iTromsø · IBM, VC2
10 minutes per day — time spent reviewing potential stories
Independently reportedMedia & entertainment · Document processing · NLP · In production
Document processing
United Trust Bank (UTB) · NashTech
Vendor-reportedBanking & financeUnited KingdomPilotAI agents
Vendor-reported. The customer is named and the numbers are quoted from the source page, but the account comes from the vendor. No independent confirmation.
Reported by nashtechglobal.com (vendor self-report). Checked against the source page on 2026-09-06. 1 of 1 figures below appear on that page word for word.
The problem
UTB’s QS team managed a high-volume, manual workflow where each report required careful extraction, validation and reconciliation of key financial and compliance elements.
What was deployed
- Vendor
- NashTech
- Products
- Gemini AI, Azure OpenAI, n8n, SharePoint Online, MongoDB, Azure container
- Models
- GPT-4
- Technique
- AI agents
- Build or buy
- Built in-house or to order
- Deployment
- Hybrid
- Data used
- QS PDFs, mocked financial data
What changed
Each row is quoted from the source. Figures we could not find on the page in those words are marked — they are kept, not deleted, so you can judge them.
5–6 minutesQS report analysis timefrom 30–60 minutes
“QS report analysis reduced from 30–60 minutes to 5–6 minutes, achieving an 80–90 per cent improvement.”
Quoted word for word from the source
uniform summary for every QS report, ensuring consistencyconsistency of QS report summaries
“The PoC produced a uniform summary for every QS report, ensuring consistency across assessments and enabling a clearer audit trail for future decision-making.”
Quoted word for word from the source
consistently identified discrepancies and exceptions that can be overlooked during manual checksDiscrepancies and exceptions identified
“The PoC consistently identified discrepancies and exceptions that can be overlooked during manual checks. This shows how automation can support a stronger control environment by providing consistent, rule-based validation.”
Quoted word for word from the source
“The proof of concept demonstrated just how transformative agentic AI can be. By engaging with NashTech, we were able to take requirements directly from our business experts, rapidly evaluate AI tools and language models and see tangible early results. At scale, we expect this capability to reduce a process that currently takes an hour to just a few minutes, freeing our experienced team to focus on exceptions and higher-value work rather than routine data evaluation.”
Difficulties and limits
- traditional extraction libraries proved inconsistent for UTB’s varied formats
Sources
- nashtechglobal.com — vendor self-report, 2026-01-12
Similar deployments
Amplitude · Wolfia
4X — throughput of security questionnaires
Vendor-reportedTechnology & software · Document processing · AI agents · Scaled
Adyen · Rossum
70 — automation rate
Vendor-reportedBanking & finance · Document processing · NLP · Scaled · 2023
Skaylink · Moterra
dropped by around 80% — NIS2 preparation time
Vendor-reportedTechnology & software · Document processing · LLM · Scaled
OM Pharma · DrugCard
3 minutes — review time per article
Vendor-reportedPharma & biotech · Document processing · NLP · In production
AMPREP (Amalgamations Repco) · Cimplify.AI
100% — invoice automation
Vendor-reportedManufacturing · Document processing · NLP · In production