending how the budget plays out. For example, maybe you’d like your new flooring to have radiant heating. That’s nice, but not an absolute necessity. If money opens up later in the planning, you can toss it in.Finally, there’s the Won’t Have bucket. This might seem like an unnecessary step, since anything not already sorted into a bucket could be considered a Won’t Have. But the exercise of specifically labeling it as a Won’t Have is useful because it brings clarity to your priorities. If you find more budget later, you’ve already prioritized the Could Haves as more deserving of rescue. Won’t Haves aren’t necessarily things you’ll never do—they’re just things you’re not doing now. For example, maybe your HVAC system is a little old and you’re thinking it will need to be replaced in the next few years—but not at this moment, because you’re spending your money on all these other projects. So you put that into the Won’t Have bucket because you know you’ll be returning to it in the future.Once you’ve done an initial categorization of your home renovation, you can start crunching numbers to see if changes are necessary or desired. Maybe you decide, on reflection, that a Should Have is really a Must Have, or vice versa. And if your budget can’t deliver on every priority, you can shift some things into the Could Have bucket and hold them in reserve for the future.Any successful home renovation is as much about planning and prioritization as it is about budget and schedule management, and the MoSCoW Method can help ensure your project is on track before anyone touches a power tool.

Nvidia's RTX Remix is a remarkable tool that allows game modders to bring state-of-the-art path traced visuals to classic PC games. We've seen Portal RTX from Nvidia already, along with the development of a full-on remaster of Half-Life 2 - but I was excited to see a community of modders take on 2004's Painkiller, enhanced now to become Painkiller RTX. It's still a work-in-progress project as of version 0.1.6, but what I've seen so far is still highly impressive - and if you have the means, I recommend checking it out. The whole reason RTX Remix works with the original Painkiller is due to its custom rendering technology, known as the PainEngine. This 2004 release from People Can Fly Studios was built around Direct X 8.1, which gave it stellar visuals at the time, including bloom effects – specular lighting with limited bump mapping and full framebuffer distortion effects. Those visuals dazzled top-end GPU owners of the time, but like a great number of PC releases from that era, it had a DX7 fallback which culled the fancier shading effects and could even run on GPUs like the original GeForce. RTX Remix uses the fixed function DX7 path and replaces the core rendering with the path tracer - and that is how I have been playing the game these last few days, taking in the sights and sounds of Painkiller with a new lick of paint. It's an upgrade that has made me appreciate it all the more now in 2025 as it is quite a special game that history has mostly forgotten. To fully enjoy the modders' work on the path-traced upgrade to Painkiller, we highly recommend this video.Watch on YouTube Painkiller is primarily a singleplayer first-person shooter that bucked the trends of the time period. After Half-Life and Halo: Combat Evolved, many first person shooters trended towards a more grounded and storytelling-based design. The classic FPS franchises like Quake or Unreal had gone on to become wholly focused on multiplayer, or else transitioned to the storytelling route - like Doom 3, for example. Painkiller took all of those 'modern' trappings and threw them in the garbage. A narrative only exists in a loose sense with pre-rendered video that bookends the game’s chapters, acting only as a flimsy excuse to send the player to visually distinct levels that have no thematic linking beyond pointing you towards enemies that you should dispatch with a variety of weapons. The basic gameplay sounds familiar if you ever played Doom Eternal or Doom 2016. It is simple on paper, but thanks to the enemy and level variety and the brilliant weaponry, it does not get tiring. The game enhanced its traditional FPS gameplay with an extensive use of Havok physics – where a great deal of the game’s environmental objects could be broken up into tiny pieces with rigid body movement on all the little fragments, or environmental objects could be manipulated with ragdoll or rope physics. Sometimes it is there for purely visual entertainment but other times it has a gameplay purpose with destructible objects often containing valuable resources or being useful as a physics weapon against the game's enemies. So, what's the score with Painkiller RTX? Well, the original's baked lighting featured hardly any moving lights and no real-time perspective-correct shadows - so all of that is added as part and parcel of the path-traced visuals. The RTX renderer also takes advantage of ray-traced fog volumes, showing shadows in the fog in the areas where light is obscured. Another aspect you might notice is that the game’s various pickups have been now made to be light-emissive. In the original game, emissives textures are used to keep things full bright even in darkness, but they themselves emit no light. Since the path tracer fully supports emissive lighting from any arbitrary surface, they all now cast light, making them stand out even more in the environment. To see this content please enable targeting cookies. The original game extensively used physics objects, which tended to lead to a clash in lighting and shading for any moving objects, which were incongruous then with the static baked lighting. Turn on the path tracer and these moving objects are grounded into the environment with shadows of their own, while receiving and casting light themselves. Boss battles are transformed as those enemies are also fully grounded in the surrounding environments, perfectly integrated into the path-traced visuals - and even if the titanic enemies are off-screen, their shadows are not. The main difference in many scenes is just down to the new lighting - it's more physicalised now as dynamic objects are properly integrated, no longer floating or glowing strangely. One reason for this is due to lighting resolution. The original lighting was limited by trying to fit in 256MB of VRAM, competing for space with the game’s high resolution textures. Painkiller RTX's lighting and shadowing is achieved at a per-pixel level in the path tracer, which by necessity means that you tend to see more nuance, along with more bounce lighting as it is no longer erased away by bilinear filtering on chunky light map textures. Alongside more dynamism and detail, there are a few new effects too. Lit fog is heavily used now in many levels - perhaps at its best in the asylum level where the moonlight and rain are now illuminated, giving the level more ambience than it had before. There is also some occasional usage of glass lighting effects like the stain glass windows in the game now filtering light through them properly, colouring the light on the ground in the pattern of the individual mosaic patterns found on their surface. Half-Life 2 RTX - built on RTX Remix - recently received a demo release. It's the flagship project for the technology, but modders have delivered path traced versions of many modern games.Watch on YouTube New textures and materials interact with the path tracer in ways that transform the game. For some objects, I believe the modders used Quixel megascan assets to give the materials parallax along with a high resolution that is artistically similar to the original game. A stoney ground in the graveyard now actually looks stoney, thanks to a different texture: a rocky material with craggy bits and crevices that obscure light and cast micro shadows, for example. Ceramic tiles on the floor now show varying levels of depth and cracks that pick up a very dull level of reflectivity from the moon-lit sky. Some textures are also updated by running them through generative tools which interpret dark areas of the baked textures as recesses and lighter areas as raised edges and assigns them a heightmap. This automated process works quite well for textures whose baked features are easily interpreted, but for textures that had a lot of noise added into them to simulate detail, the automated process can be less successful. That is the main issue I would say with the RTX version so far: some of these automated textures have a few too many bumps in them, making them appear unnatural. But that is just the heightmap data as the added in material values to give the textures sheen tend to look universally impressive. The original game barely has any reflectivity, and now a number of select surfaces show reflections in full effect, like the marble floors at the end of the game's second level. For the most part though, the remix of textures from this mod is subtle, with many textures still being as diffuse as found in the original game: rocky and dirty areas in particular look much the same as before, just with more accurately rendered shadows and bounce lighting - but without the plasticy sheen you might typically find in a seventh generation game. Whether maxed on an RTX 5090 or running on optimised settings on an RTX 4060, the current work-in-progress version of Painkiller RTX can certainly challenge hardware. | Image credit: Digital Foundry Make no mistake though: path tracing doesn't come cheap and to play this game at decent frame-rates, you either need to invest in high performance hardware or else accept some compromises to settings. Being a user mod that's still in development, I imagine this could improve in later versions but at the moment, Painkiller RTX maxed out is very heavy - even heavier than Portal RTX. So if you want to play it on a lower-end GPU, I recommend my optimised settings for Portal RTX, which basically amounts to turning down the amount of possible light bounces to save on performance and skimping a bit in other areas. Even with that, an RTX 4060 was really struggling to run the game well. With frame generation on and DLSS set to 1080p balanced with the transformer model, 80fps to 90fps was the best I could achieve in the general combat zones, with the heaviest stages dipping into the 70s - and even into the 60s with frame generation. The mod is still work-in-progress, but even now, Painkiller RTX is still a lot of fun and it can look stunning if your hardware is up to it. But even if you can't run it, I do hope this piece and its accompanying video pique your interest in checking out Painkiller in some form. Even without the path-traced upgrade, this is a classic first-person shooter that's often overlooked and more than holds its own against some of the period's better known games.

The online servers for platform fighting game MultiVersus have officially been taken down. While players will no longer to match up with others online, WB Games had noted that players that had logged in between between February 4 and May 30 will still be able to play the title. WB Games had originally announced that May 30 would be the last day for the MultiVersus servers all the way back in January. In its announcement, the company had said that the game’s fifth season would serve as its final update. Along with taking the servers offline, MultiVersus will also be delisted from all digital storefronts, including Steam, Epic Games Store, PlayStation Store and Microsoft Store. This means that no new players will be able to buy the game going forward. “This decision was not made lightly,” posted developer Player First Games back when it had announced the servers being shut down. “All of us on the Player First Games and Warner Bros Games teams have put our heart and soul into MultiVersus, creating a unique social and cooperative experience from the ground up that brings together a variety of popular characters and worlds.” “We are very proud of the game we have built and are incredibly thankful for the support of the MultiVersus community throughout this journey. Your passion and enthusiasm have been unmatched.” The servers being shut down comes just a few months after Player First Games itself was also closed. The studio’s closure came alongside the closure of WB Games San Diego, and Monolith, as well as an in-development game based on the Wonder Woman franchise. “This is a strategic change in direction and not a reflection of these teams or the talent within them,” said the company in a statement announcing the studio closures. While MultiVersus had scene some success back when it originally launched in closed alpha all the way back in 2022, the game lost most of its player base quickly, before the alpha was shut down. It later got a full relaunch in May 2024 on PC, PS4, PS5, Xbox One and Xbox Series X/S. However, despite the relaunch, the game was never able to get back the traction it had originally built up. Rumours have indicated that the commercial failure of MultiVersus may have contributed to Warner Bros. Games facing quite a bit of heat in terms of its financial state, leading the company to shut down development on Monolith’s Wonder Woman game. MultiVersus, along with Rocksteady’s Suicide Squad: Kill the Justice League had been implicated in this by Bloomberg’s Jason Schreier. “Last year Suicide Squad was a humongous flop – they wrote off $200 million because of that. MultiVersus and Quidditch Champions, also both flops, [they] wrote off another $100 million because of that,” said Schrier, pointing out that WB Games’ forays into live-service games had just not been working out for the company under the leadership of former head David Haddad. Haddad has since stepped down from his position, being succeeded by JB Perrette, who also serves as the CEO and president of Warner Bros. Discovery’s global streaming and games division.

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Token Monster, a new AI chatbot platform, has launched its alpha preview, aiming to change how users interact with large language models (LLMs). Developed by Matt Shumer, co-founder and CEO of OthersideAI and its hit AI writing assistant Hyperwrite AI, Token Monster’s key selling point is its ability to route user prompts to the best available LLMs for the task at hand, delivering enhanced outputs by leveraging the strengths of multiple models. There are seven major LLMs presently available through Token Monster. Once a user types something into the prompt entry box, Token Monster uses pre-prompts developed through iteration by Shumer himself to automatically analyze the user’s input, decide which combination of multiple available models and linked tools are best suited to answer it, and then provide a combined response leveraging the strengths of said models. The available LLMs include: Anthropic Claude 3.5 Sonnet Anthropic Claude 3.5 Opus OpenAI GPT-4.1 OpenAI GPT-4o Perplexity AI PPLX (for research) OpenAI o3 (for reasoning) Google Gemini 2.5 Pro Unlike other chatbot platforms, Token Monster automatically identifies which LLM is best for specific tasks — as well as which LLM-connected tools would be helpful such as web search or coding environments — and orchestrates a multi-model workflow. “We’re just building the connectors to everything and then a system that decides what to use when,” said Shumer. For instance, it might use Claude for creativity, o3 for reasoning, and PPLX for research, among others. This approach eliminates the need for users to manually choose the right model for each prompt, simplifying the process for anyone who wants high-quality, tailored results. Feature highlights The alpha preview, which is currently free to sign up for at tokenmonster.ai, allows users to upload a range of file types, including Excel, PowerPoint, and Docs. It also includes features such as webpage extraction, persistent conversation sessions, and a “FAST mode” that auto-routes to the best model without user input. At the heart of Token Monster is OpenRouter, a third-party service that acts as a gateway to multiple LLMs, and into which Shumer has invested a small sum, by his admission. This architecture lets Token Monster tap into a range of models from different providers without having to build separate integrations for each one. Pricing and availability As of right now, Token Monster does not charge a flat monthly fee. Instead, users only pay for the tokens they consume through OpenRouter, making it flexible for varying levels of usage. According to Shumer, this model was inspired by Cline, a tool that enables high-spending users to access unlimited AI power, allowing them to achieve better outputs by simply using more compute resources. Multi-step workflows produce richer LLM responses Token Monster’s AI workflows extend beyond simple prompt routing. In one example, the chatbot might start with a research phase using web search APIs, pass that data to o3 for identifying information gaps, then create an outline with Gemini 2.5 Pro, draft text with Claude Opus, and refine it with Claude 3.5 Sonnet. This multi-step orchestration is designed to provide richer, more complete answers than a single LLM might be able to generate alone. The platform also includes the ability to save sessions, with data securely stored using the open source online database service Supabase. This ensures that users can return to ongoing projects without losing their work, while still giving them control over what data is saved and what is ephemeral. A non-traditional CEO In a notable experiment, Token Monster’s leadership has been handed over to Anthropic’s Claude model. Shumer announced that he is committed to following every decision made by “CEO Claude,” calling it a test to see whether an AI can manage a business effectively. “Either we’ve revolutionized management forever or made a huge mistake,” he wrote on X. Emerging from the Reflection 70-B controversy Token Monster’s launch comes less than a year after Shumer faced controversy over his launch and ultimate retraction of Reflection 70B, a fine-tuned version of Meta’s Llama 3.1 that was initially touted as the most highly performant open source model in the world, but which quickly became subject to criticism and accusations of fraud after third-party researchers were unable to reproduce its stated performance on third-party benchmark tests. Shumer apologized and said the issues were born out of mistakes made due to speed. The episode underscored the challenges and risks of rapid AI development and the importance of transparency in model releases. MCP integrations coming next Shumer said his team on Token Monster is also exploring new capabilities, such as integrating with Model Context Protocol (MCP) servers that allow websites and companies to have LLMs make use of their knowledge, tools, and products to achieve higher-order tasks than just text or image generation. This would enable Token Monster to connect with a user’s internal data and services, opening possibilities for it to handle tasks like managing customer support tickets or interfacing with other business systems. Shumer emphasized that Token Monster is still very much in its early stages. While it already supports a suite of powerful features, the platform remains an alpha product and is expected to see rapid iterations and updates as more users provide feedback. “We’re going to keep iterating and adding things,” he said. A promising experiment For users who want to take advantage of the combined power of multiple LLMs without the hassle of model switching, Token Monster could be an appealing choice. It’s designed to work for people who don’t want to spend hours tweaking prompts or testing different models themselves, instead letting the system’s automated routing and multi-step workflows handle the complexity. As Token Monster’s capabilities grow, it will be interesting to see how users and businesses adopt it — and how its experiment with AI-led management pans out. For now, it’s a promising addition to the rapidly expanding landscape of AI chatbots and digital assistants. Daily insights on business use cases with VB Daily If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI. Read our Privacy Policy Thanks for subscribing. Check out more VB newsletters here. An error occured.