Enterprise AI in Practice: How Leading Firms Move from Strategy to Production

This article is sponsored by HTEC and was written, edited, and published in alignment with our Emerj sponsored content guidelines. Learn more about our thought leadership and content creation services on our Emerj Media Services page.

Operating models, workflows, and software development processes were never designed for AI — resulting in stalled initiatives, lost momentum, and an inability to deliver measurable business value.

Research from RAND found that 84 percent of business leaders believe AI will significantly impact their organization — yet only 14 percent report being fully ready to integrate it, and more than 80 percent of AI projects fail at twice the rate of comparable non-AI technology projects. Stanford HAI’s 2025 AI Index confirmed that while organizational use of generative AI more than doubled in a single year, most companies that report any financial impact from AI estimate those benefits at low levels. The result is stalled initiatives, lost momentum, and an inability to deliver measurable business value.

Emerj featured three leaders from HTEC on the AI in Business Podcast — Lawrence Whittle, Chief Strategy Officer; Ronny Fehling, Chief AI Transformation Officer; and Tim Sears, Chief AI Officer — in a series examining how enterprises can move AI from isolated wins to repeatable, business‑visible impact.

This article examines four insights that clarify the conditions under which enterprise AI produces real, repeatable business outcomes rather than isolated pockets of progress:

End‑to‑end workflows as the real unit of AI value: AI only produces measurable ROI when teams build connected workflow sequences instead of isolated pilots, enabling momentum, repeatability, and business‑visible impact.

AI built inside the live workflow: The first AI slice must run in the real system of record — narrow enough to finish in six to twelve weeks and important enough that users feel pain if it disappears — because only in‑workflow usefulness creates the pull and operational proof required to scale.

Scaling AI from solo users to teams: Apply AI to shared workflows — the steps teams rely on together — so delivery speeds up across the whole group instead of producing scattered individual improvements.

Enterprise AI succeeds when the work changes, not just the tools: The real constraint isn’t model capability — it’s how teams choose, build, sequence, and deliver the work that AI touches.

End‑to‑End Workflows as the Real Unit of AI Value

Episode 1: What Enterprise AI Looks Like When It’s Real – with Lawrence Whittle of HTEC Group

Guest: Lawrence Whittle, Chief Strategy Officer at HTEC Group.

Expertise: Digital Transformation, AI Strategy, Data & Analytics, Go-to-Market Strategy

Brief Recognition: Lawrence Whittle is a seasoned technology executive with leadership experience across AI, data platforms, and enterprise software organizations. He currently serves as Chief Strategy Officer at HTEC, following executive roles including President and Chief Commercial Officer at Verana Health, where he helped scale a real-world data platform focused on healthcare insights, and CEO of Parsable, a connected worker platform acquired in 2024. Previously, he served as Chief Revenue Officer at Persado, an AI company focused on language intelligence, where he helped drive growth through enterprise adoption across financial services, healthcare, telecommunications, and retail. Lawrence has also been involved as an investor and advisor to technology companies, with experience spanning multiple IPOs and M&A transactions.

Lawrence Whittle makes a clear distinction among users, use cases, and end‑to‑end workflows, explaining why only the third produces measurable ROI. His perspective is grounded in what he has seen across global clients: pilots succeed technically but fail commercially because they never engage the full business sequence in which value is created and measured.

He explains that organizations spent 2024–2025 validating concepts rather than validating value. Pilots ticked technically, but they were scoped around narrow use cases that couldn’t demonstrate business impact. Leaders could see activity, but they couldn’t see return. Whittle argues that enterprises unlock momentum when AI is deployed across the actual workflow sequence — the chain of steps where cost, velocity, and conversion metrics live.

He frames the difference between experimentation and enterprise‑level value creation:

“There’s a big difference between a user, a use case, and an end‑to‑end use case. A user is just an individual experimenting with tools, and a use case is a small slice of a business process. Neither of those creates measurable enterprise impact. You only see real ROI when AI spans the full workflow — the sequence of steps that lets you say, ‘I spent X and got Y back,’ whether that return shows up as higher conversions, lower costs, or faster velocity across multiple teams.”

— Lawrence Whittle, Chief Strategy Officer at HTEC Group

Workflow sequencing — not tool selection, not pilot volume — is the real unit of value. Evaluating AI across the full workflow sequence makes the return visible, because value accumulates across steps rather than within isolated tasks. When AI is deployed at that level, even small projects show clear cause‑and‑effect, creating the momentum that accelerates adoption across the enterprise.

AI Built Inside the Live workflow

Episode 2: Fixing the Pilot‑to‑Production Gap in Enterprise AI – with Ronny Fehling of HTEC

Guest: Ronny Fehling, Chief AI Transformation Officer at HTEC

Expertise: Generative AI Strategy, AI Transformation, Enterprise AI Implementation, Digital Strategy

Brief Recognition: Ronny Fehling is a technology and AI transformation leader with more than 20 years of experience spanning software engineering, digital transformation, and enterprise AI. He currently serves as Chief AI Transformation Officer at HTEC, where he focuses on scaling production-grade AI capabilities across operating models, engineering teams, and client solutions. Previously, Ronny was Partner and Vice President of Generative Artificial Intelligence at BCG X, where he worked with Fortune 500 organizations on AI strategy and implementation, built AI solutions delivering measurable enterprise value, and led global teams of AI scientists, engineers, and strategists. He also founded and scaled Spend AI, a patented AI solution that delivered significant cost-optimization value for enterprise clients. Ronny holds a Master’s degree in Computer Science from the University of Freiburg with a specialization in Artificial Intelligence and completed studies in Computer Science, Mathematics, Robotics, and related fields at MIT.

​

AI scales when the first slice runs inside the real workflow, where real users, real systems, and real constraints force it to prove its value.

Ronny emphasizes that most pilots fail because they’re built next to reality rather than inside it, which means that the moment production begins, everything the pilot avoided becomes the work.

He defines the first production slice as a deliberately narrow, bounded step inside a real workflow — small enough to finish in six to twelve weeks, but real enough that users feel the impact immediately.

When that slice is built outside reality, all the work that was deliberately left out hits at once: the systems it has to integrate with, the validation and certification it must pass, the governance and monitoring it must satisfy, and the way real users actually behave. What looked like momentum becomes slowdown — not because the pilot was flawed, but because it never touched the operational reality it was meant to survive.

Ronny’s manufacturing example shows the opposite pattern. His team built a very small system for blue‑collar operators handling non‑quality events in a regulated environment — a workflow that routinely caused delays and blame. The slice sat lightly integrated into their existing system, required no new learning, and removed a source of daily friction.

In this case, operators previously handled non‑quality events through manual reporting and escalation steps, which slowed resolution and created ambiguity about accountability. The AI system inserted directly into that workflow reduced the time required to identify and resolve issues, eliminating redundant steps without requiring users to change systems or retrain.

“Their work just got easier,” and that usefulness created the pull enterprises struggle to manufacture: real users who would feel pain if the system were taken away.

Ronny draws a sharp line between experimentation and enterprise‑level value:

“Adoption is a symptom of a useful first slice, not something you engineer separately. The work you do has to remove the pain from the people doing it. If it doesn’t, no change program will save you.’

— Ronny Fehling, Chief AI Transformation Officer at HTEC

From his perspective, a first slice is ready to scale when it meets four conditions:

Removes real pain — Operators feel the difference. If the slice doesn’t eliminate friction, “no change program will save you.”

Lives in the system of record — It runs inside the actual workflow, not in a parallel environment built for experimentation.

Matters enough that someone cares — It touches a step that shows up in the P&L or in operational accountability.

Is painful to remove — Users generate pull because the slice makes their work meaningfully easier.

When the first slice runs inside the live workflow — with real users, real data, and real constraints — it produces the only signal Ronny trusts: proof that the system can survive inside the environment that must ultimately carry it.

Evaluate whether an AI initiative is ready for scaling by applying a simple set of production criteria:

It runs inside a system of record, not a test environment.

It removes a clearly defined operational pain point.

It is delivered within a 6–12 week timeframe

Its removal would create immediate workflow regression

When these conditions are met, AI moves from theoretical value to operational proof, creating the internal pull required for broader deployment.

​

Scaling AI from Solo Users to Teams

Episode 4: How AI Is Reshaping the Way Enterprises Build Software – With Tim Sears of HTEC

​Guest: Tim Sears, Chief AI Officer at HTEC

Expertise: Artificial Intelligence, Machine Learning, AI Strategy, Data Science & Engineering

Brief Recognition: Tim Sears is an AI and technology leader with a background spanning machine learning, data science, finance, and entrepreneurship. He currently serves as Chief AI Officer at HTEC, where he focuses on helping organizations apply AI strategically and operationally. Previously, Tim led Software Applications at Groq, where he directed engineering teams building software and tools around Groq’s AI inference technology, contributing to advancements in AI workload performance and scalability. Before Groq, he built and managed Target’s Data Science & Engineering organization, applying data and AI capabilities to improve business outcomes at scale. Tim has also advised organizations, including Bain & Company’s Advanced Analytics Group, and holds a Ph.D. in Computer Science and Machine Learning from The Australian National University, with research focused on machine learning model structures.

AI delivers meaningful enterprise impact when it accelerates the way teams work together, not just when individuals experiment on their own. Tim is explicit that today’s productivity gains are uneven because they depend on who is personally excited about AI, who has learned the tools, and who is willing to experiment. That creates scattered individual improvements — helpful, but not transformative.

The shift he describes is that AI must become a catalyst for teamwork. In software engineering, that means moving from isolated chats and personal boosts to shared wor

[truncated for AI cost control]