Reimagining the enterprise innovation engine in the agentic era

Four years ago, I outlined a playbook for building an enterprise innovation engine. The goal was straightforward: establish a repeatable model for discovering, aligning, developing, and commercializing emerging technologies before market shifts pass you by. While those foundational phases still apply, the technology landscape under our feet has shifted dramatically.

The arrival of agentic workflows has fundamentally changed how quickly an organization can translate research into portfolio impact. Prototyping cycles that previously took months now happen in days. Yet, this explosion in process velocity introduces a fresh challenge for IT leaders. Speeding up execution without updating your operational and communication models leads to friction, unmanaged risk, and disconnected teams. To build a sustainable innovation engine today, organizations must adapt their management structures and alignment mechanisms to keep pace with agentic execution.

Balancing velocity with organizational risk

In any organization, different parts of the ecosystem move at different speeds. Core infrastructure and compliance frameworks demand high stability, while emerging AI technologies evolve daily. Agentic tools allow engineering teams to build prototypes, generate tests, and iterate software at unprecedented rates.

Higher velocity can create severe organizational risk if stakeholders are surprised by rapid shifts in direction. The primary objective for IT leaders is not simply maximizing raw output speed. The true objective is increasing organizational velocity in their support of the business while keeping risk predictable.

Achieving that balance requires guardrails rather than gates. Instead of slowing down engineering teams with complex approval checklists, establish clear architectural boundaries early. When agentic workflows operate within well-defined operational guardrails, teams iterate rapidly while remaining aligned with enterprise standards.

How discovery and prioritization have evolved at Red Hat

We believe good ideas can come from anywhere, whether they originate from Red Hat, third parties or via collaboration. Historically, we’ve identified which research topics and emerging technologies to explore based on a set of strategies. Typically this involves identifying and collaborating with academic and open source organizations, commercial laboratories, venture capital firms, and startups, plus individual open source contributors, researchers, engineers, and entrepreneurs that understand the domain. 

Today, we use agents to streamline the process. These agents not only help us build a more comprehensive list of potential topics, technologies and collaborators, they also apply upstream metrics, product roadmaps, and portfolio strategy so we know which ideas to prioritize based on expected impact and applicability.

With the help of these agents, a team can synthesize a 30-page market research document, detailed technical specs, or exhaustive strategy memos more efficiently than ever. However, hitting executive leadership with massive volumes of AI-generated documentation creates acute information indigestion. As innovators, the responsibility falls on us to make information as digestible as possible for our stakeholders. 

Instead of asking stakeholders to read lengthy point-of-view papers, we can leverage the discovery and prioritization output to generate interactive applications and visual dashboards to demonstrate the business case for commercialization. Delivering the minimum effective dose of information provides the critical context and information needed throughout the software development lifecycle (SDLC) and drives faster alignment among teams and leadership. 

Democratized ideation, prototyping and the path to commercialization

Agentic systems also widen the funnel for who can participate in early-stage innovation. By lowering technical barriers for rapid prototyping, deep research agents serve as a broad ideation screen for the entire organization and prevent ideas from getting stuck behind review bottlenecks created by tribal knowledge or specialized tooling. Agent harnesses, such as OpenCode and OpenClaw, used in conjunction with our Model-as-a-Service (MaaS) capabilities give our engineers access to approved frontier and open weight models. This allows them to achieve prototyping objectives while staying within their token consumption quota.

However, generating ideas faster makes the commercialization phase even more critical. Taking an emerging technology from prototype to production still requires a clear handover to an engineering team responsible for long-term operational maintenance.

We’ve historically addressed this transition through technology transfer agreements. When a project matures to an alpha or beta stage, the emerging technology team pairs directly with the receiving product or IT delivery team. Rather than handing over code over a wall, engineers temporarily embedded with the receiving team assist with initial deployment and skill transfer. Once the receiving team is fully equipped to maintain the lifecycle of the service, the emerging tech engineers return to explore the next frontier. However, as receiving teams transition to centralized agentic SDLC pipelines, research and tech teams must shift focus toward seamless handoffs—providing prototyping artifacts like detailed context, specs, and tests directly into commercialization pipelines.

Building for continuous adaptability

Adopting emerging technologies is no longer about surviving a single shift like cloud or mobile. It is about creating a flexible operational muscle that continuously absorbs new paradigms. Agentic tools provide the raw power to build software faster than ever before. IT leaders who combine this technical velocity with flatter team structures, dynamic communication models, and clear risk guardrails will build an innovation engine capable of leading their industry into the future. 

Red Hat doesn’t just build agentic platforms for our customers—we use them to power our own business. What we learn along the way shapes our entire portfolio, carrying insights from early research all the way into our commercial products and customer workstreams.