Scalability assumptions

1.8 Scalability assumptions

This section expresses capacity through workload variables. The architecture evaluates user count, receipt density, line-item count, retry rate, and batch policy together.

Symbol	Meaning
`U`	Daily active users
`r`	Average receipts per user per day
`a`	Share of receipts submitted for processing
`L`	Average line items per receipt
`v`	Share of verified receipts eligible for rewards
`e`	Average event records per receipt
`ρ_ocr`	OCR retry rate
`ρ_llm`	LLM retry or self-consistency rate
`B`	On-chain settlement batch size

Daily processed receipt volume:

text

R_d = U × r × a

Approximate monthly hot-data growth:

text

Rows_m ≈ 30 × R_d × (1 + L + e)

Daily model-call volume:

text

OCR_d ≈ R_d × (1 + ρ_ocr)
LLM_d ≈ R_d × (1 + ρ_llm)

Daily reward and settlement volume:

text

Verified_d ≈ R_d × v
Onchain_batches_d ≈ ceil(Verified_d / B)

Public variable-cost form:

text

Cost_d ≈ OCR_d × c_ocr + LLM_d × c_llm + Storage_d × c_storage + Settlement_d × c_chain

Capacity decisions are made by measuring U, r, L, retry rates, and batch policy together. Numeric thresholds and provider-specific cost coefficients remain in operational planning.