How to Leverage AI for Autonomous and Intuitive Product Creation

General-purpose AI technologies have helped and augmented human efforts for a long time now: from completing code and predicting email text to generating advanced copy and deep personalization. As generative AI and large language models enter the wider imagination, we can see more clearly the deeper impact these technologies will have on creative and knowledge work.

We’re exploring how enterprises should leverage generative AI in a series of articles. This fourth one in the series will look at how generative AI techniques can help organizations create products intuitively and autonomously.

How New Product Design Process Will Benefit from Generative AI

The initial promise of AI from the 1950s was that it could mimic human activity. Over the decades, AI systems and technologies have been advancing into more and more highly nuanced areas we thought only humans could handle. Late in the last millennium, Deep Blue or Deep Mind chess-playing systems from IBM eventually defeated the world champion Garry Kasparov. Even the game of Go bit the dust a few years ago when DeepMind’s AlphaGo defeated the world champion.

Increasingly, the word “creative” has been applied to AI systems. We have seen time and again how generative AI can deliver human like, creative content both in text and in rich media. The next frontier for AI is to enable “new product imagination.”

The following steps are part of the process of envisioning and realizing new products. Each of these steps can be imbued with AI and generative AI capabilities.

Using AI in the Steps for New Product Ideation

• Introspection: AI can absorb internal design documents, market research, customer segmentation, text from product development collaboration and feedback on intermediate releases to identify product capability that could be built in the future. The knowledge hub could be prompted for ideas. Tools include Google’s DocumentAI and Azure’s Document Generative AI.
• Retrospection: AI tools can analyze vast volumes of user feedback, feature requests, issue reports, resolution reports and market success. The resulting knowledge hub can be interrogated for ideas on what works and what has not worked and what could work in the future. Tools include Lexalytics, MeaningCloud and Provalis QDA Miner.
• Proactivation: AI can assimilate intellectual property, including internal knowledge and owned patents, with competitive research, including strategy pronouncements and publicly available information, to identify addressable market gaps. These market gaps are opportunities that can be addressed with creative bundling of existing products or net-new products. Tools include PatSnap & Innography (for patent research), Receptiviti (for sentiment analysis in competitive research) and hyperscaler APIs from Azure, Google and IBM Watson are being explored for comprehensive proactive solutions.

Integrating AI into New Product Development

• Code development: AI-based code co-authoring tools can aid digital elements of the new product. Tools include OpenAI Codex (powers GitHub Copilot) and Repl.it Ghostwriter.
• Code explainer: AI can significantly reduce the time it takes to read code and identify the area to fix by explaining architecture and code in natural language for maintenance teams to come up to speed quickly. Tools include Feenk Glamorous Toolkit, Denigma and SourceGraph Cody.
• Training, learning and development: Generative AI is being used to create immersive learning experiences and virtual simulations. Learning pathways are customized to the learning style, immediate needs and context of the user (introductory courses, deep dive courses, refresher material etc). Tools include Cognii (personalized AI virtual tutors), ScribeSense (summarize learning material) and Anatomage (for surgical and medical students).

Integrating AI into New Product Market Access

• Outbound product management: AI tools are being widely used to generate leads, personalize outreach and customize nurturing. Tools include Pipedrive (upsell and cross-sell), Crystal (customize communication to the style of the recipient) and CloserIQ (personalized pitches).
• Marketing copy creation: Much has been written about generative AI’s capability to aid in marketing copy. Tools include Copy.ai and Persado (customized marketing content creation).
• Customer journey narratives: Zero-shot learners are being applied to historic customer journeys to visualize their pains and gains. Tools include UserPilot.

Integrating AI into Continuous Feedback and Adaptation

• Feedback collection: With new embedded AI, data collection and modelling can be supercharged to become more intelligent. Tools include Userpilot and BetterFeedback.
• Digital threads: “A customer retained is a customer made.” Digital-enabled, edge AI products that can adapt to the customer’s needs based on feedback retain more customers and morph into the new product without having to go through the new product cycle. AI-imbued digital threads can enable such flexible and auto-adapting products and experiences. In the future, we can expect software to be game-like, adapting to customer facial feedback and neural-type connections. This is still emerging. Look for facial recognition and game mode being embedded into edge-AI software systems.

In addition to the above topics, generative AI will impact the way organizations conduct designing, prototyping, market sizing and pricing.

Adoption of Generative AI in Product Creation across Industries

While the steps laid out above pertain to new product ideation with which most discreet manufacturers are familiar, these approaches find applicability across various industries.

Here is how AI will impact new products across verticals:

AI in Life Sciences

• Drug discovery: This is greatly enhanced by designing new drug molecules. Running simulations reduces the time taken to test these drugs and improve production quality. Synthesizing previous trial data (from both successful and unsuccessful trials) into an expert assistant eliminates the need for wasted time on tests that would fail.
• Medical research: Zero-shot learners continuously working EHRs, population health data and other determinants of health can make surprising predictions and connections that can suggest new ways to prevent and combat disease. Tools include Atomwise (new molecule design), Benevolent AI (identify new targets for drug development) and DeepMind Health (medical research on large data sets).

AI in Banking and Financial Services

• Investment research and risk assessment: All the published material on a sector can be consumed by a custom-built large language model. This knowledge hub can then be interrogated against the investment hypotheses and trigger points to automate investment recommendations. Tools include Finteza and Narrative Science.

AI in Insurance

• Abuse and fraud identification: Large data sets can be mined for heretofore unknown methods of fraud and abuse. Generative AI techniques can be applied with zero-shot learners to come up with new hypotheses and abusive patterns of actions to track and add to the arsenal. American insurer Lemonade is using generative AI to analyze claims data and identify hard-to-spot instances of fraud.
• Risk assessment: AI can take into consideration more risk factors than were previously on the radar. This could include large weather patterns and potential harm that could arise out of them. In other cases, hidden factors for life insurance risk can be modeled. Munich Re (large weather simulations) and Aviva (life insurance) are both using these ideas.

AI in Healthcare

• Patient diagnosis: Image interpretation can be improved with general-purpose AI.
• Patient journey: AI can help explain complex personal care aspects to patients and caregivers. It can help dosage discipline by providing consistent and personalized prompts to the user. AI assistants can proactively and reactively track patient wellness.
• Deep personalization: Choosing the best mechanisms for delivering care and having an AI assistant as part of the care team leads to better health outcomes. Tools include IDx-Dr (for retinopathy), Lunit CXR (Xray interpretation), Path.ai (diagnosis) and IBM Watson Oncology (treatment recommendations).

Future Digital Fabric Will Be Programmable and Intelligent

Enterprises that have made significant investments in digital threads stand to gain ground. Their value chain is already instrumented with digital aspects across the lifecycle. It is easy to add AI capability to that footprint to make them “intelligent threads.”

We can foresee a world where personalization goes from saving preferences to delivering truly customized experiences the user designs. The democratization of applications (from internal users to the entire customer base), intelligence at the edge and instrumentation with digital threads allow for a golden age for customers to securely create, design and perceive their experiences that fit their needs. The market will increasingly reward companies that are “platformized” to enable this vision.

To enable this future vision, we will have to ensure proper models for creation, unbiased data for product ideation and invisible and meta-adaptive security. This requires a thoughtful approach, ambitious visioning, disciplined investment and follow through.

ISG helps organizations design an AI strategy that works for them, choose the right technology mix and find the right partner to bring your AI solutions to life.