Type to search every page. Use the arrow keys to move through results.
Industries
Financial services
Leaders in banking, asset management, fintech and real estate answer to supervisors, investors and lenders for what their organizations do with AI. Published research shows AI in wide use at financial firms and still at the pilot stage in much of real estate, and supervisors and professional bodies now expect that use to be explained.
Where the industry stands
In Canada, finance and insurance businesses are among the most likely to use AI. In Statistics Canada’s spring 2026 survey, finance and insurance was one of the three industries most likely to use it. Of its businesses, 40.4% reported use in the previous 12 months, against 19.2% of all businesses. The agency gives no such figure for real estate. In its summer survey of plans, 23.6% of businesses in real estate and rental and leasing expected to use AI over the next 12 months, against 25.2% of all businesses.
What leaders answer for has moved on from whether to use AI. Canada’s federal financial supervisor, OSFI, has issued a model risk guideline that covers AI and takes effect on 1 May 2027. Staff of Canada’s securities regulators have said that using AI directly in suitability and trading decisions for client accounts will be problematic unless a high degree of explainability is assured. The Bank of Canada began supervising retail payment providers in September 2025, and says about 1,600 have registered. RICS made its standard on the responsible use of AI mandatory for its members and regulated firms from 9 March 2026.
The four sub-industries differ in what they hold and whom they answer to. The research points to the same questions in each: how far the use of AI can be explained, how it is governed and what it returns. Those are the questions we work on. We help you decide which uses are worth finishing, govern the ones you keep and report on them in terms a board, an investor and a supervisor can follow.
The figures on this page are third-party research. Each links to the organization that published it, and none describes our own work.
In the data
Statistics Canada asked businesses across industries about their plans for AI. Finance and insurance sits well above the all-industry figure. Real estate sits just below it.
Share of Canadian businesses planning to use AI over the next 12 months, by industry% of businesses in each industry, third quarter of 2026 (9,601 responses)
Professional, scientific and technical services54.8%
Information and cultural industries48.0%
Finance and insurance35.5%
Health care and social assistance33.8%
All industries25.2%
Arts, entertainment and recreation24.5%
Manufacturing24.4%
Real estate and rental and leasing23.6%
Administrative and support services19.6%
Construction18.3%
Accommodation and food services17.4%
Wholesale trade16.8%
Other services (except public administration)16.7%
Retail trade16.0%
Mining, quarrying, and oil and gas extraction14.8%
Agriculture, forestry, fishing and hunting13.2%
Transportation and warehousing8.0%
These figures are plans, and the survey table does not measure use. Finance and insurance includes insurers. Real estate is real estate and rental and leasing. The survey’s table covers 16 industries, all shown here.
Leaders in banking and capital markets answer to supervisors for decisions across lending, payments, trading and compliance. That now includes decisions in which AI has a part.
Supervisors expect a bank to know which models drive its decisions, to test them and to explain them. That now reaches AI.
OSFI’s final guideline on model risk, published on 11 September 2025 and effective from 1 May 2027, covers AI models. It expects explainability to be assessed in line with a model’s intended use.
Source: OSFI
Our part
We set up the governance that gives each AI tool an owner, a stated purpose and a record you can show a supervisor.
Financial firms use outside suppliers for a growing share of their AI, and a few suppliers carry much of the load. A leader answers for what happens if one of them fails.
In the Bank of England and FCA survey of 118 UK financial firms, a third of AI use cases were implemented by third parties, up from 17% in 2022. The top three providers accounted for 73% of reported cloud use.
Source: Bank of England and FCA
Our part
We map which of your services depend on which providers, and prepare systems to be recovered or moved if one becomes unavailable.
Tools watch payments and accounts for unusual activity and flag cases for a person to review. A bank answers for what it catches and for what it wrongly blocks.
The Bank of England reported that 64% of UK financial firms surveyed in 2024 were using or planning to use AI for fraud detection, and 62% for cybersecurity.
Source: Bank of England
Our part
We build automations and AI agents that sort alerts and route them to your analysts, with a person making each decision.
Regulators have begun to warn about AI tools that act on their own. A trading firm needs to be able to show afterwards what a tool did, on whose instruction and with what result.
FINRA’s 2026 oversight report for US broker-dealers warns that AI agents may act beyond their intended scope and that how they reach a result can be hard to trace. It suggests that firms keep records of what was asked and what was produced, track which version of a tool is in use and have people review the results.
Source: FINRA
Our part
We build the reporting that lets a desk, a risk team and a supervisor see what an AI tool did, when and on whose instruction.
A bank’s risk, compliance and operations teams are asked to stand behind the tools it uses. They can do that only if they understand a tool well enough to question it.
In the Bank of Canada’s 2026 survey of 54 senior risk-management experts, 56% named talent-related constraints as a barrier to using AI more.
Source: Bank of Canada
Our part
We train your risk, compliance and operations teams to use and question AI tools, and help you hire the skills you lack.
Integration comes first. The Bank of Canada surveyed 54 senior risk-management experts in 2026. Of them, 58% named difficulty integrating AI into existing infrastructure and workflows, 33% data security and privacy concerns and 31% the cost of implementation. Source: Bank of Canada
Understanding is incomplete. In the UK survey, 34% of firms reported a complete understanding of the AI they use and 46% a partial one. Source: Bank of England and FCA
The guidance is still taking shape. The Federal Reserve’s Vice Chair for Supervision said that in April 2026 the US federal banking agencies amended their model risk guidance so that it does not apply to generative AI or to AI agents. Source: Federal Reserve
The questions we start with
Which of the AI tools in use at your firm could you explain to a supervisor, decision by decision?
How many of your AI uses depend on an outside provider, and what would you do if one became unavailable?
When a tool flags a payment or an account, who reviews it, and is that review recorded?
Which of your AI tools will the model risk guidance that applies to you cover, and from when?
Could your risk and compliance teams say what each AI tool may and may not be used for?
Asset management
Leaders in asset management answer to clients and securities regulators for their recommendations. They also answer for what they say about how AI informs them.
Two pressures meet there. Fees have fallen: in the US, the average expense ratio paid by equity mutual fund investors fell from 1.04% of assets in 1996 to 0.40% in 2025. Clients and regulators have also begun to test what managers say about their methods.
Among alternative fund managers, use of generative AI is now near universal, and more than half expect to use more of it in investment work within a year.
Generative AI at alternative fund managers% of fund managers surveyed (150 in 2025)
20232025
Use generative AI in their work2023: 86%2025: 95%
Expect more generative AI use in investment processes over the next year2023: 20%2025: 58%
Alternative investment managers, which include hedge, private market and digital asset funds, so not all asset management. The 2023 figures are those AIMA reports for its earlier survey, when nearly half were unsure whether they would use more.
Investors and regulators read what a manager says about its AI. The statements have to match what the firm does and the controls around it.
In March 2024 the US Securities and Exchange Commission settled charges against two investment advisers for false and misleading statements about their use of AI, with penalties of US$225,000 and US$175,000.
Source: US Securities and Exchange Commission
Our part
We review how you describe your AI use against what your tools do and the controls around them, and help you put governance in place.
The closer an output sits to a recommendation, the more a person has to stand behind it. In Mercer’s 2026 survey of 131 asset managers, 5% granted AI autonomous or semi-autonomous authority over investment recommendations or trades.
Staff of Canada’s securities regulators said in December 2024 that using an AI system directly in suitability and trading decisions for client accounts will be problematic unless a high degree of explainability is assured.
Source: Canadian Securities Administrators
Our part
We train your investment and client teams in what each tool can be relied on for and which outputs a person must approve.
Managers report gains in efficiency and insight. Few report better investment returns, so a board can fairly ask what the spending has produced.
In the same Mercer survey, 69% reported enhanced operational efficiency and 55% faster or higher-quality insights, against 8% who reported a measurable improvement in investment returns.
Source: Mercer
Our part
We work out what each AI use costs and returns, and which are worth keeping.
Some institutional investors are adding questions about AI to their due diligence on managers, so a manager’s answers have to be ready, consistent and checkable.
In AIMA’s 2025 survey, 29% of 18 institutional investors already included generative AI questions in their due diligence questionnaires for fund managers, and a further 29% planned to this year.
Source: AIMA
Our part
We help you assemble, from your own records, the figures and descriptions of AI use that investors ask for, in a form that can be traced and repeated.
Data and compliance concerns are widely cited. Of 131 asset managers surveyed by Mercer, 69% cited data quality or access and 59% regulatory or compliance concerns as material barriers to broader AI adoption. Source: Mercer
Employers want industry-wide standards. Among the employers in the CFA Institute’s 2024 survey of 200 investment industry representatives, 85% saw a need for industry-wide standards for AI in investing and 82% said their absence hinders faster adoption. Source: CFA Institute
Governance varies with size. Half of the respondents at fund managers running under US$1 billion reported no restrictions on their use of generative AI tools, according to AIMA. Source: AIMA
The questions we start with
Could every statement you make to clients and investors about your use of AI be traced to a tool, an owner and a control?
Which outputs of your AI tools could a client read as advice, and who approves them before they reach the client?
Which of the AI uses you pay for today can you show a return on, and in what terms?
Do your records of holdings, accounts and clients agree closely enough to automate reporting?
What restrictions apply to the AI tools your people may use, and who checks that they are followed?
Fintech
Leaders at fintech firms answer to supervisors, customers and partner banks for the money their firms move and hold, often while the customer base grows quickly. By fintech we mean firms that deliver financial services mainly through their own technology. Each study cited here defines the term in its own way, and we say which.
A survey by the World Economic Forum and the Cambridge Centre for Alternative Finance covered 240 fintechs in six retail-facing areas of finance across six regions. Of these, 80% had implemented AI or were implementing it in at least one of the five application areas studied. The firms were invited to take part and the figures are self-reported.
In one global survey, fintechs were further along with AI than traditional financial institutions, apart from generative AI, where traditional institutions reported slightly more.
AI at fintechs and at traditional financial institutions% of respondents in each group; 203 fintechs and 149 traditional financial institutions surveyed worldwide, October 2025 to January 2026
FintechsTraditional financial institutions
At an advanced stage of AI, on the report’s four-stage scaleFintechs: 47%Traditional financial institutions: 30%
AI in software development, at the piloting stage or beyondFintechs: 84%Traditional financial institutions: 72%
AI agents, at the piloting stage or beyondFintechs: 57%Traditional financial institutions: 45%
AI-powered customer support, at the piloting stage or beyondFintechs: 82%Traditional financial institutions: 67%
The Cambridge Centre for Alternative Finance defines fintechs as digital-first firms in the financial sector, and its survey covers firms in banking, insurance, payments and capital markets, among others. The survey was voluntary and self-reported, and the number answering varies by row. Each row keeps its own measure. AI agents are what the report calls agentic AI. Traditional institutions reported slightly more generative AI adoption (75% against 69%), which is not charted.
When AI helps answer customers, the firm answers for what they are told. A leader decides where a person takes over and how each exchange is recorded.
In the Cambridge Centre for Alternative Finance’s 2026 survey, 82% of fintechs reported AI-powered customer support at the piloting stage or beyond, against 67% of traditional financial institutions.
Source: Cambridge Centre for Alternative Finance
Our part
We build automations and AI agents that handle routine customer requests, pass harder ones to your staff and keep a record of each exchange.
Fintechs check new customers, screen them against sanctions lists and monitor transactions, and those checks have to grow with the customer base. In the World Economic Forum survey, fraud detection was the main use of AI in risk management, and 46% of the fintechs reported implementing it.
In October 2024 the UK’s Financial Conduct Authority fined a bank £28,959,426, after a 30% settlement discount, for weaknesses in its financial crime controls. It noted that growth had outpaced the development of those controls. The bank’s customers had grown from 43,000 in 2017 to 3.6 million in 2023.
Source: Financial Conduct Authority (UK)
Our part
We build the reports that show, for each control, how many checks ran, what they flagged and how long review took, for your risk team and your supervisor.
The Bank of Canada says that from 8 September 2025 payment service providers in Canada must have risk management and funds safeguarding frameworks. New rules for UK payment and e-money firms, in force from 7 May 2026, require daily reconciliation, annual audits and a monthly return to the FCA. Firms holding under £100,000 are exempt from the audits.
The FCA reports that for UK payment firms that failed between the first quarter of 2018 and the second quarter of 2023, the average shortfall in customer funds was 65%.
Source: Financial Conduct Authority (UK)
Our part
We help you work out which safeguarding rules apply to you and set up the controls and records that show you meet them.
A fintech that relies on a partner bank or an intermediary to hold customers’ money answers for records that sit in more than one place. Keeping them in agreement depends on consistent data and regular checks.
In August 2025 the US Consumer Financial Protection Bureau alleged that an intermediary between fintech brands and partner banks failed to keep adequate records of where consumers’ funds were. It alleged that the records did not match its partner banks’ records and that the banks held less than the records showed, a shortfall of US$60 million to US$90 million. It also alleged that consumers were left without access to their money for weeks or months.
Source: Consumer Financial Protection Bureau (US)
Our part
We help you choose and oversee the technology providers and intermediaries in your chain, setting out the records and reporting you require from each.
Fintech leaders decide which AI uses are worth their cost and how to show investors and boards what each returns. The published figures are what firms report about themselves.
In the Cambridge Centre for Alternative Finance’s 2026 survey, 56% of 151 fintechs reported increased profitability from AI, against 34% of 80 traditional financial institutions. The report cautions that these are correlations and self-reported.
Source: Cambridge Centre for Alternative Finance
Our part
We work out what each AI use costs and what it returns, and set the measures that let you report it to your board and investors.
Growth can outrun controls. In July 2025 the FCA fined a second UK bank £21,091,300, after a 30% settlement discount, for failings in financial crime controls, saying they had not kept pace with growth in customers and products. Its customers rose from about 600,000 in 2018 to over 5.8 million in 2022. Source: Financial Conduct Authority (UK)
Cost and unclear rules come first. In the World Economic Forum survey of 240 fintechs, 87% said the cost of implementing and maintaining AI is a hurdle, 41% a significant one. Regulatory uncertainty about AI was a hurdle for 77%, 32% a significant one. Source: World Economic Forum and Cambridge Centre for Alternative Finance
Wind-down planning is immature. In the FCA’s review of 14 UK e-money and payments firms in 2024/25, none fully met its expectations for wind-down planning. Almost all of the plans were disconnected from the firm’s risk management framework, and liquidity risk management was immature. Source: Financial Conduct Authority (UK)
Data holds back fewer fintechs than traditional institutions. In the Cambridge report, 40% of the industry overall cited data availability and quality as a barrier to AI: 49% of traditional institutions and 34% of fintechs. Source: Cambridge Centre for Alternative Finance
The questions we start with
If your customer base doubled next year, which of your checks on new customers and transactions would have to be rebuilt, and who would decide?
If a partner bank or intermediary failed tomorrow, could you show every customer’s balance from your own records?
How often do your records of customer money agree with each partner’s, and who reviews the differences?
When an AI tool answers a customer, how does the customer reach a person, and is the exchange kept on record?
Which of your AI uses can you show returning more than they cost, and in what terms?
Real estate
Leaders in real estate answer for what their buildings cost to run, what they are worth and the professional advice their decisions rest on.
The Federal Reserve reports that inflation-adjusted US commercial real estate prices fell about 1.5% in the year to the fourth quarter of 2025. That followed significant declines between mid-2022 and early 2024, and the report describes prices as showing further signs of stabilization. Decisions to buy, hold or sell are made against that record.
Among commercial property professionals, pilots are common, regular use is less so and widespread use is rare.
Where commercial property professionals stand on AIcommercial property respondents surveyed, first quarter of 2026
39%are running early-stage pilots
29%use AI regularly in specific processes
6%report widespread or full use
Each figure is the share of respondents at that stage. More than three-quarters reported some level of use. Respondents are surveyors and other professionals worldwide, so owners and occupiers were not surveyed directly.
Owners and occupiers answer for what each building costs to run and how it performs on energy. AI adjusts building systems to how a building is used and shows where energy is wasted.
In BOMA Canada’s survey of 35 commercial real estate respondents, 70% used AI for energy management and optimization and 30% for predictive maintenance. The sample is small and indicative only.
Source: BOMA Canada
Our part
We build automations and AI agents that watch energy and equipment readings across your buildings and raise what needs attention, with your people deciding what to do.
Investors, lenders and boards rely on the figures behind each property’s value. Commercial property professionals expect AI to matter most in the tools that support those figures.
RICS reports that commercial property respondents expect AI to have the most impact over five years in decision-support tools for valuation, market analysis and portfolio choices.
Source: RICS
Our part
We build the reports and dashboards that bring market, valuation and portfolio figures together for your decision makers.
When valuers and surveyors use AI, the owners and lenders who rely on their advice have a stake in how it was produced. RICS made its standard on the responsible use of AI mandatory for its members and regulated firms from 9 March 2026.
Among commercial property respondents, 22% reported strong or full alignment with the RICS standard on AI, 32% were not currently aligned and 19% were unaware it exists.
Source: RICS
Our part
We help you identify which AI uses fall under the rules that apply to you, and how each is reviewed and recorded.
Most commercial property respondents to the RICS survey have tried AI and few use it widely. Finishing a pilot takes an owner, a budget and a measure of what it returns across the portfolio.
RICS reports that 47% of commercial property respondents plan to increase their AI investment in the next 12 months, while 6% report widespread or full use.
Source: RICS
Our part
We plan which of your pilots deserve to be finished and work out what each would return.
A portfolio can be compared against itself only once the data on its buildings, occupancy, leases and costs is consistent.
Commercial property respondents to the RICS survey named data quality and availability as the barrier to AI they met most often, then integration with existing systems.
Source: RICS
Our part
We analyze in person how your property, leasing and finance teams work and which systems they use, and build a Consier AI solution around that.
Data comes first. Commercial property respondents to the RICS survey named data quality and availability most often, then integration with existing systems, with a lack of skilled people third. Source: RICS
A smaller survey points to skills and return. Of the 35 commercial real estate respondents to BOMA Canada’s survey, 63% named lack of internal expertise, 52% unclear return on investment and 44% integration with existing systems. Source: BOMA Canada
The questions we start with
Which of your pilots has an owner who is accountable for finishing it, and what would finishing it return?
Which valuations or reports you rely on were prepared with the help of AI, and how do you know?
Where is the data on each building held, and could you compare every building in a single view?
Which of your buildings cost the most to run for their size, and what is done about them today?
How we can help
These are problems in the sectors above that AI is being used on today. For each, we set out what published evidence shows and what we would do.
Fraud takes money from customers, and they look to the bank that moved it. The Canadian Anti-Fraud Centre says Canadians reported over 112,000 instances of fraud in 2025, with total losses exceeding C$704 million.
In the UK, the banking industry’s own figures show how much gets past the controls already in place. Criminals stole £1.28 billion through payment fraud in 2025, while the industry prevented £1.68 billion of unauthorized fraud. Losses to scams in which customers are deceived into approving a payment rose 19% to £576.4 million.
What AI is doing about it today
Commonwealth Bank of Australia reports that it uses AI to alert customers to suspicious transactions, with up to 3,000 confirmed fraudulent cases a day. In its year to June 2025 it invested more than A$900 million to help protect customers from fraud, scams, cyber threats and financial crime. It says that investment helped reduce customer fraud and scam losses by over 30% in the year. In February 2026 the bank said AI was used to help cut customer fraud losses by 20% or more in the first half of its 2026 financial year. The comparison is with the same period a year earlier.
The Bank for International Settlements and the Bank of England have tested AI on payment system data. Their Project Hertha used a simulated set of 1.8 million bank accounts and 308 million transactions. In the simulation, using its findings helped banks and payment providers identify 12% more illicit accounts than they otherwise would have. No real customer data was used, and the project reported limits to the approach.
What we would do
Sit with your fraud and payments teams and follow alerts from the first flag to the final decision, measuring what is caught, missed and wrongly blocked.
Compare the detection tools open to you, including those you already pay for, against criteria drawn from your own cases and with no preference for any supplier.
Build automations and AI agents that gather the evidence behind each alert for your analysts, who make every decision.
Give each tool an owner, a stated purpose and a record of what it flagged and why, ready to show a supervisor.
Build the monthly reports that show your executives and board the losses, alerts, wrongly blocked payments and time to decision.
What you would have
You would have a measured account of how fraud alerts are handled today and a documented comparison of the tools open to you. You would also have automations your analysts use on every alert, a record of each tool to show a supervisor and a monthly report for your board.
The UK industry’s own figures show costs growing faster than the revenue that pays for them. In the Investment Association’s annual survey, revenues have grown by 15% since 2019 while operating costs have risen by 36%. Operating margins fell to 18% in 2024, from 31% in 2019.
The same survey reports that direct employment fell for a second consecutive year in 2024, while assets managed in the UK rose to £10.0 trillion. The association reads that gap as firms turning to technology, outsourcing and efficiency measures to expand capacity.
What AI is doing about it today
Norges Bank Investment Management manages the Government Pension Fund Global. At the end of 2025 it had 678 employees, and it reports that all of them now use AI tools in their daily work. In an internal survey, its employees reported an average productivity increase of 20% when using those tools.
Its annual report describes the program behind that. During 2025 the manager ran mandatory training, workshops and technology days at all its offices, and set up a network of 71 AI ambassadors. Over the year it moved from experiment and building skills to developing the solutions of greatest value to the organization. The 20% is what employees reported, and the report gives no measured saving in hours or cost.
Growth in the revenue and operating costs of the UK investment management industry since 2019% growth, 2019 to 2024
Revenues15%
Operating costs36%
Over the same period operating margins fell from 31% to 18%. The figures are the Investment Association’s estimates for the UK industry.
Spend time with your investment, operations, client service and compliance teams to see where their hours go and which tasks suit AI tools.
Train every team in the tools you approve, agree what each tool may be used for and prepare some of your own people to support their colleagues.
Build automations and AI agents for the routine work that takes the most hours, with a person approving anything that reaches a client.
Measure the time and cost of that work before and after, so that what the tools return is counted and can be reported to your board.
What you would have
You would have teams trained in the tools you have approved, with written rules for what each tool may be used for. You would also have automations running in the work that takes the most hours and a before-and-after measure of time and cost for your board.
Every new customer brings identity checks and payments to monitor, and the firm answers for each alert those checks raise. Wise reports that around a third of its global team works on preventing financial crime, keeping customers safe and meeting the requirements of over 70 regulatory licences. At the end of March 2025 it had over 6,500 employees, and in the year to that date it moved money across borders for 15.6 million people and businesses.
In April 2025 New York’s Department of Financial Services fined a US company that runs a money transfer service US$40 million for compliance failings that included its anti-money laundering program. The department found that alerts awaiting review grew from approximately 18,000 in 2018 to over 169,000 by 2020. In part, the company could not predict how its growing customer base would affect alert volumes and staffing. Between February 2021 and September 2022, reports of suspicious activity were at times filed more than a year after the alert. The order required the company to engage, for 12 months, an independent monitor chosen by the department. Three months earlier, 48 state regulators had fined the same company US$80 million over its anti-money laundering compliance.
What AI is doing about it today
A large bank has published what AI changed in this work. HSBC’s group head of financial crime wrote in June 2024 that the bank checks about 980 million transactions a month for signs of financial crime. She reported that with an AI system first piloted in 2021, the bank finds two to four times more financial crime than before, with much greater accuracy.
The bank also reported 60% fewer false positives. Before, it had been calling customers unnecessarily about activity that proved legitimate. The time needed to analyze billions of transactions across millions of accounts fell from several weeks to a few days. These are the bank’s own figures. In its order against the money transfer company, the New York department credited the company’s efforts to resolve its backlogs, add compliance staff and limit how many accounts one customer can create.
Transaction monitoring alerts awaiting review at one US money transfer companyalerts in the backlog, as found by the New York State Department of Financial Services
2018approximately 18,000
2020over 169,000
From the department’s consent order of April 2025. It attributes the growth in part to the company’s inability to predict how its growing customer base would affect alert volumes and staffing, and in part to new monitoring tools that raised more alerts.
Measure with your financial crime team how many alerts each check raises, how long each waits for review and how many end in a report.
Build a forecast that turns your growth plans into the alerts they will raise and the reviewers needed to clear them within your own deadlines.
Test each alert rule against your past cases to find those that mostly flag legitimate activity, and agree any change with your compliance officer first.
Build automations and AI agents that gather each customer’s history and earlier alerts for your analysts, who decide every case and every report.
Report each week on the alerts waiting, their age against your deadline and the reviewers available, for your board, your partner banks and your supervisor.
What you would have
You would have a measure of how your alerts are handled today and a forecast linking your growth plans to alert volumes and the reviewers they require. You would also have a documented review of your alert rules, automations your analysts use on every alert and a weekly report of what is waiting and for how long.
Energy is a cost every owner and occupier pays, building by building. The US Energy Information Administration estimates that the 5.9 million commercial buildings in the US spent US$141 billion on energy in 2018.
For a portfolio, the question is where that money is wasted and which buildings would repay new technology.
What AI is doing about it today
The US General Services Administration, which manages many federal buildings, tested an AI control system with the National Renewable Energy Laboratory at four of them. The system brought each building’s data into one view and used AI to adjust heating, cooling and ventilation to weather and occupancy. Modelling showed savings of between 5% and 11% of whole-building energy. The savings were modelled because the pandemic had changed how the buildings were occupied and run.
The agency also found that the return depended on the building. At the test sites the yearly subscription cost more than the energy saved, because local energy prices were low. The laboratory also analyzed 504 of the agency’s buildings. With a 10% saving on heating, ventilation and cooling, 64% of them would save more than the system cost. With a 5% saving, 18% would.
Federal buildings where the tested system would pay for itself% of 504 buildings owned by the US General Services Administration and subject to energy measures, by the saving assumed
64%would save more than the system costs with a 10% saving on heating, ventilation and cooling
18%would do so with a 5% saving
A market analysis by the National Renewable Energy Laboratory, assuming a yearly subscription of US$0.10 per square foot and blended energy prices. These are estimates. The savings in the field test were modelled.
Visit your buildings with the people who run them, and bring energy, occupancy and equipment data for the whole portfolio into one view.
Estimate building by building what AI control would save and what it would cost, using each building’s own energy prices.
Rank the buildings by expected return and flag those where the data points to faults that should be repaired first.
Compare the systems on offer against criteria drawn from your buildings, with no preference for any supplier, and help you choose.
Set up automations that raise what needs attention to your building teams, and report measured savings against the estimate.
What you would have
You would have one view of energy use and cost across the portfolio, with an estimate for each building of what AI control would save and cost. You would also have a ranked list of where to start, a documented choice of system and regular reports of measured savings against the estimate.
The results described here were published by the organizations named. These are examples of how we would approach the work, and none describes an engagement of our own.
Researchers at the University of Cambridge surveyed fintechs and traditional financial institutions worldwide on how far along they are with AI and what it has returned.
Canada’s central bank sets out what payment providers have had to put in place since September 2025, including risk management and the safeguarding of customers’ funds.
A briefing is a confidential conversation with the executive accountable for a technology decision. We set out how we would approach the question and what evidence it would require. Scope and fees are agreed in writing before any work begins.
