Mozilla.ai

Owners, not renters. Our Alejandro González Pérez and Thomas Bille joined Mozilla's live on open, local AI. Watch it here: https://lnkd.in/e29Ns3ck

Your agent searches the web dozens of times a day. Do you know whose search engine is reading those queries? They can include the name of an internal service or the customer you're researching, and none of the big model providers says which engine sits behind their default search. It can also change without your code noticing. David de la Iglesia Castro and Davide Eynard wrote about why this matters and how Otari lets you keep the model you already use while choosing the search backend yourself, whether that's your own Brave key or a SearXNG endpoint you host. Read the full post: https://lnkd.in/evpgxdD9

Ever swapped a hosted model for an open one and watched your agent quietly fall apart? The model usually isn't the problem. It's everything the hosted API was doing for you that you never noticed. In this engineering spotlight, David de la Iglesia Castro breaks down what we learned moving a real agent app off a proprietary API, and how Otari by Mozilla AI fills those gaps without locking you in. Inside: → The hidden jobs hosted APIs do behind the scenes → How Otari lets you choose your own search and sandbox providers → The hard problems we're still working on Read it here: https://lnkd.in/eHc8begu #OpenSource #AIAgents #Otari #MozillaAI

We're live at Shift AI Barcelona! The Otari by Mozilla AI team is at 𝗕𝗼𝗼𝘁𝗵 𝟭𝟯 and ready to chat. The morning has been busy, with plenty of conversations about what it really takes to run AI in production, from keeping control as usage grows to staying flexible across models and providers. Adrian G. M., Caroline Bohu and Niko Thielsch are here today and tomorrow. Whether you want to talk AI infrastructure, open source, or share what's working (and what isn't) in your own setup, come and say hello. We'll also have stickers and some goodies on the table. 👋 #ShiftAIBarcelona #MozillaAI #Otari #OpenSourceAI #AIInfrastructure

Meet Le Chonk. 🐱 Mistral Large 4, Mistral's largest model yet, is live in Otari by Mozilla AI from day one, along with the full Mistral model family. → 1.05T-parameter Mixture of Experts, 49B active → Natively multimodal → Strong reported results on coding, agentic and enterprise workloads In Otari, trying it means changing one model string. Your existing policies, guardrails, traces and cost tracking apply from the first request. A new model shouldn't mean a new stack. That's what Otari, Mozilla AI's open-source gateway and policy layer, is built for. Read the announcement: https://lnkd.in/evXFwsvZ #OpenSource #LLM #AIAgents #MistralAI

𝗪𝗵𝗮𝘁 𝘄𝗼𝘂𝗹𝗱 𝗴𝗶𝘃𝗲 𝘆𝗼𝘂 𝗺𝗼𝗿𝗲 𝗰𝗼𝗻𝗳𝗶𝗱𝗲𝗻𝗰𝗲 𝘁𝗼 𝘀𝗰𝗮𝗹𝗲 𝗔𝗜? That’s the question on our backdrop at Shift AI in Barcelona. We’re asking the engineering leaders we meet to share their answers. As AI moves into production, choosing the right models is only part of the challenge. Teams need to understand how they're using AI and maintain control as adoption grows. They also need the flexibility to work across models and providers without unnecessary lock-in. That’s where Otari by Mozilla AI comes in. Otari is our open-source gateway and policy layer for AI infrastructure. It gives organisations a unified control layer to connect applications and agents to models and tools, govern how AI is used, and understand what happened, why it happened, and what it cost. But we’d rather hear your answer than guess it. If you’re at Shift AI, stop by and tell Adrian G. M., Caroline Bohu, Niko Thielsch, or share your answer through our survey: https://bit.ly/4dmFYvd #ShiftAIBarcelona #LLMOps #OpenSourceAI #SoftwareEngineering #MozillaAI

Your AI agent has access to your repo, your tools, and your credentials. So what is it actually 𝘢𝘭𝘭𝘰𝘸𝘦𝘥 to do? Most teams answer with a line in AGENTS.md. But an instruction isn't a constraint. A rule followed 99% of the time still leaves you with 1% of thousands of actions going wrong. Alejandro González Pérez breaks down why the next big AI question isn't "what can the model do?" but "what is it allowed to do, and can you prove it?" In the post: • Why limits must hold even when model judgment fails • Why every agent action needs a record of the rule behind the decision • Why that layer should be open and yours, so it outlives any vendor This is the thinking behind Otari and our Agent Guardrails work. → Read the post: https://lnkd.in/eVpnPCbW → Explore Otari on GitHub: https://lnkd.in/efzWiqSR #AIAgents #AIGovernance #OpenSource #CodingAgents #Otari

Adrian G. M. is at Shift AI Barcelona representing Otari by Mozilla AI. For teams putting AI agents to work, from individual developers to large enterprises, Otari by Mozilla AI is the open-source agent governance layer that enforces the rules your team writes, on any model or cloud, and keeps every agent action on the record. If you're interested in AI infrastructure, governance or building production-ready AI systems, come and say hello. 📍 Find Adrian G. M. at Shift AI or send him a DM to connect. #ShiftAI #MozillaAI #Otari #AIInfrastructure #OpenSourceAI #AIDevelopers

Niko Thielsch is at Shift AI in Barcelona representing Otari by Mozilla AI. For teams putting AI agents to work, from individual developers to large enterprises, Otari by Mozilla AI is the open-source agent governance layer that enforces the rules your team writes, on any model or cloud, and keeps every agent action on the record. Niko is meeting engineering leaders and business decision makers to discuss how organisations can build and scale AI with greater visibility, governance and flexibility; without unnecessary vendor lock-in. If you're evaluating AI infrastructure or thinking about what's next beyond individual models, stop by for a conversation. Or send Niko Thielsch a message to connect during the event. #ShiftAI #MozillaAI #Otari #AIInfrastructure #OpenSourceAI #EnterpriseAI

Caroline Bohu is at Shift AI Europe in Barcelona representing Otari by Mozilla AI. For teams putting AI agents to work, from individual developers to large enterprises, Otari by Mozilla AI is the open-source agent governance layer that enforces the rules your team writes, on any model or cloud, and keeps every agent action on the record. As organizations move AI into production, they're looking for more than access to models; they need the infrastructure to connect AI applications and agents, apply governance, and stay in control as they scale. If you're exploring how to build AI infrastructure that's open, flexible and enterprise-ready, Caroline would love to chat. 📍 See you at Shift AI Europe or DM Caroline to book some time. #ShiftAI #MozillaAI #Otari #AIInfrastructure #OpenSourceAI #EnterpriseAI