sample-note.md

Brooks | Qpoint Introduction+ Demo

Meeting UUID: 2bbbfa5d-029f-4d18-9376-938418a56213

Participants

19:33 Qpoint’s eBPF-based sensor provides precise pre-encryption visibility into application traffic and supports customizable metadata tagging via environment variables or Kubernetes labels.
21:27 Agent benchmarks indicate under 3% CPU and a few hundred MB of RAM per node, with detailed performance metrics to be shared for larger deployments.
23:07 Selective payload capture is handled through “stacks” routing rules, enabling per-endpoint privacy controls such as header-only logging or full body capture.
25:59 A follow-up demo focused on agentic AI workflows will be scheduled in 2–3 weeks, and a shared Slack channel will be set up for ongoing collaboration.
26:46 Standard pricing is quoted at $360 per node per year, with flexibility for large-scale environments through negotiation.

2:17 Previous discussions with Brooks, Constantine, and others suggested broader applicability of Qpoint beyond initial contacts
2:48 Dan Fuhry initiated contact, highlighting Qpoint’s capabilities
3:15 Brooks leads the infrastructure cloud security team, focusing on AI security, LLM firewalls, network security, proxies, outbound access control, and AI product security infrastructure

0:37 Manual Cloudflare SSL certificate updates occur yearly; moving to monthly rotation would be onerous
5:55 Teams maintain outbound proxies despite significant computational and operational overhead, often leading to abandonment

4:15 Need visibility into pre-encryption data flows to identify where customer data is shared with vendors and AI tools
7:31 Require instant detection of TLS hygiene issues and compliance proof with live evidence rather than documentation

4:15 Customers often discover more vendor endpoints than expected, expanding visibility
5:10 Real-time detection of third-party rate limits and failures reduces troubleshooting time from hours to minutes

5:55 EBPF-based sensor observes network activity at the kernel level before encryption without code changes or TLS termination
20:34 Metadata tagging is customizable from environment variables, container labels, Kubernetes labels, or annotations
23:07 Routing “stacks” allow per-URL rules to include or ignore payload capture and scanning workflows

27:06 Standard pricing model set at $360 per node per year, adjustable for larger deployments

21:00 Unscoped eBPF capture could impact node performance if configuration is not properly limited
22:34 Privacy concerns require customizable payload capture to avoid logging sensitive information in central services

8:15 Apollo.io praised Qpoint as the first technology showing what’s happening and how to fix issues rather than just flagging them
9:05 Brooks described the solution as very helpful for visibility into agentic lab settings

0:37 Manual monthly certificate updates are burdensome and need automation
9:05 Agentic AI workflows go unmonitored as autonomous systems communicate without oversight
19:33 No service identity attribution yet, delaying mTLS proxy migration and source tracking
22:34 Privacy concerns demand granular control over which traffic data fields get captured and stored

Devin Bernosky
- 21:27 Send resource usage figures for large deployments to Brooks McMillin and the Dropbox team
- 25:59 Schedule a follow-up demo in two to three weeks with the Office of the CTO on agentic workflows for Brooks McMillin
Dan Fuhry
- 28:03 Determine internal process to set up a shared Slack channel with Qpoint
Andrew Valencia
- 27:36 Follow up in two to three weeks with Brooks McMillin regarding agentic workflow demo readiness
- 28:13 Coordinate creation of a shared Slack channel with the Dropbox team if they request it
Brooks McMillin
- 27:21 Explore deployment scenarios, budgeting, and customization needs for the Qpoint solution and report findings

3:15 Infrastructure cloud security team builds AI security measures including LLM firewalls, network security, outbound proxies, and AI product security infrastructure
9:05 Need visibility into autonomous ‘agentic lab’ environments where AI-driven machines interact without monitoring

3:41 Plan to discuss agentic workflows integrated outside of eBPF in a few weeks based on partner demand
8:15 Targeting delivery of AI security and observability features within the next month
25:59 Office of the CTO is preparing an agentic workflows demo (Cloud Code/Openclaw focus) to share in the next couple of weeks

20:27 Ability to tag connections via custom metadata sources such as environment variables, container labels, Kubernetes labels, or annotations for service account identification
24:19 Define per-URL routing stacks to customize data capture—e.g., ignore bodies for certain endpoints or enable sensitive-data scanning for LLM URLs