Gimei Alex

"The great endeavor of our age is an act of alchemy: the transmutation of humanity’s informational disorder into artificial intelligence." @ShayneRedford

Co-founders @sarahookr and @sudip_r0y talk adaptive intelligence. AI shouldn’t be model-centric. It should be task-centric. It should adapt to you.

Stop inheriting intelligence. Shape it. Join us to build the next era of intelligence @adaption_ai Hiring across research and engineering. Very exciting time to join. We cooking something very special for upcoming releases bridging adaptive intelligence + environments.

You don't have a model problem. You have a data problem. Most fine-tuning teams already know this. Almost none have the tooling to fix it systematically. Today that changes. Together AI fine-tuning is now live inside Adaptive Data. Dataset shaping to deployed model. LoRA to 100B+. One platform.

𝟔 𝐀𝐌. 𝐀𝐩𝐩 𝐮𝐩𝐝𝐚𝐭𝐞𝐝 𝐢𝐧 𝐭𝐡𝐞 𝐜𝐚𝐫. 𝐖𝐚𝐥𝐤𝐞𝐝 𝐢𝐧𝐭𝐨 𝐚 𝐜𝐥𝐢𝐧𝐢𝐜 𝐢𝐧 𝐊𝐚𝐭𝐢𝐤𝐚𝐦𝐮. 𝐇𝐞𝐫𝐞'𝐬 𝐰𝐡𝐚𝐭 𝐭𝐡𝐞 𝐃𝐢𝐬𝐭𝐫𝐢𝐜𝐭 𝐇𝐞𝐚𝐥𝐭𝐡 𝐎𝐟𝐟𝐢𝐜𝐞𝐫 𝐬𝐚𝐢𝐝. Our Lead Designer Catherine Tuka and the field team are deploying the AI Health Companion across health facilities in Luweero District this week. The routine: wake before dawn, update the app on the drive, arrive at the facility, meet the in-charge, set up research devices, and watch health workers interact with AI clinical decision support — offline, on the phone, in the clinic where they see patients every day. At Katikamu Health Centre III, the District Health Officer shared his perspective on what this technology means for his district. 📍 Katikamu HC III, Luweero District This study is conducted under the ethical oversight of @Makerere . Supported by @Google.org | Built on @GoogleDeepMind 's MedGemma & MedASR | Optimised on @nvidia DGX Spark

This month, we’re celebrating the anniversaries of two Open Science Community Leads whose contributions continue to shape and strengthen our open science community.

We believe that intelligence should not arrive preconfigured. @togethercompute is now available directly inside the Adaption platform, connecting Adaptive Data with large-scale training in a single workflow. One platform, end to end. Stop inheriting intelligence. Shape it.

🚀 Our community-led ML Agents group is kicking off a new collaborative project to build a Street Navigation Agent for more inclusive, region-aware local search. In many parts of the world, businesses exist physically — but not digitally. They're exploring how AI can use tools like Google Street View to read storefront signs, apply distance & category constraints, and reason step-by-step to identify real-world services. We’re also building a global benchmark across countries and languages to evaluate visual verification.

What is crazy is we adapt data so fast you can still join the competition today, think about a meaningful long-tail problem to work on and be in a position to win by this Friday. So join us. :) Big shoutout to @sudip_r0y for building incredibly fast infrastructure.

At scale, volume isn't the problem. Precision that adapts to the real world is. @sarahookr and @sudip_r0y on why we built Adaptive Data.

200 𝐡𝐞𝐚𝐥𝐭𝐡 𝐰𝐨𝐫𝐤𝐞𝐫𝐬 𝐭𝐫𝐚𝐢𝐧𝐞𝐝. 𝐍𝐨𝐰 𝐰𝐞 𝐠𝐨 𝐛𝐚𝐜𝐤 𝐟𝐨𝐫 𝐭𝐡𝐞 𝐞𝐯𝐢𝐝𝐞𝐧𝐜𝐞. 𝐓𝐡𝐞𝐲 𝐭𝐫𝐢𝐞𝐝 𝐭𝐨 𝐝𝐨𝐰𝐧𝐥𝐨𝐚𝐝 𝐢𝐭 𝐛𝐞𝐟𝐨𝐫𝐞 𝐬𝐡𝐞 𝐟𝐢𝐧𝐢𝐬𝐡𝐞𝐝 𝐭𝐚𝐥𝐤𝐢𝐧𝐠. 𝐍𝐨𝐰 𝐰𝐞'𝐫𝐞 𝐠𝐨𝐢𝐧𝐠 𝐛𝐚𝐜𝐤. Two weeks ago, our Lead Designer @tukacatts stood in front of 200 health workers at the Luweero District Health Office and walked them through an AI clinical decision support tool she helped design one that runs entirely on their phones, offline, with no data leaving the device. She explained what it does. She showed them how it works. She watched a room full of nurses, midwives, clinical officers, and CHEWs lean forward. Then they started searching for the app on the Google Play Store , during her session. Others offered to delete TikTok to make room for it. A nurse with 30 years of clinical experience was told her phone couldn't run it. She refused to accept it. That was the workshop. That told us demand exists. 𝐓𝐨𝐝𝐚𝐲, 𝐂𝐚𝐭𝐡𝐞𝐫𝐢𝐧𝐞 𝐡𝐞𝐚𝐝𝐬 𝐛𝐚𝐜𝐤 𝐭𝐨 𝐋𝐮𝐰𝐞𝐞𝐫𝐨 , 𝐭𝐡𝐢𝐬 𝐭𝐢𝐦𝐞 𝐟𝐨𝐫 10+ 𝐝𝐚𝐲𝐬 𝐚𝐜𝐫𝐨𝐬𝐬 𝟏𝟔 𝐡𝐞𝐚𝐥𝐭𝐡 𝐟𝐚𝐜𝐢𝐥𝐢𝐭𝐢𝐞𝐬. The question shifts: it's no longer "𝘥𝘰 𝘩𝘦𝘢𝘭𝘵𝘩 𝘸𝘰𝘳𝘬𝘦𝘳𝘴 𝘸𝘢𝘯𝘵 𝘵𝘩𝘪𝘴?" It's "does it change how they practice?" Structured questionnaires. Clinical confidence assessments , before and after using the tool. In-depth interviews with the health workers who use it. Device performance data from every session. All under the ethical oversight of Makerere University. Research clearance secured. This is not a demo anymore. This is a clinical feasibility study. We'll be sharing the journey as it unfolds from the first facility visit to the last. Supported by Google.org | Built on Google DeepMind's MedGemma & MedASR | Optimised on @nvidia DGX Spark We thank @aistudiouganda our implementation partner.

The Uncharted Data Challenge closes Friday. $20,000 prize pool for open-source datasets that fill the gaps mainstream AI keeps missing. Evolve your data. Shape your AI.

𝐂𝐫𝐚𝐧𝐞 𝐀𝐈 𝐋𝐚𝐛𝐬 𝐣𝐨𝐢𝐧𝐬 𝐭𝐡𝐞 𝐔𝐍 𝐨𝐧 𝐓𝐮𝐞𝐬𝐝𝐚𝐲: 𝐔𝐍𝐃𝐏 (𝐀𝐈 𝐇𝐮𝐛 𝐟𝐨𝐫 𝐒𝐮𝐬𝐭𝐚𝐢𝐧𝐚𝐛𝐥𝐞 𝐃𝐞𝐯𝐞𝐥𝐨𝐩𝐦𝐞𝐧𝐭) × 𝐈𝐓𝐔. Frontier AI safety evaluation built from the Global South not a future possibility, a present reality. We're joining a panel jointly hosted by International Telecommunication Union , UNDP, and the AI Hub for Sustainable Development, AI Trust, Safety and Diffusion in Small States and Low- and Middle-Income Countries, alongside the African Observatory on Responsible AI, FAR.AI, and Arab American University Palestine. Moderated by Dr. Craig Ramlal of UWI St Augustine. 𝐖𝐡𝐚𝐭 𝐰𝐞'𝐫𝐞 𝐛𝐫𝐢𝐧𝐠𝐢𝐧𝐠 𝐭𝐨 𝐭𝐡𝐞 𝐜𝐨𝐧𝐯𝐞𝐫𝐬𝐚𝐭𝐢𝐨𝐧: → UCCB : the Ugandan Cultural Context Benchmark we built in Kampala, now part of the UK AI Safety Institute's Inspect Evals framework alongside TruthfulQA and HumanEval. The first African-origin benchmark in the AISI stack. → UCCB-Health : supported by a Google.org Health AI grant, an evaluation specification for frontier medical AI grounded in the Uganda Clinical Guidelines and tested on MedGemma X MedASR. → An offline-first clinical decision support stack now reaching ~200 Community Health Workers in Luweero District, with clinical leadership from Makerere University and compute from NVIDIA. The panel description says most LMICs are adopters, not builders. Accurate for the landscape. Incomplete as a prescription. 𝐓𝐮𝐞𝐬𝐝𝐚𝐲, 𝐂𝐄𝐒𝐓. 𝐋𝐢𝐯𝐞𝐬𝐭𝐫𝐞𝐚𝐦: lnkd.in/dYy4JKcv #AIforGood #AISafety #SovereignAI

🚀 Sovereign AI for the world. Cohere & Aleph Alpha form transatlantic AI powerhouse anchored in Canada & Germany! Combining our global scale with European R&D excellence to build sovereign, enterprise-grade AI. Security, privacy & trust for businesses & governments worldwide. #SovereignAI #AIPartnership Learn more: businesswire.com/news/home/2026… Image from left to right: Rolf Schumann, Schwarz Digits, Samuel Weinbach, Aleph Alpha, Aidan Gomez, Cohere, Minister Solomon, Canada, Minister Wildberger, Germany

Cohere Labs x ICLR 2026:Verification Limits Code LLM Training Binary verification (pass/fail) filters out too much potentially useful data. Better test coverage + softer scoring signals recover more usable data and lead to improved model performance. 💪

Incredible reading through ambassador applications. 🔥 Let’s go. First round of acceptances will go out next week Friday. Join us and shape the next era of intelligence.

Cohere Labs x ICLR 2026: SimMerge: Learning to Select Merge Operators from Similarity Signals In this work, we introduce SimMerge—a method for automatically choosing merge operators and merge order for LLM checkpoints using inexpensive similarity signals, removing the need for costly search or per-task tuning.✨

Our goal is to continuously learn from the world. This requires unlocking the 90% of data that is unstructured. Today we extend our adaptive data algorithms to work with unstructured data. ✨

It’s been an exciting week for Cohere Labs and @Cohere at ICLR 2026 in Rio de Janeiro, Brazil. 🇧🇷 In addition to main track papers, we’re also presenting new work in workshops. Check them out!

Full article: a16z.com/why-we-need-co…