Graphistry @Graphistry

Discord: discord.gg/QXaFt8wk8Q Training: louie.ai/trainings Newsletter: Get emailed monthly roundups via our newsletter at hub.graphistry.com GFQL demo: pygraphistry.readthedocs.io/en/latest/gfql… Louie.ai EAP signup - Only agent team to score an A+ on Splunk Boss of the SOC CTF Graphistry - free at graphistry.com/get-started

GFQL, Graphistry, and Louie.ai are moving fast, so we’re starting community updates plus a monthly newsletter roundup. This week: GFQL has a new kraken mascot, expanded Cypher support, a new query planner, typed graph support, and stronger CPU/GPU consistency. Help give him a name, link below! Louie.ai improved Kusto/Splunk query synthesis, added Kubernetes support, UI fixes, and new botsbench.com updates on Opus 4.7 and gpt-5.5. Graphistry shipped security and performance upgrades, streamlined Kubernetes multinode mode, and we are opening experimental access to conversational visual graph analytics. Links in comment below, including help us name the new GFQL mascot

louie.ai/trainings

After a sold-out Black Hat training and a highly successful first online cohort We’re excited to launch our new US timezone virtual training Led by the team behind an AI agent that autonomously solves every challenge in Splunk Boss of the SOC CTF, this program is built for practitioners who want to move from theory to real-world AI deployment in security operations. Participants will dive into hands-on scenarios across investigation, triage, and reporting, while learning how to design agentic systems that can plan, reason, and act with human oversight. The mission: help teams operationalize AI across SOC, IR, threat intel, vulnerability management, GRC, and SecOps. What’s inside: Hands-on labs, peer collaboration, and AI cyber range exercises Pre-configured environments (Splunk, Kusto, Databricks) End-to-end investigation and contextualization workflows Natural language + code-first approaches Deep dives into LLMs, embeddings, RAG, agents, MCP, semantic layers, evals, and trustworthy AI Vendor-neutral architectures and tooling perspectives Who should join: Security leaders Analysts Security & data engineers 📅June 23-24, 2026 🕘 US time zones 🌍 Live online 🎓 16 CPE + certificate 👥 45 seats 📼 Recordings + materials included Agenda and registration link in the first comment. #AISOC #AIAgents #RAG #GraphRAG

It's free to try, drop in a csv or throw a coding agent at it to get a different perspective on your data and see how it helps with trickier relationship & correlation questions We've run the backend at 1B+ edges / second (the new open source GFQL engine), and recommend starting frontend flows at 100K edges and working up to 5M edges. Both modes are actively improving, and we're happy to work with folks who need more.

Visualize massive graphs with GPU acceleration github.com/graphistry/pyg…

Meet the new face of GFQL: The mythical kraken! Its many arms represent how graphs help us reach across many things, and special to GFQL’s GPU approach, the power of the kraken is legendary. Our young mascot is a small addition, but one we hope makes the project feel a little more recognizable and fun as our community keeps growing. At the same time, we’re moving the GFQL community over to the Graphistry Discord. Slack was great for getting started, but to solve limitations such as in searchable chat histories and easy invitation links, Discord gives us a better place to share ideas, help each other, show projects, and talk graph compute in real time. 👉 Join the Graphistry Discord: discord.gg/QXaFt8wk8Q Inside: • GFQL discussions & support • Graphistry + Louie updates • Community demos & experiments • A place to help shape what comes next If you’re working with Cypher, GPU analytics, graph ML, cuGraph, Neo4j, Arrow, RAPIDS, or just exploring graph-shaped data in Python, come hang out. Happy graphing. #GFQL #Graphistry #GraphData #GraphComputing #OpenSource

Spoke at @graphtheplanet during RSAC week on building investigative security datasets with Claude Code + tools. Went from O(person-months) to O(person-days). Realistic threats, correlated data, emergent character interactions, and quantitative eval built in. Thanks @lmeyerov and @Graphistry for having me. gamma.app/docs/Bad-Actor…

If your model is 70% accurate per step across a 10-step investigation, your actual success rate isn’t 70%. It’s ~2.8%. Even pushing to 90–95% per step doesn’t fix it: → 90% per step = ~35% overall → 95% per step = ~60% overall That’s still not reliable enough for real-world security investigations. This is the gap between demo-grade AI and production-grade systems. Accuracy compounds. So does failure. Sign up for early access and receive updates with full videos → louie.ai Our upcoming AI agents for security training → louie.ai/trainings #ai #cybersecurity #aiagents #securityoperationscenter #threatdetection #graphanalytics

For teams building AI-driven investigations, speed isn’t a luxury, it's the difference between reactive and real-time. Leo Meyerovich founder at Graphistry shares what we’re seeing with the new GFQL Cypher : → ~4s on 2M edges (vs 22s on Neo4j) → ~7s on 30M edges (where others time out) → Up to ~50× faster performance → 1–2B edges/sec on a single GPU Click the link below to sign up for early access and receive updates with full videos → louie.ai Check out our upcoming AI agents for security training → louie.ai/trainings/ai-a… #AI #Cybersecurity #MachineLearning #DataScience #GraphAnalytics #BigData

After the most popular Black Hat training and a successful first online cohort We are happy to announce our new EU timezone virtual training Run by the team that built an AI agent that autonomously solves every challenge in Splunk’s Boss of the SOC CTF Participants will work through hands-on use cases across investigation, triage, and reporting, and learn how to build agentic systems that can plan, reason, and act with oversight. The goal is to help teams operationalize AI across SOC, IR, threat intel, vulnerability management, GRC, and SecOps. Inside the training: - Hands-on labs, peer sessions, and AI cyber range exercises - Pre-configured environments like Splunk, Kusto, and Databricks - End-to-end workflows for investigation and contextualization - Natural language and code-first approaches - Coverage of LLMs, embeddings, RAG, agents, MCP, semantic layers, evals, and trustworthy AI - Vendor-neutral perspective on tools and architectures Who should join: Security leaders Analysts Security and data engineers 📅 April 14–15, 2026 🕘 Europe / EMEA time zones 🌍 Live online 🎓 16 CPE + certificate 👥 45 seats 📼 Recordings + materials included Agenda and registration link in the first comment. #AISOC #AIAgents #RAG #GraphRAG

This year’s Graph the Planet kicked off RSAC week by bringing together leaders across AI × Security to explore agentic systems, graph intelligence, responsible AI, and investigation workflows, from enterprise-scale deployment experiences to cutting-edge model training research. Breakout sessions and hands-on demos sparked conversations that carried across the day and beyond. A huge thank you to our speakers: Justin McCarthy • Sarah Bird • @aminkarbasi • @daveherrald • @dawnsongtweets • @vicfcs • Shruti Datta Gupta • Chandrani Mukherjee • @joshua_saxe • Diana Damenova • David Cass • Josiah Lashley • Matthew Maisel. Your insights made the day unforgettable. We’re also proud of our team & sponsors (Graphistry/Louie.ai, AnChain, Bloomberg, DeepTempo)  for making this possible, coordinating sessions, live-streams, and everything behind the scenes to deliver a seamless experience. 🔴 Full recordings of all talks are available on our YouTube channel, and subscribe to catch every edited session we upload them throughout the week + follow-on interviews with this year’s speakers. Leo @lmeyerov 's keynote (AI That Investigates & GPU Cypher) link youtu.be/sXn-FeCYriA - The Graphistry Team #RSAC #RSAConference #CyberSecurity #InfoSec #AIinCybersecurity #CloudSecurity

keyboard_arrow_left Previous keyboard_arrow_right Next

GraphThePlanet is excited to announce another featured speaker for 2026: Justin McCarthy, Co-Founder & CTO of StrongDM Talk Topic: Software Factories Join executives, senior practitioners, researchers, and startup founders for discussions on AI, graph intelligence, and data-driven investigations. 🔴 Watch live on YouTube and LinkedIn Event Details: • Date: March 23, 2026 • Location: San Francisco, CA • Registration & More Info: graphtheplanet.com Happy graphing, — The Graphistry Team #CyberSecurity #AI #RSAC2026 #GraphThePlanet #DevOps #PlatformEngineering #SoftwareEngineering

GraphThePlanet is excited to announce another featured speaker for 2026: @joshua_saxe , Former AI Security Lead (Stealth, ex-Meta) Talk Topic: Evaluating and Hill Climbing Efficacy in Cyber Defense Agents Joshua will dive into how to measure and improve the performance of cyber defense agents, focusing on practical evaluation methods and iterative approaches to make systems more effective over time. Join executives, senior practitioners, researchers, and startup founders for discussions on AI, graph intelligence, and data-driven investigations. 🔴 Watch live on YouTube and LinkedIn during #RSAC2026 week Event Details: • Date: 23 March RSAC Week 2026 • Location: San Francisco, CA • Registration & More Info: graphtheplanet.com Happy graphing, — The Graphistry Team #AIforSecurity #AgenticAI #GenerativeAI #InfoSec #SecurityAI #RSAC2026 #GraphThePlanet

GraphThePlanet is excited to announce another featured speaker for 2026: DianaDamenova, Security Researcher at @Microsoft Talk Topic: Lifting Knowledge Graphs from Security Logs (Without ETL) Diana will show how to turn raw security logs into usable knowledge graphs without heavy ETL pipelines, enabling faster investigation workflows and more flexible analysis across large-scale data. Join executives, senior practitioners, researchers, and startup founders for discussions on AI, graph intelligence, and data-driven investigations. 🔴 Watch live on YouTube and LinkedIn during #RSAC2026 week Event Details: • Date: March 23, RSAC Week 2026 • Location: San Francisco, CA • Registration & More Info: graphtheplanet.com Happy graphing, — The Graphistry Team #GraphThePlanet #GTP2026 #RSAC2026 #KnowledgeGraphs #CyberSecurity #SecurityAnalytics #GraphIntelligence #DataEngineering #AIforSecurity

GraphThePlanet is excited to announce another featured speaker for 2026: David Cass, CISO at @keyrock 🗣 Talk Topic: AI Safety in Regulated Enterprise Deployments David will share how regulated organizations are approaching AI safety in production from risk management and controls to what it takes to deploy AI systems responsibly in high-stakes environments. Join executives, senior practitioners, researchers, and startup founders for discussions on AI, graph intelligence, and data-driven investigations. 🔴 Watch live on YouTube and LinkedIn during #RSAC2026 week Event Details: • Date: March 23, 2026 RSAC Week 2026 • Location: San Francisco, CA • Registration & More Info: graphtheplanet.com Happy graphing, — The Graphistry Team #GraphThePlanet #GTP2026 #RSAC2026 #AISafety #ResponsibleAI #CyberSecurity #EnterpriseAI #RegTech #GraphIntelligence

Splunk BOTS CTF is Dead The Louie.ai AI team has now solved Boss of the SOC, and we uncovered something important along the way - First, the Louie.ai team has built an AI agent that autonomously solves every challenge in Splunk’s Boss of the SOC CTF. For those new to our work, we initially went public around this time last year, sharing the first public autonomous AI agent to achieve a passing score. Our video on beating BOTS (link below) resonated widely because BOTS is one of the most realistic SOC/IR investigation CTFs, with 100+ data sources and 50+ questions covering core workflows like timelining. This year, we autonomously cleared all the challenges. Our latest stats: ~96% accuracy ~1 minute median per question Getting an A+ is a meaningful step forward. It doesn’t mean investigations are “solved,” but it does signal it’s time to raise the bar again. What drove our latest gains comes down to two themes we’ve been pushing on: ◼️ Advancing agent techniques around context engineering and tool use for data-heavy scenarios ◼️Shifting the mindset from an AI coder to a professional analyst Intuitively, AI agents investigate better when they act like professional analysts, not chatbots or AI coders. The second takeaway came from evaluating these systems: AI investigation benchmarks are starting to break As benchmarks circulate more widely, we’re seeing clear signs of benchmark contamination in frontier models on BOTS-style tasks. That makes it harder to measure real progress — and harder to meaningfully compare models and agents. To address this, we’re launching BotsBench: ◼️Vendor-neutral benchmarking for agent and model performance ◼️Standardized runs on operationally relevant challenges (starting with BOTS) ◼️Leaderboard across models and agents Open submissions: submit your agent — we’ll run it and publish results Our goal: make AI investigation systems measurable, comparable, and harder to game 🔴 If you’re building agents, submit yours and check the leaderboard: botsbench.com We’ll share more on how this works — and where it’s going — on Monday at our Security x AI summit (Graph the Planet 2026). Join online or grab one of the final in-person spots. #AIAgents #CyberSecurity #SecOps #AIinSecurity #ThreatDetection #LLM

GraphThePlanet is excited to announce another featured speaker for 2026: @dawnsongtweets , Professor at @UCBerkeley Talk Topic: Towards Building Safe & Secure Agentic AI Dawn will share perspectives from leading research on how to design and deploy agentic AI systems that are secure by default, with a focus on real-world risks, safeguards, and what it takes to make these systems reliable in practice. Join executives, senior practitioners, researchers, and startup founders for discussions on AI, graph intelligence, and data-driven investigations. 🔴 Sign up to watch live on YouTube and LinkedIn Event Details: • Date: March 23, 2026 RSAC Week 2026 • Location: San Francisco, CA • Registration & More Info: graphtheplanet.com Happy graphing, — The Graphistry Team #GraphThePlanet #GTP2026 #RSAC2026 #AgenticAI #SecureAI #AIResearch #CyberSecurity #TrustworthyAI #GraphIntelligence

Benchmark - 50x+ speedups with GFQL Cypher: pygraphistry.readthedocs.io/en/latest/gfql… GTP 2026: graphtheplanet.com

You can now run Cypher without a database - and 50× bigger & faster. Just directly query your dataframes and go. Last year, Graphistry announced GFQL, the first open source GPU property graph query engine. This year, we've brought it to Cypher. Think of GFQL as what Pandas, DuckDB, or Spark do for tabular analytics, dashboards, and ML - but now for graph-shaped data like networks, relationships, and paths. The new GFQL support brings massive performance, cost savings, and architectural freedom for the most widely used graph query language used by half the Fortune 500. GFQL Cypher is the latest member of the @ApacheArrow and @NVIDIAaidev @RAPIDSai OSS ecosystem. We will announce more at GTP 2026 (RSAC week in SF + livestream) and are happy to chat directly Notebook is linked below so you can try for yourself.