黑洞社区

Have you implemented an AI solution/tool into your business in the last 12 months? If yes, what was it and do you have results or impacts from it you can share?

Turner & Townsend: 鈥淚n addition to trialling commercially available AI for general productivity, we are piloting OpenAI and Google Gemini solutions within our cloud applications to improve the quality of our core deliverables.

鈥淩esults so far are promising but appear limited by the current capability of AI. Collaborating with clients and partners, we鈥檝e introduced AI tools like Buildots, Nodes & Links, and BuildPrompt. Buildots monitors site progress, Nodes & Links analyses schedules for efficiencies, and BuildPrompt discreetly prompts data for client queries.鈥�

Arup: 鈥淲e have already deployed multiple AI systems as a way of augmenting human intelligence, to enhance our design, engineering and consulting offers to our clients, with a particular focus on achieving sustainability outcomes. One example of this is Uheat, a digital solution designed to help urban planners and city authorities to bring down the temperature 鈥� also known as the urban heat island effect. By using a combination of satellite imagery and AI trained on open-source climate data, we are now able to analyse huge areas of cities and the particular buildings, structures and materials that are causing temperatures to rise.

Mace: 鈥淲e currently use specialised AI tools across a number of very specific use cases. In the last 12 months Mace has undertaken trials of a number of more general-purpose generative AI tools across a variety of office productivity tasks. We have yet to progress to widespread deployment of generative AI technology.鈥� 

JLL: 鈥淲e have implemented our own secure AI tool for internal use. JLL GPT is an AI assistant. Since its launch, JLL GPT has offered numerous benefits including enhanced efficiency, improved decision-making capabilities and simplified data analysis. By automating routine tasks, providing real-time insights, and leveraging advanced natural language processing, JLL GPT has empowered users to streamline workflows, save time and make more informed business decisions.鈥�

Todd Architects: 鈥淢S Copilot 鈥� helped with responding to development of documents and reports and input on submissions. Revit Suite 鈥� using software to allow parametric design, our Revit group have been exploring the ability to establish for initial feasibility studies such as capacity studies on residential schemes.鈥�

Systra: 鈥淲e have an internal large language model tool that allows staff to access the benefits of a ChatGPT type solution but within a secure organisation environment.

鈥淭his is increasingly being used to help with reviewing documentation, initial understanding of opportunities, basic background research in new subject areas.

鈥淲e also have an AI-powered automated tool for checking standard design documents. We have an AI supported knowledge-sharing internal tool. We are in development of a number of other AI applications across the business and we have developed a number of machine learning applications for clients.鈥� 

Magnitude Quantity Surveyors: 鈥淲e have been building a proprietary software for construction cost benchmarking, which following launch will have an integrated carbon calculator.

鈥淲e have utilised AI to assist with coding, which has significantly reduced the cost of external consultancy assistance.鈥�

Hydrock Group: 鈥淲e鈥檙e in the process of implementing an AI tool, which is currently being trialled across our business, that reviews our invoices and reflects the likelihood of those invoices being paid on time.鈥�

Stantec: 鈥淲e are testing Microsoft鈥檚 Copilot AI assistant, allocating licences across our global organisation 鈥� the first phase aims to quantify productivity gains across six of the most used applications. The second phase assigned users to focus groups to comment on their learnings, most practical uses, accuracy and perception of its value. Results indicate that Copilot is resoundingly useful. The most frequent uses were generating Word content, summarising Teams meetings and searching emails, meetings and chats using Microsoft365.鈥�

Aecom: 鈥淲e are building an AI knowledge platform to support our teams, starting with AI-driven software to significantly speed up bid information-gathering, so that experience and expertise globally can be drawn together in seconds. This is now being expanded into enterprise resource planning and customer relationship management.鈥�

Hoare Lea: 鈥淲e鈥檙e using an open-source large language model running off-grid within our organisation with secure access to our design data and internal guidance and standards. We鈥檙e seeing some very encouraging early results, saving engineering teams hours of work to manually find answers. We鈥檙e also getting some exciting results in getting the LLM to directly query our own engineering data platform to provide predictive support in early-stage design.鈥�

鈥淲e鈥檝e applied machine learning to uncover insights from large public sector datasets to identify work-winning opportunities. We have developed a generative AI chatbot for client biodiversity enquiries and are prototyping a generative AI chatbot for employees to better access our knowledge base.鈥�

Barker Associates: 鈥淲e have implemented Synthesia 鈥� an AI solution to create video content for internal and external communications.鈥�

Baily Garner: 鈥淵es, Autogen [an open-source programming framework for building LLM applications] This does positively impact efficiency and productivity, however requires trained, experienced and skilled operatives to bespoke and be of use, otherwise it can be misleading.鈥�