Build in Digital’s latest microconference heard how the use of AI agents speeds up bid reviews, prompting a rethink in how estimators work
Preconstruction is where projects are won or lost, but as Build in Digital’s recent microconference heard, the use of AI is helping teams to tip the balance in their favour.
Agents in Action: How BW Is Transforming Preconstruction with a Smarter Tech Stack, sponsored by Egnyte, brought Thomas Hollingsworth, director of technology at BW: Workplace Exerts, together with Pavlina Nikolova, AEC practice leader, EMEA at Egnyte, to discuss the topic.
The format of a ‘fireside chat’ is a little different to the usual presentational style of the microconference events, but it was no less engaging and informative for that.
As the event heard, teams can spend countless hours reviewing specifications, drawings, and tender documentation; there’s got to a better way, right? That’s where Hollingsworth comes in, as he shares how BW is using AI agents and connected digital ecosystem to accelerate bid reviews, bring to the fore critical project information, and help teams make faster and more informed decisions.
Hollingsworth set out BW’s scale before turning to technology: the firm posted a record turnover of £327m, with a target of £500m within two years, and draws 30% of its business from technology-sector clients. He said BW had been experimenting with AI and machine learning “before it became widely known as AI”, citing an early facial recognition system for site access as its first application.
The conversation then moved to preconstruction – a stage Hollingsworth described as time-consuming and document-heavy, with roughly half of resources spent on bids that do not succeed.
Data structure
BW’s response, he said, has been to integrate data from preconstruction and construction platforms so estimators can draw on intelligence about subcontractors, pricing trends and long-lead items, helping the business make more selective tendering decisions.
Much of the discussion centred on how BW restructured its data before applying AI to it. Hollingsworth described a shift from siloed spreadsheets to a unified data portal drawing on Salesforce, Procore and Egnyte.
On the agents themselves, he explained that BW had first built AI around job roles, an approach that “became inefficient and expensive” as models were overloaded, before moving to skill-based agents that route tasks according to the capability required rather than a specific role.
That approach now underpins BW’s handling of pre-qualification questionnaires (PQQs), where agents extract and process information from tender documents, produce structured summaries and generate SWOT analyses. Hollingsworth said the speed of prototyping had helped win over sceptics within the business.
Security and governance featured prominently in the discussion. Hollingsworth pointed to BW’s ISO 27001 certification, its Cyber Essentials Plus status and progress toward ISO 42001, alongside access controls managed through Egnyte and Microsoft Entra. Nikolova acknowledged the technical safeguards but pressed him on the people side of adoption.
Here, Hollingsworth argued that change management matters as much as the technology itself. BW gives staff ownership of projects through an internal innovation website and a leadership development programme, he said. While acknowledging a generational experience gap: some younger staff are highly capable with AI tools but risk missing the underlying processes, which he likened to the historical ban on calculators in exams.
Always moving
BW is currently running around 20 initiatives at once, Hollingsworth said, describing a “never stand still” approach that now extends to programme building and early moves toward more autonomous robotics. His advice to other technology leaders was to start now: current tools and data capabilities offer a strong foundation, he said, unconstrained by legacy production systems.
On analysis of 2D and 3D data, Hollingsworth said AI models are increasingly able to read complex drawings, with 2D data already used for contract review and 3D being explored for generating walkthroughs, despite ongoing computational limits. He credited AI with improving efficiency in processing tender documents and contributing to a strong bid win rate, though he attributed that success primarily to the preconstruction team rather than the technology alone.
Hollingsworth closed by stressing that BW’s approach to fit-out work is deliberately selective, a strategy he said has strengthened client loyalty rather than costing the business work, and by emphasising BW’s preference for collaborating openly with partners over guarding innovations for itself.
Read next: £107m NHS National Rehabilitation Centre completed
Are you a building professional? Sign up for a FREE MEMBERSHIP to upload news stories, post job vacancies, and connect with colleagues on our secure social feed.
