01 • Overview
From Fragmented Project Data To A Single AEC Operating Picture, With Gen AI Live In The Hands Of Studio Principals.
02 • Problem statement
Professional services organisations rarely have an easy, company-wide platform to search previous projects, so business development rebuilds the case from scratch, delivery re-solves solved problems, and the value already paid for once is quietly paid for again.
We had the systems. We didn't have the answers. Every Monday felt like starting from scratch.
Studio principal, anonymised
Starting point • Jan 2024
14 studios, four core systems, no shared spine. Strategy decisions were being made on Monday-morning spreadsheets reconciled by hand.
Project truth was split across CRM, finance, the asset library and dozens of drives. Nobody could answer “have we done this before?” without ringing round, and the answer depended on who picked up.
Bid teams spent 5-8 hours per pursuit hunting comparable projects, fee benchmarks and imagery, and still missed the strongest examples because they simply never surfaced.
Details, specifications and design solutions already paid for on one job were rebuilt on the next, because finding the original cost more than starting again. New joiners learned the back catalogue by osmosis.
03 • The impact
Users by discipline
Architecture first, then interiors, landscape and planning.
People using the platform each month
Benefit by month
$7.9M total over three years. Value climbs as teams take it up, then holds.
Benefit per month, $ thousands
04 • Role
A single delivery team covering the data foundation, the interface design, the web application and the agentic layer, so it landed as one product rather than four workstreams handing off to each other.
Lead
Data platform, UX design, web application and agent layer, end to end.
Partner
Information governance, IP and confidentiality policy alongside internal IT and practice leadership.
Handover
Product ownership, backlog and trained internal owners by month 9.
Stood up an AWS Databricks lakehouse with the Digimasters AEC data model pre-built. Governed pipelines from Deltek, Dynamics and Revit landed under Unity Catalog with a shared semantic layer.
Re-implemented Dynamics 365 / AEC360 with a clean pursuit-to-project lifecycle. Integrated to Vantagepoint so finance and pursuit data stopped diverging at handover.
Built an agentic project-intelligence assistant for studio principals: RFP triage, fee benchmarking, resource forecasts and similar-project lookups, all running against the governed semantic layer.
Ran data literacy and Gen AI coaching across 14 studios. Measured success by weekly active usage, not training attendance.
05 • Timeline
9 months · embedded delivery
Month 1
Discovery across bid, design and delivery. Data audit and roadmap signed off.
Month 2
Figma prototype of search, filter and compare tested with users in three offices.
Month 3
Fabric lakehouse live with CRM, finance and asset-library pipelines under one semantic model.
Month 5
Project Design Service released to first two offices with 20 years of history searchable.
Month 7
Natural-language lookup, similar-project matching and precedent packs in production.
Month 9
All offices live. Product ownership, backlog and support model handed to the internal team.
06 • The team involved
A Small, Senior Squad: Data, Design, Engineering And AI In One Team, Embedded With The Practice For The Duration Of The Engagement.
Roles
For the first time, twenty years of our own work is something you can actually search. New starters find things in minutes that used to take a decade of relationships.
Design director, anonymised
07 • Technologies used
Technology partners
Data platform
AI & agents
Product & build
Governance
Get in touch