AI Systems Built & In Use

A $200M+ portfolio across 5 partner segments generates hundreds of daily signals. Critical items get buried without a forcing function, and a single leader can't manually process the volume with the consistency and timeliness the role needs.

Built an orchestration layer of 17 agents across 5 functional teams — Pipeline & Revenue, 2x2, OpenAI Partnership, Executive Prep, Team & Comms — running on a 06:30 AM morning brief and 05:00 PM EOD wrap cadence. I sit at the decision layer only: review, approve, direct. Everything below that is standardized and audited.

~8–10 hours/week recaptured (inbox triage, pipeline scanning, market monitoring, meeting prep, coaching nudges). Zero missed signals, standardized outputs, full audit trail. Zero net-new headcount required to operate the book.

Bi-weekly 2x2 reporting was consuming 4–6 hours/week across 10+ PAMs and 3 management tiers. The output was inconsistent, late, and impossible to roll up cleanly for VP-level visibility.

Built a 3-tier pipeline: PAM Scout collects at the IC level, LT Aggregator rolls up leadership team, Master Aggregator produces the VP-ready view. Wired it into the same Chief of Staff cadence so it runs itself.

Adopted independently by two peer managers on their teams. The mechanism replaced a manual process that had been inconsistent and late. A personal tool became org infrastructure.

Managing OpenAI partnership launch, VP reporting, cross-org approvals, and pipeline hygiene simultaneously would have required a team of program managers I didn't have.

Shipped 10 discrete mechanisms in six months: WWPS ISV Pipeline Intelligence, Pipeline Hygiene Agent, OAI Delegation-of-Work Review workflow with SharePoint tracking, SAP NS2 Monthly Partner Snapshot for VP leadership's dashboard, OpenAI executive leadership 1:1 Briefing Automation, MBR consolidation (4 calls → 1), Weekly OpenAI Activity Summary (VP leadership requested expansion to two versions on quality alone), NS2 First GTM Funding Mechanism, SaaS Co-Sell Benefit Resolution, and the 2x2 automation above.

13+ GovCloud access requests processed on cadence. a peer IC manager: 12 red flags across 39 opportunities actioned in a single week from the Pipeline Hygiene Agent. SaaS Co-Sell Benefit Resolution resolved a co-sell benefit gap affecting seller compensation at scale — a systemic blocker removed by identifying it in the agent output.

Two direct reports had domain expertise that was bottlenecked on manual work: a team solutions architect covering Esri geospatial across 20 SAs, and a team member building NS2's first GenAI workload with no compliance scanner that fit federal constraints.

Coached and unblocked both builds. The team solutions architect built the ArcGIS SA Agent inside the same operating patterns as the Chief of Staff system. The team member shipped Efterlev as an open-source FedRAMP scanner alongside 'AI Core-Bedrock' — the first-ever GenAI production workload at SAP NS2.

ArcGIS SA Agent saves 400+ hours/month across 20 Solutions Architects. Efterlev shipped open source. NS2 GenAI revenue: +3,472% YoY on the strength of that first production workload. The pattern travels — the team ships operator-grade AI, not decks.

Outside AWS: Budco estimators needed help turning inbound RFQs into consistent, defensible line-item quotes without adding headcount or introducing autonomy on pricing.

Claude Sonnet as the reasoning engine, Supabase (Postgres + Auth + RLS) for state and audit trail, TypeScript app front-end. Every line item ships with a confidence score and a mandatory human review step — no autonomous decisions on money.

In active use by the team. Contributed to Budco DSO improvement to 35–40 days. Same pattern as the AWS work: practical, boring, and it earns its keep every week.