The gap between what companies invest in AI and the results they actually achieve is wider than most people realize. In its 2024 report, Where's the Value in AI?, BCG surveyed 1,000 senior executives across more than 20 sectors in 59 countries. It found that only 26% of companies had the capabilities needed to move beyond proof of concept and generate tangible value. The remaining 74% had yet to demonstrate meaningful value from AI. Anthropic, citing data from the U.S. Census Bureau, also reported that AI adoption among U.S. businesses rose from 3.7% to 9.7% over two years, but the absolute figure remains low.
The same BCG survey broke down the challenges companies face: roughly 70% involve people and processes, 20% are technical, and only 10% relate to AI algorithms. This is not a new pattern. MIT Sloan Management Review cited a NewVantage Partners survey of senior executives at Fortune 1000 companies in which 90% of respondents identified people and processes as the main barriers to data-driven transformation. Only 9.1% pointed to technology itself.
In other words, buying the right tools will not change the outcome if the right people are not using them in the right way. Most companies spend their budgets on technology selection and tool procurement, yet few plan for who should take responsibility at each stage.
Three stages, each requiring different people
Based on Intellicon Solutions’ experience implementing AI in Taiwan’s traditional industries, the process can be divided into three stages: awareness, seed team development, and organizational integration. This phased implementation method is known as the TURBO framework. Its central principle is that each stage requires a different combination of people and practices. Skipping any stage creates more resistance later.
Awareness stage: Help decision-makers understand the problem, not just sit through a presentation
Most companies skip this step and start directly with training. But the purpose of the awareness stage is not to teach people how to operate a tool. It is to help key decision-makers understand three things: which problems AI Agent can solve in their departments, how fully it can solve them, and what implementation will cost.
The role needed at this stage is an executive sponsor, typically a business unit head or chief operating officer. Technical expertise is not important. What matters is that this person has the authority to decide whether processes should change and whether people should be reassigned.
As discussed in Enterprise AI Anxiety Is Not the Same as Real Demand, an executive’s anxiety and a valid business need are two different things. The main task at the awareness stage is to turn that anxiety into specific, actionable requirements.
The milestone for completing the awareness stage: At least one business unit head can say, “I want to use AI Agent to solve problem Y in process X,” and is willing to allocate people and time to the effort.
Seed team: Start with five to ten people
Once the organization has established awareness, the next step is not a company-wide rollout. It is to build a seed team. Intellicon Solutions selects five to ten people from one department. They do not need technical backgrounds, but they must meet two criteria: they are curious about AI, and they experience clear, repetitive pain points in their current work.
The seed team’s responsibilities are to:
- Identify repetitive tasks in their daily work and work with an FDE implementation consultant to turn them into tasks that can be assigned to AI Agent
- Learn to use prompt engineering to convert business rules into AI Agent skills
- Run the tasks in real working conditions for two to four weeks and gather evidence of results. Based on Intellicon Solutions’ implementation experience, this is the typical timeframe for launching the first group of tasks
The role needed at this stage is a prompt engineer. In practice, this means seed team members learn to express business rules in natural language and structure them in a format that AI Agent can execute. EgentWrX is designed so that nontechnical users can create skills, tasks, and workflows directly through conversation, without writing code.
The platform’s administrative settings must also be configured during this stage. EgentWrX’s 21-step implementation checklist divides prelaunch preparation into four groups. First, build the organizational structure, including the unit hierarchy, roles, and time zones. Next, configure budget and data management, including AI credentials and knowledge base partitions. Then invite seed team members. Only after that should advanced capabilities such as skill sharing, integrations, and channels be enabled. Step 15 of the checklist explicitly recommends inviting only the first five to ten seed team members in the initial group.
At this stage, the FDE implementation consultant helps the seed team complete the full process, from workflow assessment to prompt engineering. For a detailed explanation of this methodology, see How One Person Can Build the AI Capacity of a Fifty-Person Team: The FDE Methodology.
The milestone for completing the seed team stage: At least three tasks have operated reliably for more than two weeks, and team members can modify the tasks’ instruction fields without external assistance.
Organizational integration: Expand from the seed team to the entire company
A company should begin a broader rollout only after the seed team has proven the approach. At this stage, the challenge is no longer whether AI Agent works. It is whether other departments are willing to use it and how management will govern it.
Organizational integration requires an AI product manager or AI transformation consultant. The role focuses on cross-departmental coordination: standardizing the practices validated by the seed team, adapting them for other departments, and resolving questions involving permissions, budgets, and data ownership. Seed team members are the best candidates to advance into this role because they understand both the business and the realities of implementation.
At the company-wide governance stage, someone must take responsibility for the organization’s overall AI strategy and risk. This is the role of a Chief AI Officer or Chief AI Strategy Officer. Small and medium-sized businesses do not necessarily need to create a dedicated position, but someone must assume the responsibility of deciding which processes to prioritize, how to allocate budgets, and what principles should govern data.
The EgentWrX administration system is designed for this stage. Four-tier budget controls at the tenant, unit, member, and Agent levels keep spending manageable at every level. Cube partitions in the knowledge base isolate each department’s data while allowing information to be shared when needed. A skill review queue ensures that skills developed by individuals must be approved before they are rolled out across the company.
The factor most often underestimated at this stage is how much the underlying processes need to change. 74% Need to Redesign Processes, but Only One in Five Are Ready examines this bottleneck in detail, so it will not be repeated here.
The milestone for completing the organizational integration stage: More than three departments are using the system, and the company has a clear AI SOP that defines what can be delegated to AI Agent and what requires human judgment.
The talent ladder determines how far implementation can go
The three stages correspond to three tiers of talent.
Prompt engineers form the foundation. Every employee who uses AI Agent needs this capability: describing business rules in clear natural language, specifying output formats, and setting trigger conditions. This capability should become part of how every employee works rather than a separate position.
AI product managers and AI transformation consultants make up the middle tier. They determine which processes are worth implementing, how to measure results, and how to manage organizational resistance. Once the seed team has demonstrated results, its members are the best candidates to advance into these roles.
Chief AI Officers and Chief AI Strategy Officers form the top tier. They do not need to know how to write prompts, but they must be able to determine AI’s place in corporate strategy, assess returns, and manage risk.
Deloitte’s 2024 survey confirms this reality. Among nearly 2,000 enterprise technology leaders, 37% acknowledged that their organizations were insufficiently prepared for AI-related talent issues, while approximately 75% planned to adjust their talent strategies within two years. Most companies recognize the problem but have not yet acted.
Treating implementation as a one-time procurement project means buying the tools without developing the people who know how to use them. Based on Intellicon Solutions’ experience implementing AI in Taiwan’s traditional industries, moving from awareness to organizational integration typically requires three to six months of organizational change. Each stage requires different roles to assume different responsibilities.
FAQ
What should a company do first when implementing AI Agent?
Start by building awareness among senior executives. The goal is for at least one business unit head to articulate the specific problem they want to solve. A single AI presentation is nowhere near enough. Without this foundation, the technical implementation will remain stuck in a cycle of asking, “What can this actually do?”
What kind of people should join the seed team?
They do not need technical backgrounds, but they must be curious about AI, experience clear and repetitive pain points in their work, and be willing to spend two to four weeks testing the system in practice. Based on Intellicon Solutions’ experience, five to ten people is the right size. A smaller team will struggle to generate organizational momentum, while a larger team will slow down iteration.
Can a traditional business without a technical team implement AI Agent?
Yes. EgentWrX is designed for direct use by nontechnical employees. They can create skills, tasks, and workflows through natural-language conversations without writing code. However, an FDE implementation consultant is still needed during implementation to assist with workflow assessment and prompt engineering.
What is the most common reason AI implementations fail?
According to the survey cited by MIT Sloan Management Review, 90% of senior executives consider people and processes the main barriers, while only 9% point to technology. A common failure pattern is skipping the seed team stage and moving directly to a company-wide rollout. When unproven practices are replicated across the organization, resistance is likely to be high.
How long does it take to implement AI Agent?
Based on Intellicon Solutions’ experience, the first group of tasks can go live within two to four weeks after the seed team starts. The full process, from the awareness stage through organizational integration, typically takes three to six months. These figures are based on experience and are not a guarantee. The pace depends on whether a sponsor with decision-making authority remains actively involved.
References
- MIT Sloan Management Review, “Why Culture Is the Greatest Barrier to Data Success” — A NewVantage Partners survey of senior executives at Fortune 1000 companies found that 90% considered people and processes the main barriers, while only 9.1% pointed to technology
- Deloitte, “State of Generative AI in the Enterprise Q2 2024” — A survey of nearly 2,000 enterprise technology leaders found that 37% were insufficiently prepared for AI talent needs, while approximately 75% planned to adjust their talent strategies within two years
- BCG, “AI Adoption in 2024: 74% of Companies Struggle to Achieve and Scale Value” — The report Where's the Value in AI? surveyed 1,000 senior executives across more than 20 sectors in 59 countries. Only 26% had the capabilities needed to move beyond proof of concept; approximately 70% of the challenges involved people and processes, 20% involved technology, and 10% involved algorithms
- Anthropic, “Anthropic Economic Index September 2025 Report” — Citing data from the U.S. Census Bureau, the report states that AI adoption among U.S. businesses rose from 3.7% to 9.7% over two years
- MIT Sloan Management Review, “Why AI Demands a New Breed of Leaders” — 91% of data executives at large companies consider culture and change management their greatest barriers