Feel Behind on AI? Here Are 8 Practical Ways to Catch Up in 2025

You probably suspect AI can dramatically improve your product – and you’re right. 

Surveys show that AI adoption has surged worldwide (from 55% of companies using AI in at least one function in 2023 to about 72% in 2024​) and nearly all companies plan to increase AI investments in the next few years [1, 2]​. With this rapid growth, business leaders are asking anew: “How do we spot the right AI opportunities in our product?”

Finding great AI opportunities in 2025 is about cutting through the hype to identify projects that truly add value, those that solve high-impact user problems AND are feasible with today’s AI capabilities. This updated guide will walk you through the latest AI market trends, illustrate how generative AI is driving innovation, and outline 7 practical steps to pinpointing AI initiatives that can completely transform your product and user experience.

1. Understand the AI Landscape in 2025

AI has rapidly moved from niche to necessity. Business adoption of AI hit record levels in 2024, and investment is following suit. What does the current AI market actually look like? 

Widespread Adoption 

According to PwC's October 2024 Pulse Survey, one-third of technology leaders reported that AI was fully integrated into their companies' products and services [3].  Nearly 92% of companies plan to increase AI investments over the next three years [2]​. Across regions, Asia-Pacific firms are racing ahead – APAC is now second only to North America in generative AI adoption​, with over 90% of APAC companies planning to scale up AI use in the next two years​. Simply put, AI has gone global and mainstream [4].

ROI is Real (for Those Who Commit)

The good news for AI adopters: Companies that embed AI into their products are seeing significant returns. AI-powered personalization tools, for example, increased conversion rates by 3.6x for No7 and boosted average order values by 48% for JCPenney’s AI beauty advisor [6]. Notably, one-third of technology leaders report that AI is fully integrated into their products and services, showing a clear commitment to AI-driven innovation [7]. As businesses scale AI-powered features and learn how to deploy them effectively, return on investment continues to increase over time. 

Scaling Challenges Persist 

Despite broad adoption and optimistic spending, most companies still struggle to capture full value at scale. A late-2024 study found 74% of companies have yet to see tangible value at scale from AI projects [8]​. McKinsey similarly reports that while almost all firms are experimenting, only 1% feel they have reached “AI maturity, where AI is driving significant impact​. Many initiatives stall after initial pilots [2]. 

The takeaway: Piloting AI is easy, but 1) integrating it into your product and 2) aligning it to strategy remains difficult. 

This widespread challenge of course begs the question: What are successful AI adopters doing differently? The short answer is they start by clearly aligning AI with their most pressing, high-value user problems.

2. Align AI with High-Value Product Enhancement Opportunities

Research shows “AI leader” companies tend to focus on fewer, high-impact use cases rather than chasing dozens of experiments​. They align AI projects tightly with high-value user goals and expect significantly higher ROI – for example, top AI performers anticipated 60% higher AI-driven revenue growth and nearly 50% greater cost reductions by 2027 compared to others [10]​.

This highlights an important point: spotting great AI opportunities isn’t just about technology – it’s about focusing on impactful problems and preparing your organization to leverage AI-powered product features [8].

In sum, the 2025 landscape is one of high AI enthusiasm tempered by execution realities. AI is everywhere, and it’s delivering ROI for many, but sustained success requires choosing projects wisely and planning for scale. Keep this context in mind as we turn to how generative AI is turbocharging said possibilities, and how to zero in on the best possible opportunities for your business.

3. Leverage Generative AI to Innovate and Disrupt

One of the biggest shifts since 2022 has been the rise of Generative AI, AI systems that create content like text, images, code, designs, etc. similar to human output. Powered by LLMs and other foundation models, Gen AI has captured the world’s attention and is already transforming products and services in some pretty profound ways:

Fig 1: Generative AI enterprise spending by category, 2023 vs 2024. Enterprise investment in generative AI surged more than sixfold from 2023 to 2024, jumping from $2.3B to $13.8B​. The chart above shows how spending on foundation models and related tools skyrocketed as companies moved from pilot projects to full production deployments of generative AI features in their products [11].

New Product Features

The most successful companies in 2025 are embedding generative AI directly into their products, creating features that dramatically enhance user value. As an example, pharmaceutical firms are using AI models to generate potential drug molecules, dramatically speeding up R&D. It’s predicted that sometime in 2025, over 30% of new drugs and materials could be discovered using generative techniques​ [12].

Adobe’s Firefly, integrated across their creative cloud suite, allows users to generate and modify images through text prompts and transforming how designers create graphics. IBM's recent digital marketing campaign leveraged Adobe Firefly's generative AI capabilities to create compelling content, resulting in a 26-fold increase in engagement compared to previous benchmarks, including 20% of that audience identified as C-level decision-makers [13].

Similarly in the healthcare space, Nuance’s Dragon Ambient eXperience (DAX) uses AI to automatically document patient encounters, creating clinical notes from natural conversations between doctors and patients. This product feature has reduced physician documentation time by 50%, with 79% of physicians reporting that DAX improved the quality of their clinical notes. For healthcare systems facing provider burnout, this AI-powered feature directly addresses a critical pain point [14].

Customer Experience Transformation

Companies are completely revolutionizing their customer’s experiences by embedding AI directly into their user interfaces. It was found that those excelling at this personalization generated 40% more revenue compared to the average players [15]. Across domains, these AI co-pilots are acting as force multipliers for human productivity [16].

One example is Spotify’s AI-powered Discover Weekly, Daylist and Release Radar features that create hyper-personalized playlists. Spotify analyzes 1) listening history and 2) subtle patterns in user preferences, time-of-day habits, and even the audio characteristics of their favorite tracks. This deep level of personalization creates a continuous discovery experience that keeps their users engaged and has contributed to Spotify’s massive growth [17]. In 2024, Spotify reported an 11% increase in monthly active users, reaching 640 million, and a 12% growth in subscribers, totaling 252 million.

Competitive Disruption

While many established companies are already enhancing their products with AI features, those who fail to do so face significant disruption. A cautionary case is in online education and Q&A services – when ChatGPT went viral, it began supplanting platforms like Chegg (known for textbook solutions and tutoring). Chegg’s stock plummeted as students turned to AI answers, erasing 85% of the company’s market cap in 2023 [11]​. Similarly, the legal research industry was transformed when LexisNexis introduced Lexis+ AI, featuring context-aware legal research that understands complex legal questions in natural language and generates comprehensive answers with relevant case citations. This AI-powered feature reduced research time for associates by 65%, threatening traditional legal research providers who hadn't evolved beyond keyword search​.

These examples underline that generative AI is no longer just an internal efficiency tool, it’s a crucial component of customer-facing product features. Businesses need to proactively consider how AI might disrupt their model – or how they can use AI to disrupt competitors.

The winners in 2025 are often those who embed AI capabilities directly into their products and services, creating features that solve real customer problems.

4. Conduct a Data Asset Inventory

The first real step in spotting AI opportunities is to take stock of what data you already have. AI feeds on data so you need to know what customer information your business collects, and how it’s stored. Ask yourself and your team: What are our primary data sources?

For example:

  1. Do you have large collections of text (customer emails, support tickets, documents), images (product photos, medical scans, satellite imagery), video or audio (call center recordings, videos of operations)? 

  2. Do you have unstructured data (customer-generated social media posts, user-uploaded videos, handwritten feedback forms, recordings from customer-service interactions), or structured data (transaction records, sensor readings)? 

  3. What does this data describe? Customers, products, operations, the external environment? 

  4. And perhaps most critically, where is it stored and how accessible is it? Perhaps sales data sits in one database, production data in another, etc etc.

If your data is unstructured (images, free-form text, etc.), also consider what metadata exists (e.g. timestamps, authors, locations) that could help give it context. If it’s structured, assess whether data is siloed across various departments. In far too many organizations, valuable data is scattered across different systems. For AI to be effective, you may need to integrate or consolidate these data sources.

Understanding your data assets in 2025 also means recognizing that you don’t have to start from scratch. There are many pre-trained AI models (especially generative models and vision models) that you can adapt to your data. But you still need quality data to fine-tune models to your specific needs. 

The bottom line: audit your data – its volume, its variety, and its accessibility.

This will reveal what kinds of AI problems you can tackle, and what groundwork (like data cleaning or combining datasets) is needed to make your data AI-ready.

Companies often discover underused data that can be a goldmine for AI. For instance, a medical device company might realize it has years’ worth of imaging scan data that could power an AI-powered diagnostic assistant, helping doctors detect early signs of disease with greater accuracy. Or a retailer might have customer purchase histories that could train a recommendation engine. Take inventory of what you have – AI opportunities often hide in plain sight in your data repositories.

5. Gather Intuition from Across Your Organization

Most business leaders and domain experts have a gut feeling about how AI-powered features could enhance their products or services. Tap into this gut feeling. Talk to people across your organization about what the features they envision if certain data were harnessed intelligently. 

Consider how Waze transformed navigation by adding AI-powered traffic prediction features that analyze real-time user reports, historical patterns, and sensor data to suggest optimal routes [18]. What began as an engineer’s intuition that “maybe we could use crowdsourced data to predict traffic jams” evolved into a feature so valuable that Google acquired the company for 1.15B. The product features didn’t simply tell users where to go, rather it told them both when to leave and which route to take, creating a competitive advantage that traditional navigation apps like Google Maps couldn’t match [19]. 

Even a physician might realize: “I often struggle to classify a complex skin lesion correctly; perhaps an AI trained on thousands of images could assist.” This intuition reflects the explosion of AI in healthcare diagnostics. Today, there are over 1,000 FDA-authorized AI-enabled medical devices, many focused on medical imaging, detecting patterns that clinicians might miss [20]​. AI dermatology tools, for instance, can analyze skin images and help identify conditions with impressive accuracy, augmenting the physician’s own expertise.

Take time to document these hypotheses about what features AI could enhance or enable in your products. They don’t need to be perfectly formulated. The goal is to capture ideas – i.e., “predict workout recommendations based on fitness tracking data” or “personalize product recommendations based on past purchases” or “real-time translation of customer support 

Pro Tip: A useful rule of thumb during brainstorming is to look for features where users 1) currently make divisions with limited information or 2) perform repetitive tasks within your product. That’s often a hint that AI could step in. If a student using a tutoring software struggles to understand WHY their answers were incorrect on a practice test, that could be a great opportunity for an AI-powered answer walkthrough. 

Identify processes in your product or application where users are looking for patterns or following “if-then” rules within your products. These are prime candidates for AI-powered features. Some examples might include: 

  • Pattern recognition in visual content. Consider features that could detect and classify objects in product photos, surveillance footage, or user-uploaded images. 

  • Text analysis and information extraction. Look for opportunities where users currently read through documents, emails, or other text content to extract specific information.

  • Cross-platform data reconciliation. Identify features where users manually compare information across different sources or systems.

  • Decision rule automation. Areas where users follow consistent decision criteria or filtering rules. 

When you find these scenarios, envision how an AI might handle them. AI excels at pattern recognition at scale. For instance, Ring’s Smart Alerts features distinguish between people, packages, and animals in security footage, eliminating the need for users to manually review irrelevant videos [21]. In text, AI (especially modern NLP and LLMs) can summarize documents, classify them by topic or sentiment, or pull out entities (names, dates, product codes) far more quickly than any human reader.

A good exercise is to ask your product team: What repetitive decision processes do our users perform when using our product? Where are they manually analyzing information that could be automated? Those are prime candidates. By zeroing in on these pattern-based tasks, you’re aligning with AI’s sweet spot – crunching large data volumes for insights – and also targeting improvements that will make your user’s lives easier (nobody loves mind-numbing manual work!). This forms a practical filter for your list of AI opportunity ideas.

Ultimately, AI should enhance your products with features that solve real user problems, not just implement cool tech. So ask: What capabilities would delight our users if we could add them to our product? These could be challenges that, if solved, would significantly move the needle on revenue, customer satisfaction, or competitive position. The most compelling AI-powered features often address frustrations users have silently accepted as unavoidable limitations. 

It often helps to list out your top product pain points or feature opportunities in a simple way. For example:

Write down a set of these “wish statements.” Don’t worry yet about whether AI can enable them – we’ll get to that next. The point here is to articulate the user value clearly. These are essentially your high-value product enhancement opportunities.

Why do this? Because AI-powered features are not a goal in itself; they’re a means to an end. By anchoring on tangible product improvements, you ensure that any AI feature you develop directly enhances the user experience. It also helps communicate to stakeholders why you’re considering an AI approach. To reduce churn by matching customers to the right plan” is a compelling rationale that anyone in the business can get behind.

Once you have this list, you can start to cross-pollinate it with the earlier work: see which of these high-value product opportunities might be addressed by the data and patterns you identified. For instance, the content filtering challenge could pair with an AI pattern-recognition approach (automated content moderation feature). Users choosing wrong subscription plans might pair with a recommendation system feature (personalized plan matcher). This cross-linking starts turning abstract product challenges into concrete AI-powered feature concepts.

Keep your list of top product opportunities visible – it's your guardrail to ensure AI feature development remains grounded in genuine user value rather than technology for technology's sake.

6. Brainstorm Specific AI-Powered Features that Tackle Your High-Value Problems

This is where we translate user needs into AI project concepts. It can feel a bit awkward if you’re not used to thinking in terms of AI solutions, so let’s break it down with some basics and examples.

First, remember that most AI solutions can be categorized into a few broad types (these are your AI toolkit):

Predictive analytics (supervised learning): You have examples of past outcomes and you want to predict future ones. This includes things like predicting which customers will churn, which transactions are fraudulent, what demand will be next month, etc. You train on historical data (with known outcomes) and then predict on new data. 

  • Use case: "Predict when users will need to reorder supplies based on their consumption patterns" → supervised learning that analyzes historical usage data to proactively suggest reordering before users run out.

Personalization features (unsupervised learning). Here, you let the AI find patterns in user behavior data without explicit labels. This can surface natural groupings or anomalies that help tailor the user experience. This includes features that adapt content, interfaces, or recommendations based on the observed user behavior, or their characteristics. 

  • Use case: "Customize learning paths based on a student's performance and learning style" → AI that analyzes how users interact with educational content to create personalized curriculum progressions.

Content generation features (generative AI): A newer category (related to unsupervised+pretraining) is generative AI, where the system learns the patterns of your data and then creates new data or content on demand. This could be generating text (product descriptions, reports), images (designs, layouts), or even synthetic data to augment training. 

  • Use case: “Automatically draft responses to common customer support questions” → an LLM fine-tuned on your company’s support tickets could do this, saving agent time.

Decision automation (rule-based AI or reinforcement learning): This is AI that takes actions or recommends actions based on learned policies. This might be more advanced, but think of recommendation engines (deciding which product to show a user) or dynamic pricing algorithms that adjust prices based on market conditions. These often combine prediction with business rules.

  • Use case: "Recommend optimal times to publish content based on audience engagement patterns" → AI that analyzes historical performance data to suggest when users should schedule posts for maximum reach.

With these in mind, revisit each high-value user problem we identified and ask: “What kind of feature, if it existed, would solve this? Could AI provide that?” Don’t worry about feasibility quite yet, just ideate possible AI approaches.

At this stage, generate a list of AI project ideas corresponding to one or more user problems. It’s okay if one AI idea addresses multiple problems, or vice versa. The goal is a brainstormed set of AI opportunity candidates. For each, you should be able to say in one sentence what the AI would do (e.g., “an AI that [action] in order to [solve X]”). These are your candidates to evaluate.

Before we evaluate and rank these ideas, note that it’s normal to have more ideas than you’ll actually implement. AI leaders actually pursue fewer projects on purpose, focusing their efforts on the most promising ones [8]​. So, a big part of spotting great AI opportunities is next figuring out which ideas are the winners. That’s where feasibility & impact ranking comes in.

7. Stack Rank Your AI Opportunities by Feasibility and Impact

Now you have a shortlist of exciting AI project ideas, and it’s time to stack rank these opportunities to decide which ones to pursue first. You want to prioritize projects that are both high value and highly feasible. We already gauged value when linking to user problems; now we need to assess feasibility and ROI potential.

Key factors to consider for each AI opportunity candidate include:

Technical & Data Feasibility 

Can AI currently deliver the performance needed for this feature to provide real value? This requires an honest assessment of what today's AI can reliably do versus where it struggles. Computer vision can identify objects in images with remarkable accuracy, but may perform inconsistently in poor lighting conditions. Natural language understanding excels at sentiment analysis and classification but may still struggle with complex reasoning or nuanced context.

When evaluating feasibility, compare your feature requirements against established AI capabilities. If your feature needs AI to interpret subtle emotional cues in text with high accuracy, current NLP models might be sufficient. If it requires understanding the implicit cultural context behind user comments, that might push current technical boundaries. So ask yourself:

  • Can AI actually do what’s needed here, at least to a useful degree? 

Next, do you have access to the data required to train the AI models powering your features? This is often the most overlooked yet most critical feasibility factor. A brilliant AI feature concept is worth less if you lack the data required to make it work. Assess not just whether you have the data, but whether it’s:

  • Sufficient in quantity (enough examples for the AI to learn patterns)

  • High in quality (accurate, consistent, & representative)

  • Properly structured (organized in a way machines can learn from) 

  • Ethically obtained (with proper permissions and privacy considerations)

  • Diverse and unbiased (represents all user groups fairly)

In 2025, AI is powerful but not magic. If you're missing crucial data, consider whether you could implement a simpler version of the feature that would help collect the necessary data over time. Many successful AI products started with "training wheels" features that gradually improved as user interactions generated more training data.

And be realistic by leveraging expert input or case studies from others. You might rate each idea as High/Medium/Low feasibility based on data availability and technical difficulty.

Expected Impact (Value)

Roughly estimate how big the benefit could be for users if it works as intended.

  • Would this feature dramatically improve core user workflowers, or just add a nice-to-have capability?

  • Is this addressing a pain point mentioned in 5% of customer feedback, or 50%?

  • Would this feature theoretically increase user engagement by a small margin, or could it totally transform the user experience?

High-impact + high-feasibility ideas obviously move to the top of the list.

Resource Requirements

Consider the time, budget, and talent needed to implement each feature. Some AI projects might be quick wins using off-the-shelf models; others might require hiring data scientists or a long R&D effort.

Do you have engineers who can integrate AI into your product's technology stack? Are there annotators on hand to label the data required to train AI models? If internal resources aren’t currently sufficient, that’s an important factor in your feasibility assessment.

In 2025, smart companies often choose to partner with external AI specialists rather than build everything in-house. Instead of investing substantial time and capital into hiring a dedicated internal data science team, collaborating with an AI consultancy allows you to leverage specialized expertise right away.

Of course, working with external specialists requires budgeting for consulting fees or licensing costs, which must be accounted for in your resource evaluation. However, for most organizations—especially those without extensive internal AI experience—budgeting for external expertise is often the most practical, fastest, and lowest-risk path to making an AI vision a reality.

Timeline to Value

How long before users can start benefitting from this feature? In today’s competitive landscape, delivering value quickly is of the utmost importance. Projects that can show a prototype or ROI in months might be more attractive than multi-year moonshots (unless the payoff is truly massive).

To make this ranking more concrete, you can create a simple matrix or scorecard for each idea. For example:

This is just illustrative, but it shows how you might start ranking. Perhaps the predictive maintenance and plan recommendation emerge as top picks (high impact, feasible, data-ready). The AI coaching might be high impact but also high effort and less certain feasibility (could be a longer-term bet). Content moderation might be very feasible but maybe moderate in impact (unless content quality is core to your business).

By stack ranking, you intentionally choose to focus on the most promising opportunity first. Remember, successful AI adopters often pursue fewer projects at a time, and do those really well [8]​. It’s better to nail one great AI implementation than to juggle 10 mediocre pilots.

Next, double-click on feasibility: evaluate what level of accuracy or performance you realistically need from the AI and whether that seems attainable. This is sometimes called achievable efficacy. For instance, if your "similar product recommendation" feature achieves 80% relevance, is that good enough for users to find it helpful? Or do you need 95%+ to avoid customer frustration? Setting rough targets helps, as some use cases don’t require perfection (an AI that identifies 80% of relevant items might still save users substantial browsing time), while others (say, autonomous driving) need EXTREMELY high accuracy.

Often, the fastest way to gauge efficacy is to build a quick prototype or proof-of-concept model. For a top idea, it’s usually worth spending a few weeks to train a preliminary model and see if the predictions or outputs look remotely sensible. If they do, great!, that de-risks feasibility. If not, you might discover the problem–maybe the data is too noisy, or the objective needs tweaking–and can either adjust approach or reconsider the idea.

Also consider the type of AI technology needed: Is it standard (classification, regression, etc.) or cutting-edge (maybe requiring the latest NLP)? If it’s something like “we need AI to understand nuanced human emotions or complex reasoning” you might be pushing current tech limits. But many user problems (recommend related items, identify patterns in images, detect anomalies in user behavior) are well within what today’s AI can do when applied properly.

Prioritize and Plan

After considering feasibility, impact, and resources, you should be able to rank your list from most promising opportunity to least. The top one or two are your likely candidates to pursue first. It’s often wise to start with a project that is feasible enough to show wins quickly and valuable enough to matter – this builds momentum and executive buy-in for AI.

Once ranked, you can make the case for those top opportunities with clear reasoning: e.g. “We recommend starting with predictive maintenance AI because we have the data, proven models exist (feasible), and it could save us $5M/year in downtime (impact), with a prototype deliverable in 6 months (timely). The customer plan recommender is our next priority once we get maintenance off the ground,” etc. This kind of rationale is compelling to stakeholders.

8. Always Have a Final Sanity Check

  • Revisit the business problem. Will solving it truly “move the needle” for the business in the way you expect? Sometimes in the excitement of AI, teams realize a particular solution, while cool, might not actually yield substantial business gains. Ensure the AI project still aligns with strategic priorities and that success metrics can be defined (e.g., reduce churn by X%, increase productivity by Y).

  • Executive and team buy-in. Check that there is support from stakeholders to implement this AI solution. Change management is often the hardest part – if your AI recommends actions or changes workflows, business owners need to be on board. It helps if you can tie the AI project to an existing goal leadership cares about, something like “improve customer retention” or “increase operational efficiency”.

  • Consider any ethical, privacy, or regulatory implications. AI in 2025 faces rightful scrutiny on bias and transparency. If your opportunity involves sensitive areas (hiring, financial decisions, or healthcare for example), plan for how you’ll ensure fairness and compliance. A great AI opportunity is only great if it’s used responsibly and trustworthily. Design with a human-in-the-loop or review processes if needed to mitigate risks.

Finally, remember you don’t have to go it alone. As a McKinsey report aptly noted, even large digital natives like Amazon and Google partner externally to bolster their AI skills. This has become even more evident now – more than half of companies plan to work with external AI specialists or vendors to accelerate their AI initiatives [4]​. 

Congratulations! If you’ve followed these steps, you should have a clear, updated perspective on where AI can truly benefit your business, and a prioritized roadmap to get started.

A graphic explaining a four-step process to become an AI Innovator and to identifying high-value AI opportunities in your business.

With AI’s capabilities maturing and adoption becoming global, 2025 is a fantastic time to turn well-chosen AI opportunities into real business value. The key is to stay focused on impactful problems, leverage your data and intuition, and iterate quickly. Now, go forth and get a prototype in motion for your top AI opportunity – the sooner you test and learn, the faster you’ll join the ranks of companies reaping the rewards of AI innovation!


Need help spotting and vetting AI opportunities in your business? 

Don’t fall behind. Companies moving fast on AI are already seeing massive gains. The right strategy and team can turn your AI vision into a successful reality. Let’s talk before your competition gets ahead.


Discover how the Xyonix Pathfinder process can help you identify opportunities, streamline operations, and deliver personalized experiences that leave a lasting impact.


Check out some related articles:


Sources:

  1. Autodesk. (2024). Top 2025 AI construction trends, according to the experts. Autodesk. https://www.autodesk.com/blogs/construction/top-2025-ai-construction-trends-according-to-the-experts/

  2. McKinsey & Company. (2024). Superagency in the workplace: Empowering people to unlock AI’s full potential at work. McKinsey Digital. https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work

  3. PwC. (2024). AI predictions. PwC. https://www.pwc.com/us/en/tech-effect/ai-analytics/ai-predictions.html

  4. Boston Consulting Group. (2024). Generative AI adoption in Asia. BCG. https://www.bcg.com/publications/2025/generative-ai-adoption-in-asia

  5. Eyermann, J. (2024). Bullish AI spenders report higher return rates: EY study. CFO Dive. https://www.cfodive.com/news/bullish-ai-spenders-report-higher-return-rates-ey/735071/

  6. Vogue Business. (2024). What if robots ran the beauty industry? Vogue Business. https://www.voguebusiness.com/story/beauty/what-if-robots-ran-the-beauty-industry

  7. PwC. (2024). AI predictions. PwC. https://www.pwc.com/us/en/tech-effect/ai-analytics/ai-predictions.html

  8. Boston Consulting Group. (2024). AI adoption in 2024: 74% of companies struggle to achieve and scale value. BCG. https://www.bcg.com/press/24october2024-ai-adoption-in-2024-74-of-companies-struggle-to-achieve-and-scale-value

  9. McKinsey & Company. (2024). Superagency in the workplace: Empowering people to unlock AI’s full potential at work. McKinsey Digital. https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work

  10. Boston Consulting Group. (2024). Where’s the value in AI? BCG. https://media-publications.bcg.com/BCG-Wheres-the-Value-in-AI.pdf

  11. Menlo Ventures. (2024). The state of generative AI in the enterprise. Menlo Ventures. https://menlovc.com/2024-the-state-of-generative-ai-in-the-enterprise

  12. Gartner. (2024). Generative AI: The future of technology. Gartner. https://www.gartner.com/en/topics/generative-ai

  13. Adobe. (2024). IBM reimagines content creation and digital marketing with Adobe Firefly and generative AI. Adobe Business. https://business.adobe.com/au/blog/the-latest/ibm-reimagines-content-creation-digital-marketing-adobe-firefly-generative-ai

  14. Nuance Communications. (2021). Cooper University Health Care and Nuance Communications announce strategic collaboration and expansion of Dragon Ambient eXperience. Nuance News. https://news.nuance.com/2021-08-30-Cooper-University-Health-Care-and-Nuance-Communications-Announce-Strategic-Collaboration-and-Expansion-of-Dragon-Ambient-eXperience

  15. McKinsey & Company. (2024). The value of getting personalization right (or wrong) is multiplying. McKinsey Digital. https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/the-value-of-getting-personalization-right-or-wrong-is-multiplying

  16. Morgan Stanley. (2024). Morgan Stanley Research announces AskResearchGPT. Morgan Stanley. https://www.morganstanley.com/press-releases/morgan-stanley-research-announces-askresearchgpt

  17. PYMNTS. (2024). Spotify CEO: Upping user engagement and expanding AI use paid off. PYMNTS. https://www.pymnts.com/earnings/2024/spotify-ceo-upping-user-engagement-and-expanding-ai-use-paid-off/

  18. ScienceDirect. (2024). Waze: Engineering insights and advancements. ScienceDirect. https://www.sciencedirect.com/topics/engineering/waze

  19. Levine, U. (2023, June 9). Was selling Waze to Google a good decision? Founder of Waze reflects on the deal. Forbes. https://www.forbes.com/sites/urilevine/2023/06/09/was-selling-waze-to-google-a-good-decision-founder-of-waze-reflects-on-the-deal/

  20. U.S. Food & Drug Administration. (2024). FDA issues comprehensive draft guidance for developers of artificial intelligence-enabled medical devices. FDA. https://www.fda.gov/news-events/press-announcements/fda-issues-comprehensive-draft-guidance-developers-artificial-intelligence-enabled-medical-devices

  21. Ring. (2024). Receiving person and package alerts with smart alerts. Ring Support. https://ring.com/support/articles/9qnpt/receiving-person-and-package-alerts-with-smart-alerts