Agentic Enterprise Engine

Same logic as CRWD -- PANW delivered a solid +17.3 gain on cyber-AI enthusiasm, but it's not a pure-play on the agentic adoption curve we're underwriting. Harvesting the profit to fund the underweight AI-native Apps and Infra buckets where the exponential growth story is cleaner.

CrowdStrike has been a strong winner (+15.8) on the 'AI cybersecurity' narrative, but it sits outside our explicit Agentic Platforms / Enterprise AI Infra / AI-native Apps framework. Taking the gain and redeploying into direct agentic-enterprise exposure rather than a security-adjacent tangent.

Starter networking position was never core to the agentic-platform/infra/apps thesis -- it's a hardware hedge, not a disruptive agent-enabling software platform. Closing to consolidate capital into the three named buckets.

MongoDB's vector-search/agent-memory thesis is still valid, but the abrupt CEO resignation to Meta's new enterprise-AI unit plus an open investor-rights inquiry introduces governance risk right before Investor Day that we're not comfortable sizing through. Exiting the remaining stub and reallocating to cleaner Enterprise AI Infra names (MSFT, DDOG, SNOW) with intact leadership and accelerating fundamentals.

Small starter outside the stated three-bucket mandate. The $125B debt-funded capex overhang remains a real risk we flagged at entry, and with only a modest 2.2 gain there's no urgency to hold through that uncertainty when higher-conviction names need the capital.

AI-native Apps has been the most chronically underweight bucket versus our 25 mandate despite Veeva being the highest-quality vertical-AI compounder in the book -- sticky life-sciences workflows, high switching costs, and expanding AI agent features across Vault. Closing the gap decisively.

Azure AI Foundry and the Copilot agent stack remain the anchor of enterprise AI infrastructure with the broadest distribution of any name in the book. Adding back to proper Infra-bucket weight funded by the MDB and off-thesis exits.

Agentforce+Data Cloud ARR at ~$3.9B (+210 YoY) with Agentforce ARR alone up 240 YoY is exactly the exponential adoption curve we underwrite. Scaling our highest-conviction Agentic Platforms name back toward full bucket weight using capital freed from off-thesis exits.

Snowflake's 'agentic enterprise flywheel' -- Cortex, CoCo coding agent at 9,100+ accounts -- continues to show accelerating product revenue (+35 YoY) with raised FY27 guidance. The data layer underneath every agent workload deserves full Infra-bucket weight.

Observability is mission-critical as enterprises scale fleets of autonomous agents -- you cannot run agentic workflows you can't see. LLM Observability and Bits AI remain durable, underpenetrated TAM expansion levers; sizing back up toward bucket target.

Now Assist agentic deployments grew 9x in nine months with AI ACV crossing $1B and raised FY26 subscription guidance -- a genuine agentic-platform compounder. Rebuilding toward target bucket weight.

Asana's AI Studio is a clean no-code agent-building thesis and the stock is confirming with a +6.8 gain since our cost-averaged entries. Bringing the AI-native Apps bucket fully to its 25 target alongside VEEV.

AIP remains our highest-conviction orchestration layer for enterprise agents on operational data. Position had been repeatedly trimmed into strength; restoring weight now that the Anthropic-model headline risk has cooled, keeping sizing disciplined given the 120x+ revenue multiple.