Artificial intelligence has moved well beyond research labs and into everyday marketing workflows. From personalised product recommendations to automatically generated web copy, machine learning tools are reshaping how brands connect with customers. The examples below illustrate the practical directions AI was already taking in the retail, content, and media-buying spaces.
IBM Watson in Retail: Conversational Commerce
IBM Watson is one of the most widely recognised AI platforms in enterprise use. Capable of ingesting large volumes of unstructured data and answering questions posed in natural language, Watson has been applied across healthcare, finance, and retail.
US department store Macy’s partnered with Watson through the intelligent engagement platform Satisfi to create a mobile tool called Macy’s On Call. Shoppers can type questions — such as where to find a particular category of clothing — and receive immediate, location-specific answers without waiting for a human staff member.
Apparel brand Under Armour took a different angle, using Watson to generate personalised insights for users of its fitness apps. By analysing activity data collected through those apps, the system aimed to surface customised training recommendations for individual users.
Both examples illustrate the same underlying pattern: AI handling high-volume, repetitive interactions at a scale and speed that human staff cannot match.
Open-Source AI: Amazon DSSTNE, Google TensorFlow, and Microsoft CNTK
Amazon attributes a significant share of its revenue to its recommendation engine. Rather than keeping the underlying technology proprietary, the company released a deep learning framework called DSSTNE (Deep Scalable Sparse Tensor Network Engine) as an open-source project. The stated rationale: product recommendation is a harder problem than image recognition because labelled training data is comparatively scarce. By open-sourcing the framework, Amazon hoped the wider developer community would contribute improvements that the company could then adopt.
This move sat alongside similar open-source contributions from Google (TensorFlow) and Microsoft (Cognitive Toolkit / CNTK). Together, these releases lowered the barrier for marketers and developers to experiment with machine learning without building models from scratch.
AI-Generated Copy: Natural Language Generation in Practice
Natural language generation (NLG) platforms take structured data — spreadsheet figures, database fields, product attributes — and turn them into readable prose automatically. Quill, developed by Narrative Science, is one of the better-known examples. Publications including Forbes have used Quill to produce earnings reports, and Credit Suisse has applied it to investment research coverage.
On a smaller scale, US car dealership group Dominion Dealer Solutions used Quill to generate vehicle descriptions for its listings pages. According to figures cited at the time, this led to a 50% increase in page views for used-vehicle listings and approximately US$1 million in additional annual revenue. A comparable tool, Wordsmith from Automated Insights, offered similar NLG capabilities for marketers in other verticals.
The practical implication for content teams is significant: NLG handles high-volume, data-heavy writing tasks (product descriptions, financial summaries, sports reports) efficiently, freeing human writers to focus on strategy, narrative, and creative work.
AI Web Design: Automated Site Generation
The Grid released an AI website builder called Molly that aimed to generate custom website layouts based on the content and images a user provides, without requiring manual design decisions. The proposition was that small businesses or agencies could produce unique, content-driven sites automatically.
A concrete use case put forward at the time: a boutique travel agency could automatically generate a personalised landing page for a customer interested in a specific destination — pulling in relevant activity suggestions, restaurant proximity data, and seasonal information — rather than pointing every customer to a generic brochure page.
This kind of dynamic, data-driven personalisation represents a logical extension of what recommendation engines already do for product listings, applied to the broader customer experience.
AI in Programmatic Media Buying
Programmatic advertising automates the buying and serving of digital ads through real-time auctions, connecting advertisers to publisher inventory at scale. AI takes this a step further by optimising bid decisions dynamically.
Wise Data Media developed a real-time bid management system called Wise Air, which applies AI to analyse consumer data and adjust bids before inventory is purchased. The system attempts to match the right audience, at the right moment, through the right channel — decisions that are too complex and fast-moving for manual management.
For marketing teams managing large digital budgets, AI-driven bid management reduces wasted spend and can improve campaign performance without proportionally increasing headcount.
What This Means for Marketing Teams
The thread connecting all of these applications is scale and personalisation. AI does not replace marketing strategy; it executes that strategy at a volume and granularity that manual processes cannot achieve. Key takeaways for marketing and DevOps teams integrating AI tools:
- Conversational interfaces (chatbots, on-site assistants) reduce support load and improve response time without sacrificing personalisation.
- NLG tools are mature enough for high-volume, structured content tasks. They are not a replacement for strategic or long-form editorial work.
- Open-source ML frameworks (TensorFlow, DSSTNE, CNTK) give in-house teams a viable starting point for building recommendation or personalisation features without licensing costs.
- Programmatic AI is most valuable when campaign data volumes are large enough to train the model meaningfully. Smaller campaigns may see limited gains.
- AI-generated web experiences are an emerging space; evaluate tools critically against your actual design and accessibility requirements.
The organisations seeing the clearest returns from AI in marketing are those treating it as an operational layer — automating execution — rather than a replacement for human judgement on audience, brand, and strategy.
Original reporting sourced from memeburn.com (2017).