Case studies
Proven results.
Five engagements, with the figures the clients themselves put on them. Where a client asked not to be named, the sector is given instead.
Project Luna: document review automation
A major transformation programme required document review at a volume the firm could not staff. Doing it by hand was possible on paper and unaffordable in practice.
Automated document review and extraction across the programme, replacing the manual process end to end and reporting its own throughput so the benefit could be measured rather than estimated.
- Cost avoidance, as costed by the client
- €606KCost avoidance, as costed by the client
- Automated processing time
- 408hAutomated processing time
- Days of manual work replaced
- 34Days of manual work replaced
- Staff equivalents
- 10Staff equivalents
“The stats are astounding! Had the work been completed manually, it would have taken 10 staff working 12-hour days for 34 days at a cost of €606,000. With benefits such as these, it would be a great loss to the business to not develop this further.”
- Document extraction
- Office automation
- .NET
A multi-tenant communications platform, built and run
Member organisations were running contact lists in spreadsheets, campaigns through three unconnected tools, and donations through a payment link that never reconciled back to the person who gave.
A multi-tenant SaaS platform with per-tenant isolation, custom subdomains and Entra ID single sign-on. Campaign delivery across email, SMS and WhatsApp from one composer, with giving, Gift Aid declarations and events all reconciled back to member records. We built it, and we run it.
- Per-tenant data isolation
- Multi-tenantPer-tenant data isolation
- Delivery channels from one composer
- 3Delivery channels from one composer
- Live with paying organisations
- ProductionLive with paying organisations
- Next.js
- Node
- PostgreSQL
- Azure Container Apps
- Front Door
- Entra ID
Replacing four business systems with one
Quoting, point of sale, product specification and accounting each lived in a different product. Nothing reconciled, every order was re-keyed at least twice, and nobody could answer what a job had actually cost.
Odoo implementation mapped to the real procurement process rather than the textbook one, with migration off the incumbent tools, reconciliation at each step, and a phased cutover so the business keeps trading through the switch.
- Systems consolidated into one
- 4Systems consolidated into one
- Re-keying between systems
- 0Re-keying between systems
- Cutover, no trading downtime
- PhasedCutover, no trading downtime
- Odoo 19
- Python
- PostgreSQL
- ETL
Natural-language campaign building, with the brakes on
Building a segmented multi-channel campaign meant understanding the data model well enough to write the segmentation yourself. Most users never got past the default list, so the capability went unused.
A workflow service that turns a plain description of an audience into a validated segment and a draft campaign. Nothing sends without human review, and the generated segmentation is shown as editable rules rather than a black box. Per-tenant spend caps and provider fallback keep both cost and availability bounded.
- Review before anything sends
- HumanReview before anything sends
- Per-tenant AI spend ceiling
- CappedPer-tenant AI spend ceiling
- No single provider dependency
- FallbackNo single provider dependency
- FastAPI
- LangGraph
- OpenRouter
- PostgreSQL
Thirty minutes to thirty seconds, across the whole firm
Document processing took around thirty minutes per document. Across thousands of fee earners that is non-billable time the firm was absorbing every single day.
Enterprise-wide document processing automation rolled out across the fee-earning population, turning administrative handling time back into billable capacity.
- Reduction in processing time
- 98%Reduction in processing time
- Per document, from 30 minutes
- 30secPer document, from 30 minutes
- Fee earners affected
- 3,000+Fee earners affected
“The transformation of our document processing from 30 minutes to 30 seconds per document has revolutionized how our 3000+ fee earners operate. This represents a fundamental shift in our billable hour optimization.”
- Document automation
- Office.js
- Enterprise rollout
A document system with no tolerance for error
Case documentation in the justice system has to be right every time, and it is produced by people with very different levels of technical confidence. A tool that only works for confident users does not work.
A template and document system that guides every user to a correct result regardless of technical skill, maintained continuously across two decades as requirements and the desktop estate changed around it.
- Accuracy requirement
- 100%Accuracy requirement
- Years of partnership
- 20+Years of partnership
- Tolerance for error
- ZeroTolerance for error
“The partnership requires 100% accuracy and speed. The tool is used by varying skillset and technical competency, so it had to be easy to use, guiding users to 100% accuracy in no time.”
- Word automation
- VSTO
- Managed desktop deployment
Template modernisation and Office performance
Microsoft Office was taking fifteen to twenty seconds to start. Multiplied across a firm of lawyers opening documents all day, that is a material amount of chargeable time spent watching a splash screen.
Modernised the template management system and optimised the Office startup path: add-in management, startup process streamlining and configuration, alongside the migration to Office 365.
- Faster startup
- 70%Faster startup
- Down from 15-20 seconds
- 5-6sDown from 15-20 seconds
- Fewer IT support tickets
- 60%Fewer IT support tickets
- VSTO
- Office 365
- Template management
Self-service pitch creation after a rebrand
Following a major rebrand, fee earners needed to produce brand-compliant pitches quickly. Every request went through document services, which became the bottleneck exactly when speed mattered most.
Self-service pitch creation with brand compliance enforced automatically at the point of creation, integrated with Optimizely DAM so only permitted assets could be used.
- Fee earners create their own
- Self-serviceFee earners create their own
- Brand compliance
- AutomaticBrand compliance
- Document services bottleneck
- RemovedDocument services bottleneck
- Office automation
- Optimizely DAM
- Brand compliance
Get in touch
Recognise your situation in any of these?
A conversation with the people who would do the work, and a straight answer on whether we are the right fit for it.