SES AI Corporation: Financial Close Flux
Analysis with Claude


Summary
SES AI Corporation, a battery technology manufacturer listed on the NYSE, closes its books every month and every quarter across a US (LLC) entity and a Singapore (SGP) entity. Each close required the finance team to explain every material period-over-period variance by hand. Sumvec AI built Claude skills that encode SES AI's own flux methodology, and SES AI's finance team now runs them during the live close for both entities.
The challenge
Each monthly and quarterly close, SES AI's finance team investigated each material variance across GL accounts, departments, vendors and reconciliations, then drafted commentary into the flux workbooks. The know-how behind this work sat with the analysts: which accounts matter, which dollar thresholds trigger commentary, which supporting reconciliation explains each driver, and the house style the explanations are written in. Generic automation could not capture those rules, so the work stayed manual.
The solution
Sumvec AI worked with SES AI's finance team to document the close methodology and encode it into Claude skills, built with Claude Code and Claude Sonnet and packaged as reusable workflows.
The skills carry four things:
Materiality rules. SES AI's materiality thresholds, which decide which variances need commentary.
• Account scoping. the accounts in scope for flux review, following SES AI's own decisions.
• Source-of-truth mapping. the reconciliation or ledger that is the source of truth for each GL area.
• Analyst voice. the analyst writing style used in SES AI's existing flux commentary.
The skills write explanations directly into SES AI's existing flux workbooks, in their existing format, and keep a full tie-out trace behind every explanation, so a reviewer can follow each figure to its source. SES AI did not need to change its close process to adopt the skills.
Results
• Accuracy. Validated against SES AI's prior analyst-written explanations as the reference, the skill matched about 95% of explanations end-to-end on a full monthly Balance Sheet flux. It matches because it applies SES AI's own thresholds and sourcing rules rather than generic heuristics.
• Independent use. SES AI's controller ran the flux independently for two of their entities without any assistance
• Adoption. SES AI now uses the skills for both entities.
• Quarter-end review. "During the Q1 2026 quarter-end review, re-performing the close on the automated process validated its accuracy and reliability.
Both the balance sheet and income statement flowed seamlessly from NetSuite into the financial statements without issue, tying out cleanly alongside the expected manual adjustments from the original preparation. Entity allocations and reconciliations aligned with high precision showing complete consistency in the automation core logic while standard post-automation checks were completed smoothly right within the close window."
Customer feedback
By re-performing the Quarterly and monthly close on the automated process, the team achieved seamless accuracy and efficiency. The balance sheet and income statement pulled directly from NetSuite and formatted into the financial statements smoothly with zero issues and tying out cleanly across the board, save for expected manual adjustments from the original preparation. Additionally, entity allocations and reconciliations were completed effortlessly, and the consolidated setup pulled accurately once the crosswalks were in place. Ultimately, the automation delivered reliable, predictable results, keeping post-automation checks minimal and right within expectations."
— Aljon, Controller, SES AI
