A $1.2B-revenue specialty chemicals manufacturer cut month-end close from 11 working days to 3 with a LangGraph + Temporal multi-agent finance platform integrating SAP S/4HANA + Oracle E-Business Suite + NetSuite. Year one: $850K annualised savings, 9,400 monthly journals automated and zero SOX 404 ITGC findings.
Discuss Your Project
AI Finance Automation Agent on LangGraph + Temporal stateful workflow orchestration with Anthropic Claude 3.5 Sonnet — $850K Annual Savings for a Regional Financial Services Firm
A $1.2B-revenue mid-market specialty chemicals manufacturer with 1,400 employees, 14 plants across 9 countries and three different ERPs (SAP S/4HANA, Oracle E-Business Suite, NetSuite) was closing its books in 11 working days — well behind the Day-5 industry benchmark. DreamzTech engineered a six-agent finance automation platform on LangGraph + Temporal with Anthropic Claude 3.5 Sonnet. Result: month-end close cut to 3 days, $850K annualised savings, 9,400 journals auto-posted monthly and zero SOX findings.
The client faced significant operational and financial challenges that demanded a custom AI month-end close automation platform tailored to their multi-subsidiary month-end close workflow, vendor-format diversity, and finance compliance requirements.
DreamzTech architected a production-grade AI month-end close automation pipeline on AWS, Azure or Google Cloud with five interconnected modules — from ingestion to GL posting — delivering specialist-agent extraction, AI line-item reconciliation, PO + GR + journal entry PO + GR + journal entry three-way matching, and deep ERP integration with full audit trails.
Four integrated platform components were built and launched in a production-grade engagement on AWS, Azure or Google Cloud with HIPAA-eligible / SOC 2-aligned security, signed Microsoft SOX evidence chain, and seamless SAP S/4HANA + Oracle E-Business Suite + NetSuite (heterogeneous ERPs) + NetSuite integration.
We trained a six-agent specialist crew on 200+ historical vendor journal entry samples drawn from each subsidiary, covering structured PDF journal entrys, scanned images, faxed statements and handwritten freight bills. The model extracts header fields (journal entry number, date, vendor, totals, tax codes), line items (description, qty, unit price, GL hint, tax) and tables with row/column structure preserved. Out-of-the-box accuracy reached 92% on standard journal entrys and 95-98% on the trained vendor set after three retraining cycles. Field-level confidence scores drive the downstream review workflow.
Anthropic Claude 3.5 Service (GPT-4o) handles the unstructured comprehension layer that templates cannot solve: vendor-name normalisation across DBA / subsidiary aliases across abbreviations and DBA aliases, PO number lookup when vendors omit or mis-format it, GL account classification using the firm’s 2,400-line chart of accounts, and reasoning traces explaining each match decision. We integrated Pinecone / Weaviate vector search for retrieval-augmented generation over historical journal entry patterns, and LangChain for prompt orchestration with structured-output JSON validation.
Low-confidence pages (extraction confidence below 0.85 on critical fields, or three-way-match exceptions) route to a Power Automate review queue. staff accountants see side-by-side document + extracted JSON, edit fields directly, and approve or reject. Microsoft Entra ID role-based access enforces subsidiary segregation and SOX-required separation of duties between staff accountant, AP supervisor and CFO approval thresholds. Every correction feeds back into a weekly specialist-agent retraining job — accuracy improved month-over-month from 92% to 98% on the trained vendor set.
Approved journal entrys post to SAP S/4HANA + Oracle E-Business Suite + NetSuite (heterogeneous ERPs) (3 subsidiaries) and NetSuite (1 subsidiary, smallest entity) via AWS API Gateway with retry logic and idempotency keys. Kafka guarantees delivery; Azure Monitor + Log Analytics captures every document access, extraction call, override and approval as immutable audit trail entries for SOX 404 compliance. Customer-managed keys (CMK) in the cloud Key Vault encrypt all journal entry data at rest; TLS 1.3 protects data in transit.
Measurable business outcomes delivered in the first nine months post-launch — validated by SAP S/4HANA + Oracle E-Business Suite + NetSuite (heterogeneous ERPs) reporting and the firm's internal AP analytics dashboards.
Reduction in manual journal entry data entry across all 14 manufacturing plants
Straight-through month-end close automation — no human touch
Annual savings delivered within 12 months of go-live
Vendor journal entry formats trained on LangGraph + Temporal orchestration six-agent specialist crews
Invoice cycle time reduction (from 3 days to under 4 hours)
Subsidiary month-end close workflows unified on a single Azure IDP platform
DreamzTech delivers custom AI month-end close automation and intelligent document processing platforms for specialty chemicals, insurance, healthcare and public-sector clients. Microsoft Solutions Partner, AWS Partner and Google Cloud Partner with 200+ AI projects across 15 countries and 97% client retention.
Continue your intelligent document processing journey — pick the cloud, we build the system.
Cloud-agnostic AI IDP — extract, classify, validate and route journal entrys, contracts, claims, KYC and medical records across AWS, Azure or Google Cloud.
AWS-native IDP on Amazon Textract, Comprehend, Bedrock with Anthropic Claude, A2I and Lambda — Step Functions orchestration, GovCloud-ready.
cloud-native IDP on LangGraph + Temporal orchestration , AI Language, Anthropic Claude 3.5 and Step Functions — FedRAMP High on AWS GovCloud.
See how DreamzTech delivers AI document processing across legal and insurance verticals — with measurable ROI in months, not years.
Amazon Textract + Anthropic Claude 3.5 Sonnet on Amazon Bedrock + custom Amazon SageMaker NER trained on 45,000 prior contracts. 90+ clause types extracted at 99.1% accuracy. Paralegal review time 40h → 12h per contract. $2.4M annual billable-hour recapture for a top-100 global law firm with 6 offices.
Intelligent document processing + EXIF / metadata forensics + vision-capable LLMs (Claude 3.5 Sonnet, GPT-4o, Gemini 1.5 Pro) + graph-based cross-claim similarity. 62% improvement in fraud catch rate, $5.1M prevented losses year one, and 87% faster SIU triage (45 min → 6 min per claim) for a national P&C insurance carrier.
LangGraph + Temporal stateful workflow orchestration with Anthropic Claude 3.5 Sonnet ships with a prebuilt journal entry model out of the box that handles common formats with 88-92% accuracy on day one. For specialty vendor formats — handwritten freight bills, multi-subsidiary tax codes, faxed statements — the six-agent specialist crew trains on as few as 50-500 labelled examples and reaches 95-99% accuracy. The Document Intelligence Studio gives staff accountants a labelling UI without writing code, and the service is HIPAA-eligible under Microsoft’s SOX evidence chain — important for any financial-services workload that touches PII in vendor master.
LangGraph + Temporal orchestration handles structured extraction; Anthropic Claude 3.5 Sonnet as primary reasoning engine across six specialist agents handles the reasoning layer. We use it for: (1) vendor-name normalisation across DBA / subsidiary aliases across DBA / subsidiary aliases, (2) PO number lookup when vendors omit it, (3) GL account classification against the firm’s chart of accounts, (4) variance summarisation for the staff accountant review screen, and (5) duplicate detection across plants with explainable reasoning traces. All Anthropic Claude 3.5 calls run in the firm’s AWS account under their data-residency and SOX evidence chain — journal entry data never leaves their environment.
For this engagement we integrated with SAP S/4HANA + Oracle E-Business Suite + NetSuite (heterogeneous ERPs) (3 subsidiaries) and NetSuite (1 subsidiary). Across other DreamzTech AI month-end close automation engagements we have shipped integrations with SAP S/4HANA, Oracle E-Business Suite, NetSuite, Sage Intacct, Workday Financials, SAP S/4HANA and custom legacy AP systems. All integrations use AWS API Gateway with retry, idempotency keys and Kafka for guaranteed-delivery messaging.
Twenty-two weeks total. Phase 1 (LangGraph + Temporal orchestration multi-agent topology and ERP integration + SAP S/4HANA + Oracle E-Business Suite + NetSuite (heterogeneous ERPs) integration + go-live for 1 subsidiary) shipped in 14 weeks. Phase 2 (Anthropic Claude 3.5 PO + GR + journal entry PO + GR + journal entry three-way match + Power Automate human-review + rollout to remaining 3 subsidiaries) added 8 weeks. The first useful AP capability was in production by week 8 — enabling the corporate finance team to start extracting and posting journal entrys for the largest subsidiary while we trained models on the smaller subsidiaries’ data in parallel.
Within 12 months of go-live: 73% reduction in manual journal entry data entry, 85% auto-post rate on monthly journals (no human touch), journal entry cycle time cut from 3 days to under 4 hours, $850K in annual savings (7 reduced staff accountant FTE-equivalents plus zero SOX 404 ITGC findings on year-end audit), and a 41% drop in journal entry exception rework. ROI was achieved in 9 months. The six-agent specialist crew accuracy continues to improve month-over-month as staff accountant corrections feed back into the weekly retraining job.
