GWWO Architects
Background
GWWO, a Baltimore-based architecture firm, has crafted prominent cultural and educational spaces, including the Harriet Tubman Underground Railroad Visitor Center and the National World War I Memorial. Supporting these projects requires an extensive repository of institutional insight, highlighted by a comprehensive compilation of design standards and lessons learned started in 2008. Eric Feiss, Principal, describes this resource as “an encyclopedia of how to design and document buildings at GWWO.”
According to Dan Audette, Senior Associate and Technology in Practice Specialist at GWWO, this wealth of expertise is central to the firm’s client value proposition: “When you hire an architecture firm, you’re hiring an entity. There’s a body of knowledge here of decades and decades of experience that we all should be able to utilize and access.”
GWWO named its AI initiative Project Burnham, inspired by the Chicago architect Daniel Burnham’s famous principle, “Make no little plans.” In line with this vision, the initiative aims to consolidate decades of collective project knowledge and make it all timeless and seamlessly accessible across the entire firm. Partnering with Mindgrub for an AI Sprint 0 converted this ambitious goal into an actionable, step-by-step strategy.
The Opportunity
Around 2016, GWWO began talking about what the team called “the database of all things.” The idea was one connected source that architects, project managers, accounting, and marketing could all use. Earlier conversations with developers went nowhere. The cost was too high, and the technology of the time couldn’t make sense of inconsistently organized records without a manual cleanup that no one on the team had time for.
Generative AI made the project feasible. It also landed on Dan, a licensed architect who taught himself to program, and now spends much of his day writing code. “I had a list of responsibilities before AI hit,” he says, “and then AI hit, and I just added responsibilities and didn’t take any away.”
GWWO needed help, not just a vendor to take the project off its hands. “It was never really our intent to just hand this over to a consultant and say, here, build this for us,” Eric says. The team wanted a partner who would work alongside the firm and really empower them to carry the work forward.
When selecting a partner, Mindgrub’s Baltimore roots mattered because GWWO has found that its advisors, from accountants to attorneys, serve it best when they know the region. Breadth mattered too. Beyond the data lake itself, the project would need interfaces and applications that non-technical staff could use, and Mindgrub’s AI, web, and app teams could cover a scope that was only still taking shape. From the very first conversations, Eric also sensed the team would be easy to work with and understanding of the challenge.
Our Approach
Knowing the GWWO team had a handle on their vision, Mindgrub treated Dan as a full-on collaborator from day one.
Discovery reached well beyond the design studio. In a series of workshops led by Mindgrub, the team heard from accounting, marketing, business development, project managers, construction administration, designers, and staff at every experience level. The goal was to learn where each group loses time and what information would help them work more efficiently. In tandem, the team inventoried and identified 5 main systems holding GWWO’s project knowledge that needed to connect first.
Mindgrub approached tool selection with no vendor agenda and recommended building on the Microsoft environment GWWO already uses. The plan put governance ahead of everything else. Permissions, sensitivity labels, and privacy rules are designed before any content is indexed, and financial data is surfaced only at the project level. These are the structural questions an architecture team isn’t trained to ask, but can’t afford to skip, allowing Mindgrub’s expertise to shine and serve without overtaking the clear vision of GWWO.
“I’m a self-taught programmer, so I half expected to be told I didn’t know what I was talking about. Instead, Mindgrub brought me into the process. They explained their thinking, encouraged me, and helped me see the things I didn't know to ask about.”
Senior Associate and Technology in Practice Specialist, GWWO
A Roadmap Built to Prove Value Early
GWWO thought about their ideal system in three layers: the underlying data, a conversational layer on top of it, and applications that put it to work. Sprint 0 with Mindgrub turned those layers into a phased roadmap, with each phase scoped to justify the next.
-
Phase 0.5, Proof of Concept:
The Project Knowledge Engine, a Copilot agent in Microsoft Teams grounded in GWWO's proposal archive and design guidelines. It required no new infrastructure. A pilot group of 10 to 20 people from marketing, business development, project management and firm leadership to generate the usage data that determines where to invest next.
-
Phase 1, Data Lake Foundation:
The five priority systems flow into a single Azure data lake. Everything is organized around a shared project number and employee ID, so records from different systems finally line up, and are correctly attributed.
-
Phase 2, Search:
Firm-wide search across projects, people and documents, plus a one-click Project Info Sheet. A rollup that ended up being requested independently in four separate workshops.
-
Phase 3, AI and Automation:
A menu of 14 potential capabilities GWWO can choose from based on what the pilot data shows.
The Phase 3 options range from practical to ambitious. They include automating project setup, a 30- to 90-minute task that happens 40 to 60 times a year, and billing automation that could give the accounting team back an estimated three days a month. At the ambitious end of the Sprint would be a fully realized proposal-drafting agent that turns an incoming RFP into a first draft about 70 percent complete.
What It Looks Like in Practice
Eric says at least one architect in the office faces a first every day, whether that’s designing an accessible bathroom or working with a civil engineer on a roadway cut. For example a distinctive type of paneling for exterior cladding that rarely appears on projects.
The realized GWWO Agent would allow an architect specifying zinc for the first time to ask where GWWO has used it before. The answer would include relevant drawings and details, technical specifications, photos of finished buildings, RFIs and change orders from construction, and any documented lessons learned. It would also name the project managers who led those projects.
That last detail perfectly reflects how GWWO works. The firm opposes using AI to generate designs, and the roadmap enforces that line: the proposal-drafting agent is built to refuse requests for renderings or building forms. The information gives architects a starting point, and the colleagues who did the work fill in the nuance. “You still have to collaborate,” Eric says. Freeing up that administrative time for an architect to dedicate their skills to the complex, human work is where AI’s value is undeniable.
“Sprint 0 gave us what we needed: confidence that we were on the right track, a clear view of the roadblocks ahead, and a roadmap our team can build on. Mindgrub met us where we were and helped us see the path forward.”
Principal, GWWO
Results That Matter
-
A phased roadmap with detailed estimates,
so GWWO can fund one stage at a time and choose its AI capabilities based on evidence
-
The Project Knowledge Engine, built by Mindgrub, in use with a pilot group:
86% increase in weekly users since launch, and 1.9x increase in weekly queries.
-
Revived initiatives.
"All these dead projects that we thought were great ideas, but didn't have the resources to accomplish them, are all coming back to life," GWWO says.
-
A flexible path forward.
GWWO and Mindgrub are now exploring a retainer for scoped projects and for strategic input when they need it.
