Challenge
The team was spending significant time manually validating draft reports against a large set of regulatory and due-diligence materials. Each assessment required cross-checking claims and citations across 20–30+ reference documents(PDS, FSC materials, diligence questionnaires, manager responses, and internal assessments), while maintaining clear evidence traceability for every verdict.
In practice, this manual approach took an average of 8 hours per report. As report volumes increased, the process created structural bottlenecks and increased compliance and reputational risk. Outcomes could also vary depending on who performed the review and how evidence was tracked.
Critically, highly skilled analysts were spending the majority of their effort on administration rather than judgement. Around 60–70% of their time went to locating, reconciling, and tracking documents, leaving only 30–40% for applied analysis and decision-making.
Stack Highlights
The Approach
We designed an AWS-native Intelligent Document Review platform aligned with the Agentic AI charter, one that orchestrates foundation models, keeps outputs explainable, auditable, and governed, and scales across emerging document types.
1. Partnered with stakeholders to understand the current review workflow, clarify accuracy and compliance expectations, and prioritise the highest-value automation opportunities so effort was focused where it would have the biggest impact on turnaround time.
2. Designed an end-to-end review flow that standardises how claims are captured, verified against source evidence, cleaned of irrelevant detail, and assessed before producing a final verdict, improving consistency and reducing rework.
3. Built evidence traceability in from the start so every conclusion is supported by citations and can be inspected, explained, and audited, helping reduce citation errors and compliance risk.
4. Delivered an analyst-friendly interface to explore contradictions, review supporting evidence, and export annotated outputs in familiar ways of working, helping analysts spend more time on exceptions and oversight rather than manual cross-checking.
The Outcome
Codex delivered a deterministic, AI-driven document intelligence capability that scales across hundreds of assessments while preserving the research partner’s compliance requirements and established tone.
Key results included:
Automated fact validation across each report, producing consistent and repeatable verdicts
Significantly faster validation cycles, reducing average review time from around 8 hours to under 1 hour
Increased analyst capacity, freeing 5–6 hours per report and enabling the team to handle more funds without additional headcount
Lower review costs, with estimated annual savings of roughly $45k–$50k per 100 reports
A scalable foundation that can be quickly extended to new document families as requirements evolve
Overall, the solution streamlined the validation workflow, improved consistency and auditability, and allowed analysts to focus on higher-value oversight rather than manual checking.
Document Review Platform Highlights
Every verdict is backed by inspectable citations across 20 to 30+ reference sources such as PDS, FSC materials, diligence questionnaires, manager responses, and internal assessments. This strengthens defensibility and reduces citation risk.
A standardised flow for claim capture, evidence verification, relevance cleaning, and assessment produces repeatable, governed decisions. This reduces reviewer-to-reviewer variability and rework.
A reviewer interface brings contradictions and supporting evidence into one place, helping analysts spend time on judgement and oversight rather than manual cross-checking and administration.
Validation cycles reduced from about 8 hours per report to under 1 hour, freeing about 5 to 6 hours per assessment and enabling higher throughput without additional headcount.
Built as an AWS-native capability aligned to an Agentic AI charter, the platform extends quickly to new document types while keeping outputs explainable, auditable, and governed.
Talk to Us
We would love the opportunity to connect and understand more about the problems you are trying to solve.
Adrian Cambpell
Associate Partner, AI
Martin Campbell
Managing Partner
Get in touch to coordinate a meeting with one of our technical experts.
Australia: +61 7 3132 3002.
