Deployed with the world's largest private company.
Copilot is rolled out broadly. Claude drafts the board deck. The product itself runs ML against industrial OT signals. But finance and accounting AI usage is "basically nothing," and IT is focused on foundational infrastructure with zero tooling for an AI-native organization. Phyvant closes the gap on the systems your team already uses.
A ~$1B acquisition and a new reporting cadence to a Japanese public-company parent. The integration work lands on George and Sai. Phyvant documents the close, the reporting pack, and the cross-entity flows once, then runs them every cycle from that documentation.
You are on NetSuite but "not to full functionality." You are hiring a Financial Systems Manager to fix it and run the Advanced Revenue Module. Phyvant sits on top of NetSuite and uses what is there today, and the automations improve as the Systems Manager turns more of it on.
Copilot agents Sai built in co-work. Claude prompts in board-deck drafting. Ironclad AI on contracts. AP AI. None of it shares a maintenance, version-control, or audit story. Phyvant governs all of it on one layer: role-based permissions, audit trail, confidence weighted by user expertise, outputs sourced from your data instead of generated.
On the May 6 demo we walked the close together: Excel manipulation, AP and payroll downloads, NetSuite uploads, the reclasses and judgment calls that today live in one person's head. Phyvant's Chrome extension watches it once, documents every step, and runs the same procedure every cycle. Approvals where George wants them, automatic where he doesn't. Every step is sourced, versioned, and audited.
The May close ran on the documented procedure. Three steps required human approval, two judgment calls were flagged, and the procedure updated itself accordingly.
Second use case George surfaced on the demo: the manual work in sales operations. Deal-desk approvals, ARR bookings categorized against the Advanced Revenue Module, quota retiring, commissions reconciliation. Phyvant watches the rep-to-revenue flow, documents the procedure, and surfaces deals where the bookings shape does not match what actually posted to the ARM.
Live ARR rolled up against the Advanced Revenue Module. Phyvant reconciles deal-desk decisions to ARM postings every week, not at quarter-end.
Today: Copilot agents Sai built in co-work for the weekly summary, Claude prompts in finance for board decks, Ironclad AI on contracts, AP AI. None of it shares a maintenance, version-control, or audit story. That gap was named explicitly on the call. Phyvant governs all of it on one layer: every agent versioned, every output sourced, every action audited, role-based permissions enforced, confidence weighted by who corrected what.
Every agent your team uses, registered, versioned, and audited on one surface. Confidence weighted by user expertise. George's correction counts more than a new hire's.
AP, payroll, Excel reclass, intercompany, Mitsubishi pack, all driven from one documented procedure every cycle.
Every deal's bookings shape reconciled against the ARM posting weekly, not at quarter-end.
A customer in NetSuite, the CRM, Ironclad, and Slack, resolved to one record. Contradictions handled.
Every Copilot, Claude, and Ironclad agent registered, versioned, sourced, and auditable.