How AI does real work, one use case at a time

A survey of 4,050 sales professionals found the average seller spends about 40% of their time actually selling, and 51% of sales leaders using AI say disconnected systems are slowing their AI work.

Salesforce, State of Sales report, 2026 edition (opens in a new tab) · 2026

Across 5,179 customer support agents, an AI assistant raised issues resolved per hour by 14% on average and by 34% for novice and lower-skilled agents.

NBER, Generative AI at Work (Brynjolfsson, Li and Raymond) (opens in a new tab) · 2023

In a study of 453 professionals doing writing tasks from their own jobs, access to ChatGPT cut the time taken by 40% and raised quality, as rated by evaluators, by 18%.

Science / MIT, Experimental evidence on the productivity effects of generative AI (Noy and Zhang) (opens in a new tab) · 2023

In a field experiment with 758 BCG consultants, those using GPT-4 on tasks within its abilities finished 12.2% more tasks, 25.1% faster and at over 40% higher quality, but did worse on a task outside its abilities.

Harvard Business School / BCG, Navigating the Jagged Technological Frontier (Dell'Acqua et al.) (opens in a new tab) · 2023

Business email compromise caused reported losses of about $2.77 billion across 21,442 complaints in 2024, out of $16.6 billion in total reported cybercrime losses.

FBI Internet Crime Complaint Center, 2024 Internet Crime Report (opens in a new tab) · 2025

Third-party involvement in breaches doubled to 30%.

Verizon, 2025 Data Breach Investigations Report (opens in a new tab) · 2025

Across randomised trials, attendance was 78.6% with text reminders against 67.8% with none, and text reminders worked about as well as phone calls at lower cost per attendance.

Cochrane, Mobile phone messaging reminders for attendance at healthcare appointments (Gurol-Urganci et al.) (opens in a new tab) · 2013

Gartner predicted at least 30% of generative AI projects would be abandoned after proof of concept, citing poor data quality, inadequate risk controls, escalating costs or unclear business value.

Gartner, Gartner predicts 30% of generative AI projects will be abandoned after proof of concept by end of 2025 (opens in a new tab) · 2024

About 75% of generative AI's potential value sits in four areas: customer operations, marketing and sales, software engineering, and R&D.

McKinsey & Company, The economic potential of generative AI: the next productivity frontier (opens in a new tab) · 2023

In a survey of 31,000 people in 31 markets, 75% of knowledge workers said they use AI at work, and 78% of AI users bring their own AI tools.

Microsoft and LinkedIn, 2024 Work Trend Index: AI at work is here. Now comes the hard part (opens in a new tab) · 2024

62% of people said they spend too much time searching for information, 68% lack enough uninterrupted focus time, and 55% said next steps at the end of meetings are unclear.

Microsoft, 2023 Work Trend Index: Will AI fix work? (opens in a new tab) · 2023

US retailers expected 16.9% of 2024 sales to be returned, about $890 billion, and 93% said return fraud and abuse is a significant issue.

National Retail Federation and Happy Returns, 2024 Consumer Returns in the Retail Industry (opens in a new tab) · 2024

The average business loses almost 9% of contract value each year through poor contract management; the best lose about 3%, the worst 15% or more.

World Commerce & Contracting, Contract management: an overlooked driver of business agility and financial performance (opens in a new tab) · 2025

A US federal judge fined two lawyers and their firm $5,000 for filing a brief that cited cases ChatGPT had invented.

ABA Journal, Lawyers who defended ChatGPT's fake cases must pay $5K, judge says (opens in a new tab) · 2023

Google rewards original, high-quality content however it is produced, but using automation mainly to manipulate rankings breaks its spam policies.

Google Search Central, Google Search's guidance about AI-generated content (opens in a new tab) · 2023

The FTC's 2024 rule bans fake reviews, including AI-generated ones, undisclosed reviews by company insiders and threats used to suppress negative reviews.

US Federal Trade Commission, Final rule banning fake reviews and testimonials (opens in a new tab) · 2024

Calls using AI-generated voices count as artificial voice calls under US telephone consumer law, so they need the person's prior express consent.

US Federal Communications Commission, FCC makes AI-generated voices in robocalls illegal (opens in a new tab) · 2024

From August 2026, AI systems that talk directly to people in the EU must make clear they are AI, unless that is obvious.

EU AI Act, Article 50: transparency obligations (opens in a new tab) · 2024

AI used to recruit or select people, or to make decisions about promotion, termination, task allocation or performance, is classed as high-risk.

EU AI Act, Annex III: high-risk AI systems (opens in a new tab) · 2024

Employers in New York City may only use automated hiring tools that had a bias audit within the past year, must publish a summary of it and must notify candidates.

NYC Department of Consumer and Worker Protection, Automated employment decision tools (Local Law 144) (opens in a new tab) · 2023

NIST's framework organises AI risk work into four functions, Govern, Map, Measure and Manage, with a profile for generative AI added in July 2024.

NIST, AI Risk Management Framework 1.0 and Generative AI Profile (opens in a new tab) · 2023

In 1,921 fraud cases, expense reimbursement schemes made up 13% of cases with a median loss of $50,000, and frauds at organisations with under 100 staff had a median loss of $141,000.

Association of Certified Fraud Examiners, Occupational Fraud 2024: A Report to the Nations (opens in a new tab) · 2024

Only 12% of employees strongly agree their organisation does a great job of onboarding new employees.

Gallup, Why the onboarding experience is key for retention (opens in a new tab) · 2018

Organisations waste an average of $21 million a year on unused SaaS licences, and SaaS spend averages $4,830 per employee.

Zylo, 2025 SaaS Management Index (opens in a new tab) · 2025

Gartner research put the average cost of poor data quality at $12.9 million a year per organisation.

Gartner, Data quality: why it matters and how to achieve it (opens in a new tab) · 2020

Knowledge workers spend roughly a fifth of their working week searching for and gathering information.

McKinsey Global Institute, Capturing business value with social technologies (opens in a new tab) · 2012

The global average cost of a data breach reached $4.99 million, a record high.

IBM, Cost of a Data Breach Report (opens in a new tab) · 2026

Observed physicians spent 27% of their office day on direct time with patients and 49.2% on the electronic record and desk work, plus 1 to 2 hours of after-hours work each night.

Annals of Internal Medicine, Allocation of physician time in ambulatory practice (Sinsky et al.) (opens in a new tab) · 2016

In the first 10 weeks of a rollout, 3,442 physicians used ambient AI scribes in up to 303,266 patient encounters; use was linked with less time on documentation, though accuracy still needs a clinician's review.

NEJM Catalyst / Kaiser Permanente, Ambient artificial intelligence scribes to alleviate the burden of clinical documentation (Tierney et al.) (opens in a new tab) · 2024

In a survey of 1,000 physicians, practices averaged 43 prior authorisations per physician per week and 12 hours of physician and staff time; 94% said it delays access to care.

American Medical Association, Exhausted by prior auth, many patients abandon care: AMA survey (opens in a new tab) · 2024

43.2% of US physicians reported at least one symptom of burnout in 2024, down from 48.2% in 2023 and 53% in 2022.

American Medical Association, U.S. physician burnout hits lowest rate since COVID-19 (opens in a new tab) · 2025

In a pilot with 162 clinicians at Stanford, AI drafts were used for 20% of replies; task load scores fell, but time spent on replies did not change.

JAMA Network Open, Artificial intelligence-generated draft replies to patient inbox messages (Garcia et al.) (opens in a new tab) · 2024

Under FDA guidance, software meant to support time-critical clinical decisions does not meet the definition of non-device clinical decision support.

US Food and Drug Administration, Clinical decision support software: frequently asked questions (opens in a new tab) · 2026

HealthCare.gov insurers denied 20% of in-network claims in 2023, and fewer than 1% of denied claims were appealed.

KFF, Claims denials and appeals in ACA Marketplace plans in 2023 (opens in a new tab) · 2025

Asked to produce billing codes from descriptions, GPT-4 matched the correct code exactly only 33.9% of the time for ICD-10-CM and 49.8% for CPT.

NEJM AI, Large language models are poor medical coders (Soroush et al.) (opens in a new tab) · 2024

Across 19 studies, between 6.8% and 62% of laboratory tests and between 1.0% and 35.7% of radiology tests were not followed up, with harms including missed cancer diagnoses.

Journal of General Internal Medicine, Failure to follow-up test results for ambulatory patients: a systematic review (Callen et al.) (opens in a new tab) · 2012

The global nursing shortage was 5.8 million in 2023 and is projected to be 4.1 million in 2030.

World Health Organization, State of the World's Nursing 2025 (opens in a new tab) · 2025

The global cost of medication errors has been estimated at US$42 billion a year.

World Health Organization, Medication Without Harm (opens in a new tab) · 2017

Around 1 in 10 patients is harmed in health care, and more than half of that harm is preventable.

World Health Organization, Patient safety fact sheet (opens in a new tab) · 2023

Moving US healthcare admin transactions from manual to automated could save $20 billion a year; a prior authorisation takes providers about 24 minutes by phone, fax or email and 16 minutes through a portal.

CAQH, 2024 CAQH Index (opens in a new tab) · 2025

Only 12% of US adults had proficient health literacy; 36% had basic or below basic health literacy.

US National Center for Education Statistics, The health literacy of America's adults: results from the 2003 National Assessment of Adult Literacy (opens in a new tab) · 2006