What AI adoption actually looks like
Descriptions of AI in business often default to either breathless futurism or abstract statements about transformation. Neither is useful to a professional trying to understand what AI deployment looks like in practice.
The examples below are drawn from publicly reported implementations across a range of industries and company sizes. They are chosen for specificity: what the company did, which part of its operation it affected, and what changed as a result.
1. Morgan Stanley: AI for financial advisers
Morgan Stanley deployed an AI assistant built on GPT-4 that gives financial advisers instant access to research reports, investment strategies, and market analysis from the firm's library of over 100,000 documents.
The outcome: advisers spend less time searching for information and more time with clients. The tool does not make recommendations. It surfaces relevant content faster than a human search process.
2. Klarna: customer service AI
Klarna deployed an AI assistant for customer service that handles customer queries across 23 markets in multiple languages. The tool handles a volume of conversations equivalent to the work of several hundred human agents.
The nuanced reality: Klarna also reduced its human customer service headcount significantly. This example illustrates both the capability and the workforce implication that companies deploying AI at scale must navigate.
3. Duolingo: personalised learning at scale
Duolingo uses AI to personalise learning paths for each learner based on their progress, error patterns, and engagement. The AI adjusts the difficulty, pacing, and content mix for each individual in real time.
This would be impossible to deliver at Duolingo's scale with human instructors. AI makes the personalisation economically viable at 500 million users.
The Duolingo example illustrates a key principle: AI is most transformative when it enables something that was previously impossible at scale, not just when it makes something existing incrementally faster. Look for these unlock moments in your own industry.
4. Levi Strauss: AI-generated model photography
Levi's partnered with an AI studio to generate diverse model images for product photography without requiring large-scale photography shoots. This reduced production costs and expanded the diversity of representation in product imagery.
It also drew criticism from some photographers about job displacement. This example is included because it represents the kind of adoption that affects creative professionals directly and that is happening across retail.
5. Walmart: AI for supply chain
Walmart uses AI for demand forecasting, inventory optimisation, and supplier communication. The models predict product demand by store and region with high accuracy, reducing both overstock and stockout situations.
At Walmart's scale, a small improvement in inventory accuracy across thousands of stores produces significant financial impact. This is a back-of-house AI application that most customers never see but that affects the shopping experience directly.
6. Deloitte: AI for audit
Deloitte's AI tools scan large volumes of financial transactions during audit engagements to identify anomalies and patterns that warrant human investigation. Instead of sampling, the AI reviews complete data sets.
This changes the nature of the audit rather than eliminating the auditor. Human professional judgment on what the anomalies mean and what further investigation is required remains essential.
7. NHS: AI diagnostics support
Multiple NHS trusts are using AI tools for radiology, flagging abnormalities in scans for radiologist review. These tools do not diagnose. They prioritise the radiologist's review queue, surfacing the scans most likely to contain actionable findings first.
This reduces the wait time for patients whose scans contain concerning findings, without removing the clinical professional from the diagnostic decision.
The NHS example illustrates an important deployment pattern: AI as a triage and prioritisation tool rather than a decision-maker. This pattern is appearing across many regulated and high-stakes domains where human judgment cannot be removed from the final decision.
8. HubSpot: AI content suggestions for marketers
HubSpot has integrated AI into its marketing platform to suggest email subject lines, generate blog content drafts, and recommend social post variations based on past performance data.
This is AI as a workflow accelerator for marketing teams rather than as a replacement for marketers. The judgment about which content to pursue and how to position it remains human.
9. Stitch Fix: AI-assisted styling
Stitch Fix combines AI-generated styling recommendations with human stylist review. The AI generates an initial selection for each customer based on their style profile and feedback history. A human stylist reviews and personalises the selection before it is shipped.
The AI handles the pattern-matching at scale. The human adds the contextual judgment and personalisation that the AI cannot reliably provide.
10. A mid-size law firm: contract review
Multiple mid-size law firms report using AI contract review tools that read commercial agreements, flag non-standard clauses, identify missing provisions, and produce a risk summary for the reviewing lawyer.
The outcome is a faster first review pass. Senior lawyers spend their time on the analysis and advice that requires legal judgment rather than on the mechanical reading of contract language.
The pattern across all of these
With few exceptions, the most successful implementations combine AI efficiency with human judgment at the decision point. The companies trying to remove humans from consequential decisions entirely are encountering the resistance, regulatory challenge, and quality problems that come with that approach.
The AI courses at Deliberate Academy cover how to apply these principles in your own professional context, with role-specific training and verifiable credentials.