£3m AI Value Unlocked in Six Weeks for UK Equity Broking Firm
£3m a year of capacity release identified across 30+ costed AI use cases
Governed data platform built in four weeks using spec-driven development on Microsoft Azure
Background
Our client is a UK equity broking firm that advises small and mid-cap companies across corporate broking, market research and post-trade operations, and processes around 20,000 trades a day. The firm saw clear potential for AI, but it lacked the internal capacity to work out where the value sat and to build the AI literacy needed to adopt it safely in an FCA-regulated environment.
The pressure was felt most in research and operations. Analysts were spending hours each reporting morning comparing announcements against forecasts. Draft research notes were waiting days for a part-time editorial and compliance review, whilst the the post-trade team was manually triaging mismatched trades while preparing for T+1 settlement.
Solution
WeBuild-AI ran a six-week AI accelerator that delivered the following workstreams:
AI education. Sessions for each team covered AI fundamentals, prompt engineering, how to identify a strong use case and safe use in a regulated environment.
Use case discovery. Discovery sessions with the Research, Post-Trade Operations and Corporate teams produced 30+ AI use cases. Each was sized, costed and assigned a delivery approach (Copilot, build or buy), then placed on a roadmap that balanced value against complexity.
Data platform. WeBuild-AI built a governed Azure data platform across the firm’s trading and market data domain in four weeks. It was delivered using a spec-driven development approach, whilst leveraging infrastructure as code and GitOps automation capabilities. The platform housed a series of primary data products, whilst being made accessible through a natural language data analysis through the use of Microsoft Copilot.
Production agents. WeBuild-AI took four Copilot agents and workflows into production across the research and trading teams within six weeks. Between them, they cover compliance review of draft research, company research and data gathering, research drafting in each analyst’s own style, and daily market briefings on house stocks. Together they are forecast to release around 4,770 hours of effort a year which equates to ~£0.5m in cost from the organisation's balance sheet.
“In six weeks we went from not knowing where to start with AI to a cross cutting data platform, 4 production agentic use cases, and teams who now understand what AI can and cannot do for us.”
- Chief Technology Officer
Impact
Around 4,770 hours a year are released by the production agents, worth around £0.5m in just 6-weeks.
Research Notes compliance review process reduced from a multi-day turnaround to just minutes for new publications.
A governed Azure data platform built in four weeks using spec-driven development, infrastructure as code and GitOps automation.
Four Copilot agents and workflows taken into production across research and trading within six weeks.
100+ employees trained in practical and safe AI use, whilst surfacing 30+ AI use cases worth £3m a year of capacity release at full adoption.
Why WeBuild-AI
WeBuild-AI is an engineering-led AI consultancy that designs, builds and runs AI agents in production for global enterprises. Our experience in building agents that can operate safely in regulated environments. The customer partnered with us because we were able to develop agents that provided reliable, explainable output, whilst keeping a human in control of key decisions. Ensuring the solutions we delivered could stand up to scrutiny from risk and compliance teams as well as the regulator.
Our forward deployed engineers worked alongside the research, trading and operations teams to map how their work actually gets done, and then built agents around those workflows rather than asking the teams to change how they work.
