PointFive

Take us back to when companies banned the browser. Now it seems many are rolling back AI agent usage (3 in 4 companies) through a fear of productivity loss, privacy, and cost. Dave Anderson shares his (comical) thoughts on the stage of AI in 2026. What if there was a solution to all this? #thereis #tokenshift #tokeneconomy

Two years ago the FinOps conversation was "cloud waste." Today it's "AI waste," and it's compounding 10x faster, with far less visibility into what's actually driving the line item. Excited to partner with Ghost Technology to get this conversation in front of the executives who need to be having it

At KubeCon, our Cloud & AI Research Lead Ariel Fishman-Lichterman showed why S3 costs so often spiral past the $0.023/GB sticker price. When buckets get treated like filesystems, API call storms can multiply costs by 100x. Every LIST, HEAD, GET, and PUT is a billable event. The pattern shows up across the stack: frameworks that assume POSIX semantics, libraries that scatter unbatched writes, even the native Kubernetes S3 CSI driver that mounts buckets as local filesystems. Harmless-looking operations like metadata fetches, existence checks, and one-off writes become cost disasters at scale. His session traces them from billing reports through metrics down to execution inside systems like Spark and Airflow, then uses Cortex (the CNCF time-series database for Prometheus) as a blueprint for doing it right. Bucket indexes, shared caches, compactor workflows, and batched writes are all responses to the same underlying constraint: the object store charges for coordination. The takeaway Ariel keeps coming back to: stop treating the object store as a free-form database. Own the metadata, coordination, and routing in your application layer. A quick test for any data system you run: total S3 spend ÷ bytes stored. If it's 5x or 10x the raw storage rate, you're paying for coordination you could have tracked, batched, or cached. Watch the KubeCon session: https://lnkd.in/g6hQGvft

Our co-founder & CTO Gal Ben David joined the Amazon Web Services (AWS) for Startups AI Showcase ahead of AWS Summit NYC this week. Three takes worth sharing: Cloud cost isn't a crowded space. It's an unsolved one. Most tools aggregate spend and visualize it. Coming from cybersecurity, we treat it like detection: triage the problem, route it to the right owner, make it a team sport. That mindset is the product. Your biggest competitor isn't another vendor. It's your customer. Claude Code, Codex, and the rest make every buyer feel like they could build it themselves. SaaS has to become SaaS-plus-service, or it doesn't justify itself. Hot take on agents: they're only as good as how you instruct them. Like hiring sharp people, what matters is how they spend their time and tokens. The opportunity isn't shipping an agent. It's tuning AI for a specific role and helping customers measure the ROI. That last one is exactly why we built TokenShift: https://lnkd.in/g3Fq9qGv Thanks to AWS for Startups and to the other founders for a genuine, open conversation.

Our CEO Alon Arvatz sat down with Victor Garcia at FinOps X to talk about our $60M Series B, 6X revenue growth, and the thing he can't stop thinking about: AI spend. Teams are seeing 50% month-over-month growth in AI spend. Here's the thing nobody's saying out loud: The data about how your engineers burn tokens doesn't exist in your cloud bill. It lives on their laptops. Cursor, Claude Code, whatever they're running. You have no idea where the spend is going or what it's buying you. So PointFive shipped TokenShift. TokenShift sits on the endpoint. Compresses prompts behind the scenes. Shows platform teams exactly where tokens are going (debugging, building, architecting) at the developer level. We also built an AI Research Lab. Optimizing prompts isn't the same as rightsizing an EC2 instance. You're doing text analysis, preserving meaning, working around cache effects. New problems, new talent, new playbook. Alon's take: Cloud is still where the dollars are but AI is what's keeping everyone up at night. Watch the full convo: https://lnkd.in/gBu9heRT

That's a wrap on #FinOpsX 2026. The theme of the week was impossible to miss. This community is no longer just talking about cloud cost. It's talking about token economics, AI spend, and agentic FinOps. The Tokenomics Foundation was announced live on the keynote stage. One practitioner put it bluntly: traditional FinOps is dead. We agree and we already saw it coming. That's why we showed up to San Diego with the AI Efficiency OS, fresh off our $60M Series B led by Accel and the launch of TokenShift. PointFive is the only efficiency platform that manages AI spend everywhere it happens, from the cloud to the coding agent. But we also know it's not JUST AI, it's also everything that runs the cloud. The industry just named the shift. We had already built for it. Highlights from the week: 🎤 Our CEO Alon Arvatz on the keynote stage with Nubank Thomas Hammer, the world's largest digital bank outside Asia, on how their engineers turned efficiency into a habit. How automation, and small incrimentals can ad up to significant cost and performance efficiencies.  8–12% of total cloud spend saved. 🚢 A sunset cruise through San Diego Harbor with customers, prospects, and friends of PointFive. Best conversations of the week happened on the water. 🃏 Hundreds of demos at the booth, and a few Cloud Blackjack decks that found good homes. Overhearing people say "have you seen what PointFive is doing ... made us feel a bit special". Thank you to the FinOps Foundation and everyone who stopped by.

What turns a FinOps program into a boardroom conversation? That is the question Alon Arvatz (CEO, PointFive) and Thomas Hammer (Nubank) took on at #FinOpsX in San Diego. Nubank runs one of the world's largest digital banks: 131M customers, 10,000+ engineers on AWS, 45+ business units, each with its own Cost Champion. The FinOps practice is mature. And it earns leadership attention for one reason. It treats price and usage as two different problems, with two different owners. One number from the session says the most: unattached EBS resolution cut from 210 days to 16. Here is why that matters more than it looks. A finding you have spotted but not fixed is not a saving. It is a bill that keeps running. Every day an opportunity sits open, the waste keeps billing. Speed to resolution is speed to value, and going from 210 days to 16 means roughly seven months of cost that simply stops. The rest of the impact, all verified against the actual bill: • 8–12% of total cloud spend saved • 3,000+ DynamoDB tables optimized automatically • 92% of internal teams reached The Q&A was on fire. A lot of the questions centered on automation, remediation, and anomaly detection in highly regulated environments, which tells you where this audience's real pressure is right now. The throughline: estimates get discounted, verified savings do not, and savings you act on fast are worth more than savings you sit on. More takeaways from the session coming to this feed this week. Thank you Thomas Hammer and Nubank for your incredible partnership. #FinOpsX #FinOps #CloudEfficiency

Yesterday was huge. 1) Wall Street Journal, Catech, Forbes 2) NYSE live stream 3) And our team got recognised on trains on the way to work. At #FinOpsX our booth was run off its feet. The AI Efficiency OS and TokenShift hit like a cold drink post marathon. Thank you to our entire team, our customers, investors, partners and the queue of people signing up for our new offerings. Tonight we cruise. 🚢

Three things today. One. PointFive raised $60M in Series B funding, led by Accel, with participation from Entrée Capital and Salesforce Ventures. Two. We're launching the AI Efficiency OS. The category called "cloud cost optimization" was built for a world that's already gone. AI spend is growing faster than cloud ever did, and most FinOps tools can't see it. PointFive is the only efficiency platform that manages cloud and AI spend in one place, and puts you in control to build your own workflows, agents and apps. Three. We're launching TokenShift, the developer endpoint of the platform. Coding agents are the fastest-growing line item in engineering, and most companies have zero governance over what Claude Code, Cursor, and Copilot are spending. TokenShift fixes that in real time, without degrading model performance. And we will be live at #FinOpsX in San Diego this week. Thanks to Philippe Botteri and the Accel team, to Entrée and Salesforce Ventures, and to our founders Alon Arvatz, Amir Hozez, Gal Ben David, and every person at PointFive. So what are we optimizing today? #AIEfficiencyOS #borntoday #TokenShift #FinOps #FinOpsX

Next week will be the biggest week in our history. It's all hands on deck. There will be announcements on the future of this industry, there will be live streams, there will be knowledge, networking, and partying. Let's make some memories. Who's excited? Who's getting on board? #finOpsX #AIEfficiencyOS #FinOpsReimagined #Whatareyouoptimizingnext