Enterprise AI: Cutting Through the Noise

By Narayan Bharadwaj

The Buzz Around GenAI

In the span of 18 months, Generative AI has vaulted from niche experimentation to top-tier boardroom agenda. What began with viral demos of ChatGPT has now evolved into urgent executive discussions about productivity, innovation, and competitive advantage. Every Fortune 1000 CEO is being asked: "What is your GenAI strategy?"

But for many enterprises, the signal is still buried under the noise. While the promise of GenAI is undeniable, the path to scalable, secure, and value-generating adoption remains elusive. This is the moment for strategic clarity.

The Opportunity: Unrealized Enterprise Value

According to IoT Analytics, the GenAI software and services market has grown from $191 million in 2022 to over $25 billion in 2024. However, the majority of this revenue is concentrated in infrastructure and foundational models.

• Infrastructure Investments: Hyperscalers like Microsoft, Amazon, Google, and Meta collectively invested over $300 billion in AI infrastructure in 2024 alone—focused on GPUs, model training, and data center expansion.

• Model Revenues: OpenAI generated ~$2.7 billion in revenue in 2024, primarily from ChatGPT subscriptions and API usage—highlighting that most monetization today is at the foundation layer.

The application layer—enterprise-specific GenAI solutions that solve real business problems—remains vastly underpenetrated. McKinsey estimates that GenAI could unlock $2.6 to $4.4 trillion in annual value across industries, with a significant portion achievable through custom enterprise applications. This is the real opportunity: building “AI-native” apps that deliver end-to-end workflows, not just prompts.

A Framework for Clarity: Vishal Sikka’s TiEcon Lens

At TieCon 2025, Vishal Sikka from Vianai Systems, Inc. offered a crisp framework that resonated across the Enterprise AI track. He outlined three imperatives:

1. Acknowledge the limitations of today’s LLMs. GenAI is powerful but unreliable. Blind trust in foundation models will lead to brittle applications.

2. Focus on systems, not models. The breakthrough lies in building human-centered systems around the models—not worshiping the models themselves.

3. Prioritize real business value. The winning use cases are the ones that tangibly improve operations, decision-making, or customer experience.

Sikka also emphasized that AI is just another technology—an exceptionally powerful one, yes, but not magical. It must be treated with the same rigor, skepticism, and design thinking as any transformative tool.

This perspective is a useful north star for both enterprise buyers and GenAI builders.

Enterprise Use Cases That Matter

One emerging pattern across successful implementations is the rise of the agent platform: AI-powered agents that take contextual actions across multiple systems, rather than just returning results. These agents go beyond copilots—they initiate workflows, trigger API calls, and autonomously execute business tasks with minimal human intervention. These platforms point to a future where AI isn't just embedded in apps—it becomes the connective tissue across them.

According to McKinsey, BCG, and real-world startup activity, enterprise GenAI use cases are expanding rapidly across key domains. An increasing number of companies are building products to solve problems in these areas:

• Knowledge Retrieval & Decision Support: Automating research, surfacing institutional knowledge, and enhancing judgment-intensive workflows.

• Customer Support & Engagement: Powering virtual agents, reducing time-to-resolution, and handling long-tail queries with contextual awareness.

• Software Development: Assisting with code generation, debugging, and developer productivity through intelligent copilots and IDE plugins.

• IT Operations & Incident Management: Automating ticket triage, incident response, and system monitoring with AI-driven agents.

• Document & Communication Automation: Drafting, summarizing, translating, and contextualizing enterprise communications and documents.

• Sales & Marketing Optimization: Personalizing outreach, qualifying leads, and measuring campaign effectiveness with generative personalization.

• Legal & Compliance Automation: Accelerating contract analysis, regulatory compliance, and document review through structured LLMs.

• Finance & Procurement Insights: Automating invoice processing, generating forecasts, and streamlining vendor negotiations.

These companies are not just leveraging GenAI—they are building context-rich, system-aware applications that integrate tightly with enterprise workflows, prioritize governance, and deliver measurable ROI.

Enterprises Already Delivering Value

While many companies are experimenting, a few forward-thinking enterprises are already realizing tangible gains from GenAI—including in the high-impact domain of software development:

• Morgan Stanley deployed GPT-4-powered internal tools like the "Morgan Stanley Assistant" to give financial advisors instant access to proprietary research, reducing client response time significantly.

• eBay has used GenAI to optimize product listings, automatically generating titles and descriptions that increase conversion and searchability.

• Duolingo introduced GenAI-powered conversation partners and grammar feedback through its Max tier, significantly improving language learner engagement and retention.

• Moderna uses GenAI to accelerate drug discovery by modeling mRNA sequences and predicting therapeutic viability, reducing early-stage R&D cycles.

• Ernst & Young (EY) launched EY.ai to embed AI across its audit, tax, and consulting services—focusing on internal training, LLM integration, and client solutions.

• Google and Microsoft have reported that up to 40% of new code in some engineering teams is now authored with the assistance of AI-powered development tools like GitHub Copilot and internal copilots integrated into their IDEs and cloud platforms. This signals a shift in how software is designed, written, reviewed, tested, deployed and debugged —making GenAI an essential part of software development.

These companies demonstrate the emerging best practices: investing in foundational platforms, embedding AI in daily workflows, and measuring impact through operational KPIs rather than AI metrics.

Takeaways for Fortune 1000 companies

If you're leading a large enterprise, your GenAI strategy must go beyond pilots and press releases. The following principles can help ensure you generate sustainable value:

1. Start with the user. Identify your highest-friction customer and employee experiences and ask where AI could help—then build from there. As Brad Lightcap, COO of OpenAI, put it: "The most effective deployments we’ve seen start with user pain, not model capabilities."

2. Anchor on high-friction business workflows. Use AI to solve real inefficiencies in functions like underwriting, support, compliance, or procurement. American Express, for instance, has streamlined charge dispute resolution by embedding GenAI into its claims workflow.

3. Create a GenAI operating model. Establish cross-functional AI councils with representation from business, legal, risk, and data teams to govern responsibly and prioritize use cases.

4. Invest in data readiness. Clean, labeled, and governed enterprise data is the fuel for every meaningful AI use case. No data, no differentiation.

5. Pilot with intent. Choose proof-of-concepts that test integration depth, adoption friction, and business impact—not just model capabilities. EY’s “AI garage” model is one way to run controlled experiments with measurable goals.

6. Institutionalize trust. Deploy frameworks for explainability, monitoring, and fallback mechanisms, especially in regulated environments.

7. Up-skill the enterprise. Train knowledge workers, not just data scientists, on how to partner with GenAI tools. Walmart’s L&D teams, for example, are rolling out AI fluency programs to frontline managers.

These moves separate the leaders from the tourists in this new AI economy.

Guidance for Builders

For GenAI founders and product builders, creating lasting enterprise value means thinking beyond prompts and prototypes:

1. Get close to the customer. Embed forward engineers in user environments. Observe workflows. Co-create. This isn’t just product discovery—it’s customer immersion. Retool and Palantir popularized this model, and companies like Windsurf are doubling down on it for enterprise AI.

2. Solve for high-context use cases. The most enduring GenAI apps serve users who need deep context: analysts, engineers, legal teams—not casual consumers. As Glean’s CEO Arvind Jain notes, "Search is only useful if it understands your enterprise’s unique DNA."

3. Design for workflows, not outputs. Your product must seamlessly drive decisions or actions—not just generate content. Products like Ema and Typeface thrive because they align with CRM and marketing ops workflows.

4. Operationalize trust. Embed mechanisms for feedback, oversight, and traceability into the product by default. Harvey.ai’s legal copilot shows how confidence scores and citation tracing can win over skeptical users.

5. Win one wedge, then expand. Focus on a single high-impact use case with fast ROI and build outward from there. Mutiny, for example, started with personalized B2B landing pages before expanding into broader demand-gen tooling.

6. Build a go-to-market motion early. Enterprise AI is sold, not bought. Design your product and team for multi-month, multi-stakeholder sales processes. Founders from Aisera and Forethought have stressed the importance of customer education and ROI modeling in every pitch. PLG is necessary—but not sufficient. Credit card trials may help early traction, but they won't close enterprise deals with procurement, legal, and risk in the room.

Winning in enterprise GenAI requires a deep understanding of customer pain, clear economic value, and the grit to navigate complexity.

Looking Ahead

As we look toward the next wave of enterprise GenAI, the story is shifting from hype to execution. McKinsey estimates that GenAI could automate activities that account for 60-70% of an employee’s time in functions like marketing, software engineering, customer operations, and R&D. Yet, less than 10% of companies have scaled GenAI across business units. Why? Because most initiatives underestimate the cultural, architectural, and operational complexity of AI in the enterprise.

The leaders of tomorrow will be those who treat GenAI not as a sidecar, but as a core part of the digital operating model. That means:

• Embedding AI into the heart of decision-making and workflows

• Building the connective tissue across data, tools, and teams

• Designing for trust, accountability, and measurable outcomes

The best GenAI systems won’t be the flashiest demos—they’ll be the quiet, embedded agents that make work smarter, faster, and more human. We are still early—but the architecture of competitive advantage is being laid now.

I am interested to know your thoughts on enterprise AI.