1Password

Adarsh H., Co-Founder and Co-CEO of Mercor, joined Zero-Shot Learning to share why we need to rethink how agentic performance is measured and the infrastructure we can build to get there. Adarsh shared, "You could have a model that just answers correctly the first time, but it's making all the wrong decisions along the way. Then when you adapt it to a slightly different context, all of a sudden you've got an agent that's totally broken in production." While most teams grade an agent on whether it delivers the right result, the best teams grade an agent's task, trajectory, and output because that's what breaks when the context shifts. Zero-Shot Learning Episode 5 with Adarsh, Nancy Wang, CTO at 1Password, and Dev Tagare, Head of Engineering for Google Gemini, unpacks what it actually takes to move an agent from passing a benchmark to production. 🔹Why evals are infrastructure that has to evolve with every model and prompt change 🔹Why grading the output isn't enough, and how trajectory-based evals catch failures before they hit production 🔹The staged rollout model for deploying agents in production that limits the scope of failure at every step 🔹Why authentication for agents may already be a bigger market than authentication for humans Lock in to the full episode: https://bit.ly/4pnJASo #ZeroShotLearning #AgenticAI #1Password #podcast

1Password's product security team has designed a model-agnostic security review pipeline to evolve alongside the latest frontier models. We built our Security Analysis Guidance Engine (SAGE) to ensure 1Password products remain secure as AI-powered offensive techniques become more prominent. SAGE runs several models from different providers across its Finder, Critic, and Judge pipeline: the Finder surfaces possible issues, the Critic tries to knock them down, the Judge makes the call. Learn how we built it: https://lnkd.in/eYAzVkJf

1Password moves into AI cost management, betting that token spend is the next enterprise budget crisis https://bit.ly/3SZTYUm

Today we're introducing AI Spend and Consumption Management, a new capability within 1Password SaaS Manager that gives organizations a single, normalized view of AI consumption and spend across Anthropic, Cursor, and OpenAI. AI costs are rising fast and most teams don’t have a clear way to track or manage them. By bringing AI token consumption data into one place, IT, finance teams, and business leaders can understand what's driving costs, stay ahead of unexpected overages, and make more informed investment decisions by: ✅ Seeing AI token consumption in one place across supported vendors ✅ Staying ahead of budget risk with spend thresholds and automated alerts ✅ Understanding what's driving costs across teams, users, vendors, and models ✅ Optimizing AI investments within their broader software portfolio As our CFO Greg Henry put it: “Executives want teams to build faster with AI, but that speed is creating a new kind of spend pressure.” Learn more: https://bit.ly/4ffPdO0 #AISpend #SaaSManagement #FinOps #AIGovernance #1Password #tokenmaxxing

Last week at RAISE Summit, the 1Password team joined conversations on what it takes to build, deploy, and secure AI in the enterprise. Here's where we took the stage: ➡️ Aashish Tripathi, VP, Product Strategy, joined Snyk's AI Security Summit to discuss securing AI agents in production with leaders from Docker, Inc, JetBrains, and Snowflake. As agents become autonomous actors that authenticate, invoke systems, and even delegate work to other agents, identity systems must operate at AI speed and scale. He discussed why credentials should never be exposed directly to models, and why authorization, continuous monitoring, and auditability are becoming foundational controls for agentic systems. ➡️ Richard Kirby, Sr Director, Product, joined leaders from Airbus, Schneider Electric, Capgemini, and Third Bridge Group to discuss how organizations should measure AI ROI. His perspective is that meaningful ROI goes well beyond token costs. It comes from improving engineering throughput and software quality while accounting for risk, using metrics like DORA and integrating AI into existing engineering workflows rather than creating new ones. Different stages. Different conversations. One common thread. As AI moves from experimentation to production, organizations need to solve two challenges simultaneously: governing how AI systems access data and infrastructure, and ensuring AI delivers measurable business value. Security and ROI aren't separate conversations, they're deeply connected. That's the future we're building toward at 1Password. #RAISESummit #EnterpriseAI #AIROI #DevOps #1Password

Can your organization clearly say how much it’s spending on AI? AI token usage is creating costs that don't fit neatly into existing budgets or reporting. Finance teams are reconciling unexpected spend. IT teams are trying to understand who is using what. Governance is often spread across multiple tools. Join 1Password CFO Greg Henry and Global Advisory CISO Dave Lewis on July 21 for a candid conversation about how organizations can gain visibility into AI usage, strengthen governance, and get ahead of AI spend before it becomes a bigger problem. You'll learn: 🔹 Why AI token usage has become an unexpected challenge for Finance and IT 🔹 What every CFO and IT leader should be able to answer about AI consumption today 🔹 How leading organizations are proactively managing AI usage and spend 🗓️ July 21 at 11:30 AM PT / 2:30 PM ET Register here: https://bit.ly/4vszXTQ Can't make it live? Register anyway and we'll send you the recording. #AI #AISecurity #SaaSManagement #Finance #ITLeadership #1Password

AI is changing who writes software. Designers are prototyping internal tools. Analysts are connecting data pipelines. Developers are shipping faster with AI coding assistants. But while AI accelerates development, it also makes it easier for secrets to end up in .env files, scripts, chat threads, and code that was only meant to be temporary. Those shortcuts add up. Every plaintext credential increases the risk of secret sprawl and creates another opportunity for attackers. 1Password Developer Tools help teams keep secrets out of code and access them securely at runtime, so developers can move quickly without compromising security. With 1Password, you can: 🔹 Store and use SSH keys without keeping them on disk 🔹 Keep secrets out of code and .env files with Environments and Secret References 🔹 Access credentials securely at runtime through the CLI and SDKs, including from AI-assisted workflows 🔹 Use service accounts for automation instead of sharing personal credentials 🔹 Inject secrets into CI/CD pipelines without exposing them Learn more: https://bit.ly/4wBODkI #DeveloperSecurity #SecretsManagement #DevTools #AgenticAI #1Password

The first documented AI-run ransomware attack didn't start with AI. It started with stolen credentials. This is the new threat model, it's already here, and it's worth thinking about what we can do about it: ▶️ Standing credentials are the attack surface. If secrets aren't stored somewhere persistent, the initial compromise has a lot less to work with. ▶️ Just-in-time access limits blast radius. Scoped and time-bound access means more friction at every step. ▶️ Zero-knowledge credential handling means credentials never transit a system that can be swept. What isn't there can't be stolen. ▶️ Continuous monitoring matters at AI speed. This agent fixed a failed login in 31 seconds. ▶️ Audit logs are essential for understanding scope and containing damage. The techniques in this attack were fairly ordinary. What was different was the speed, and that changes the conversation about where controls need to live.

A prompt change doesn't touch a line of code. It can ship straight from a dashboard with no record that a security reviewer would ever see. And it can change how an AI agent interprets a task, which tools it calls, and how it uses the credentials it already has. That's the gap Ankur Goyal (Founder and CEO, at Braintrust) came to talk about on Zero-Shot Learning episode 4, and he wasn't subtle about it: "I am, almost to a scary amount, frequently surprised by what the ramifications of random prompt changes are." He joined @Nancy Wang (CTO at 1Password) and Dev Tagare (Head of Engineering, Google Gemini) to dig into what a real release gate for prompt changes actually looks like. The framing that stuck: "Instead of writing a PRD, you can write an LLM-as-a-judge with really clear criteria about what matters to you in the business. Once you do that, you have a quantifiable way of enforcing whatever guidelines make sense for you." Evals as the new PRD. That's the shift. Read the full breakdown: https://lnkd.in/e76RgC4c #ZeroShotLearning #AI #AIAgents #AIObservability #1Password

🏆 Congratulations to Larissa Crandall, Global Vice President of Channel at 1Password, on being named to the 2026 CRN Inclusive Channel Leader list! Presented by CRN and The Channel Company, this recognition celebrates channel leaders who are building cultures where everyone feels valued, respected, and empowered to reach their full potential. As Larissa shared: "Leading a global organization has reinforced how critical diverse perspectives are to better decision-making and adaptability. I've seen the barriers people face up close, which has driven me to be intentional about creating opportunities, mentoring others, and helping people grow throughout their careers." We're proud to see Larissa recognized for the leadership she brings to our partners, our team, and the broader channel community. See the full 2026 CRN Inclusive Channel Leader list here: https://lnkd.in/exa7ZsuN #CRN2026 #InclusiveChannelLeader #ChannelFirst #1Password #Inclusion #Leadership