Document intelligence

AI that reads, extracts, classifies and audits documents, with governance built in.

What document AI does, and the technology behind eachDrawn from p. 5

01Reading and extraction

Multimodal transformers
Spatial OCR
Recognising names and terms

02Summarisation and drafting

Large language models
Speech-to-text
Agentic prompting

03Classification and routing

Supervised machine learning
Semantic embeddings

Findings

01
Notably, there has been a 166 percent increase in Generative AI uptake within this sector, predominantly focused on internal operations such as compliance monitoring, auditing, and risk management.p. 3

From the report

Potential data sovereignty issues if hosted offshore, high risk of vendor lock-in, recurrent API costs, and lack of transparency into underlying model weights.p. 9

Three ways to source document AIRedrawn from the table on p. 9

Deployment ModelAdvantageDisadvantageCommercial APIsHighly scalable, low maintenance overheadHigh risk of vendor lock-inOpen-Source ModelsFull code transparencyRequires significant in-house technical talentSovereign ModelsFully aligned with national data sovereignty mandatesRequires robust internal computing infrastructure

Document intelligence

From the report

Start a conversation