Our 2025 Year in Review is live 🚀 ✅ 44 new investments ✅ 30 follow-on investments ✅ 20 M&A exits ✅ $1B AI Fund fully deployed ✅ New focus areas in robotics, defense, & energy Explore our full Year in Review: salesforceventures.com/sfv-2…

What can studying an AI 'mind' teach us about our own human brains? "Once we can understand an AI mind, we hope that this teaches about our own minds, our own consciousness, our own understanding of reality and the world," says Eric Ho, CEO of @GoodfireAI. Eric describes interpretability research as "neuroscience" for AI. If we can understand the mind of an AI model, we may learn how to design safer systems while also learning about how our own minds function. 🔗 Full interview with Eric and Emily Zhao here: youtube.com/watch?v=_b1Hs2Q7…

"We were looking for people who really get and see the journey that we're on. I think it's really important that our investors deeply share our vision of the future so that it feels like one team rowing in the same direction." Eric Ho, CEO of @GoodfireAI, explains why he partnered with @emilyyzhao and the Salesforce Ventures team 🤝

Salesforce Ventures is proud to lead the Series B of @algo_artis 🎉 Some of the world's most consequential operational decisions — at power plants, petrochemical facilities, logistics networks — are still being made manually, with spreadsheets and legacy tools. ALGO ARTIS is changing that. Their AI-driven planning platform, OPTIUM, handles the combinatorial complexity that overwhelms conventional tools, while keeping human operators in the driver's seat. ALGO ARTIS is already trusted by major Japanese enterprises like Kansai Electric Power, Cosmo Oil, and Nippon Paper — and built to scale globally. We're excited to partner with founder Kentaro Nagata and the ALGO ARTIS team. We wrote more about our investment on our blog: salesforceventures.com/persp…

If we can’t read, edit, or debug AI models, we can’t shape them safely enough for the systems they’re becoming a part of. As @GoodfireAI CEO Eric Ho explains, AI models today are vastly different from traditional software. We can write software, inspect it, change it, and debug it. With LLMs, much of the behavior is hidden inside the weights of a neural network. Goodfire is building the science and technology to change that. Their work is focused on understanding the foundations of neural networks so models can be intentionally designed, not just trained and tested from the outside. That’s important for safety, but also for progress. As models become more capable, they have the potential to develop insights that humans cannot easily access on their own. The opportunity is to look inside those systems, understand what they have learned, and translate that knowledge back into something humanity can use.

Salesforce Ventures is proud to invest in @supabase: the new backend for the AI era 🚀 Building a modern backend has always been more painful than it should be: stitching together a database from one vendor, auth from another, storage from a third, and glue code to hold it all together. Supabase eliminates that complexity. One open-source platform. Complete Postgres backend with auth, storage, realtime, and edge functions — wired together from day one. The numbers speak for themselves: 7 million developers have signed up, 55% of the latest @ycombinator batch builds on Supabase as well as leading enterprises like Figma, Vercel, and Bolt. Further, AI coding tools like Claude Code, Cursor, and Lovable are making Supabase one of the most common backends in a generation of new applications and creating a compounding distribution advantage that goes well beyond managed Postgres. Proud to back @kiwicopple, @AntWilson, and the entire Supabase team as they build the foundational infrastructure layer for how software gets built in the AI era 🎉 🔗 On our blog, @pdrews, @emilyyzhao , @KartikeyaG9, and Pascha Hao shared more about our investment in Supabase: salesforceventures.com/persp…

We've raised $500M at a $10B valuation supabase.com/blog/supabase-s…

"I would like to live in a world where we actually deeply understand the AI models that we're deploying everywhere across society." @GoodfireAI CEO Eric Ho on how even researchers at frontier labs can't fully explain how their models work, or why they become intelligent. "Never before has there been this wide of a gap between our understanding of a technology and how widely it's being deployed. Our entire mission as a company is to change that." More than half of the Goodfire team is focused on interpretability research —reverse engineering the weights and parameters of neural networks to better understand their algorithms, computations, and features. Goodfire's aim is to look "inside" a model and attach meaning to the units of computation within it. This way, they can help design better models, build safer systems, and create more effective guardrails during training and deployment. 🎥 Watch Eric in conversation with @emilyyzhao on how Goodfire works to understand AI models.

The idea that launched @GoodfireAI🔥 When ChatGPT launched, most people were blinded by the possibilities. Eric Ho saw the risk it posed. "I kind of saw the next few years unfold before me, where we were about to get increasingly powerful models ... massive amounts of compute, massive amounts of intelligence, but we wouldn't understand at all how this intelligence would actually work." Most models operate as a black box. Users can see the output, but they can't reliably see how the model reached it, why it behaved in a certain way, or whether it will behave the same way again next time. With how quickly AI is being deployed into mission-critical environments, Eric wanted to do something to ensure AI models are functioning as intended. In a new video interview with Emily Zhao, Eric explained his aim behind founding Goodfire: to build the science and technology needed to understand AI from the inside out.

We're proud to invest in @genx_corp 🤝 Japan's enterprise AI market is enormous — and untapped. GenerativeX deploys custom AI agents for Japan's largest enterprises in days, not months. Fast AI adoption at scale is possible. GenerativeX is proving it. salesforceventures.com/persp…

8️⃣ SFV portcos were just named to this year's @CNBC #Disruptor50 list: @AnthropicAI, @databricks, @MistralAI, @NotionHQ, @WeAreLegora, @Lovable, @glean, and @runwayml 🚀 Congrats to every founder and team recognized — we're proud to be in your corner. cnbc.com/2026/05/19/2026-cnb…

We've invested in @OpturaAI 🚀 Healthcare AI has a measurement problem. Orgs are running pilots — but can't prove hard-dollar ROI to their boards. Optura fixes that. A full-stack AI OS for healthcare enterprises. Proud to partner with this team 🤝 salesforceventures.com/persp…

"Peter Morales was put on this earth to solve this problem." After leading @Code_Metal_AI's Series B, @RobwKeith talks to @ryanaytay about why Salesforce Ventures invested in the business 🎥

"I think it's a great network to have. The Salesforce ecosystem is very big, and so whether it be execs or people working within the company, or the companies within the Ventures portfolio ... it's a massive network." – @ryanaytay of @Code_Metal_AI on why he partnered with SFV

Solving the AI energy crisis requires more than new generation. It also requires infrastructure that ensures energy gets where it needs to go. We just published Part 2 of our energy series, focused on modernizing the U.S. power grid. Read it here: salesforceventures.com/persp…

"There's a lot happening in the world of AI and code generation ... but there's not enough focus on what I'll call AI verification." @ryanaytay, COO of @Code_Metal_AI, on the company's mission and why he decided to join.

New on the blog today: What the Data Says About AI Pricing 📈 @Spyzguyz dug into our recent AI Pricing Report and shared some of his favorite takeaways. 🔗 Read it here: salesforceventures.com/persp…

Congrats to the 4️⃣ Salesforce Ventures portfolio companies named to @TIME's 100 Most Influential Companies of 2026 list: @AnthropicAI, @EmeraldAi_, @Lovable, and @MistralAI 🚀 Well deserved. time.com/collection/time100-…