I build the operating systems a go-to-market org runs on.

Monthly renewal reviews ran on a static spreadsheet. Meetings were spent clicking between systems, and decisions had nowhere to live between reviews.

A Claude artifact that puts the CSM's notes, the renewal rep's notes, value at risk, forecast, and health on one card per renewal. Managers send Slack pre-reads and log decisions to a shared record.

CSM assignments ran in spreadsheets and were executed one account at a time. Nothing checked whether the receiving CSM had room, whether the account fit their segment, or whether a customer family was being split.

A workbench for every kind of assignment: handoffs straight from Implementation, leave coverage, new-hire books, departures, and rebalances. It ranks candidates on capacity (account count and ARR as separate limits), sales-pod fit, complexity, and region. The manager approves and the move ships as one batch.

KPIs lived in aging dashboards that disagreed with each other. Each person's actual work (renewals, expansion, problem accounts) was spread across systems with no shared sense of what came first.

• Prioritized work: every CSM, manager, and director sees their renewals, cross-sell and upsell opportunities, and problem accounts, ranked by a custom prioritization model.
• KPIs and trends: one dated snapshot, colored against target, that clicks through to the accounts behind each number. KPI design draws on the renewal-driver analysis (F.06).
• Certified AI tools, nested: the renewal and account-assignment workbenches and other certified tools open from inside it, so there is one front door.

The prior year's carve data arrived late and hurt the start of the year. Books were balanced by hand, with no shared rules and no way to compare scenarios.

Each territory carries its agreed rules: caps, disruption limits, renewal locks, and sales-pod alignment. Auto-balance moves whole customer families into the best-fit book, and nothing saves until a person confirms. It surfaced two conflicts before sign-off that would otherwise have landed late.

Getting a trusted number, or turning a request into a correct Gainsight rule, meant pulling in Ops. Ops became the bottleneck, and people filled the gap with numbers that didn't match.

• Reporting skill: CSMs and ICs pull governed metrics on demand, in a structured format.
• Gainsight rules skill: turns an unstructured CTA request or a BRD into a step-by-step build guide for complex rules.
• Briefing generator: builds a briefing agent per stakeholder that reads five systems into a two-minute morning summary.

The team tracked about eight activities it assumed drove renewal, and its renewal metric didn't match Finance's certified rate.

One workflow that pulled two years of renewals from four systems, ran a size-controlled logistic regression, and produced a director-ready report.