70% of AI Value May Come From Redesigning the Business—Not Deploying Better AI Tools
For years, the AI conversation has centered on the technology.
But new BCG research suggests the larger source of value may be somewhere else entirely.
Its 2026 Applied AI Index found that “future-built” companies expect nearly 70% of AI value to come from reshaping and inventing business processes, not simply from deploying better AI tools. the-formula-for-agentic-ai-valu…
That is a major shift.
Because if most future value comes from redesigning the business rather than improving the technology, then AI transformation is not primarily a deployment problem. It is an operating-model problem.
Better Tools Are Not the Same as Better Business
The usual AI sequence looks something like this:
Buy the tool → Deploy the tool → Drive adoption → Measure productivity
That approach can create useful gains, but it may leave the underlying business unchanged:
- A salesperson gets a better drafting assistant.
- A developer writes code faster.
- A customer-service representative receives quicker answers.
- A finance team automates reconciliation.
All of those can be valuable. But the deeper question is: Did the process itself change?
BCG’s distinction is important because the largest AI gains may not come from making existing work slightly faster. They may come from changing:
- Which steps exist
- Which decisions occur
- Where information moves
- What AI handles
- What humans handle
- How authority is distributed
- How exceptions are escalated
- How outcomes are measured
That is a very different kind of transformation.
The Unit of Transformation Is the Process
One of the most interesting examples in the BCG report describes a global business-process outsourcing company that redesigned its operating model around increasing levels of AI autonomy.
Its progression moved from:
- Insights: AI diagnoses the work
- Copilot: AI assists employees
- Autopilot: Agents execute work end to end
But the most important detail is not the technology. BCG says autonomy was defined at the process level, not at the level of an individual tool. The process determined where decisions occurred and when a human needed to remain involved. the-formula-for-agentic-ai-valu…
That distinction matters.
The wrong question is: How autonomous should this agent be?
The better question is: Which parts of this process should be autonomous, and which decisions still require human authority?
That moves AI design out of the tool layer and into the operating system of the business.
Work Backward From Strategy
BCG makes the same point in its workforce analysis. It argues that companies should not simply “agentify” the organization they already have.
Instead, they should work backward from strategy to determine:
- What should be delegated to agents
- Where humans should remain in the loop
- Where humans should remain in the lead
- What roles should change
- What new skills become necessary the-formula-for-agentic-ai-valu…
That is a much more consequential starting point. The sequence becomes:
Strategy → Process → Decision → Authority → Human/AI Allocation
rather than:
Tool → Adoption → Hope for Value
That may be one of the most important differences between AI experimentation and AI transformation.
Strategic Clarity Multiplies Value
BCG’s broader finding reinforces this point. According to the report, companies combining high strategic clarity with high applied-AI maturity generated roughly five times as much AI value as companies lacking both. the-formula-for-agentic-ai-valu…
Strategic clarity here is not simply having an AI vision. BCG describes future-built companies as concentrating AI around a focused, multiyear program tied to the largest business opportunities, with clear ownership and explicit mechanisms for tracking value to the P&L. It reports that 95% of these companies use clear KPIs or directly track P&L value from AI. the-formula-for-agentic-ai-valu…
So clarity means making choices:
- What deserves investment?
- Which processes matter most?
- Who owns the outcome?
- What value is expected?
- What evidence determines whether to continue?
That is much closer to capital allocation and operating discipline than to technology strategy.
Fewer AI Projects. Bigger Business Bets.
This is another important implication. BCG advises companies not to spread AI investment evenly across the enterprise. Its research suggests that AI value is concentrated in a relatively small number of core functions, and it recommends focusing investment in the two or three areas that matter most for a given sector. the-formula-for-agentic-ai-valu…
That makes sense. When AI makes experimentation cheaper, organizations can produce more pilots than they can realistically scale. The bottleneck becomes selection.
The question is no longer: Where can we use AI?
It becomes: Which business processes are important enough to redesign around AI?
Governance Is Part of the Redesign
Process redesign also changes the governance problem. BCG found that 42% of companies expect agents to act autonomously by 2030, yet only 5% currently have the relevant agent controls in place. the-formula-for-agentic-ai-valu…
That is not simply a compliance gap; it is an operating-model gap.
The report identifies six important control areas:
- Memory and context
- Scope and oversight
- Evaluation and rollback
- Tools and integration
- Ownership
- Security, audit, and cost guardrails the-formula-for-agentic-ai-valu…
And BCG reports that companies with all six controls in place enterprise-wide generate roughly three times as much agentic AI value as those with only one. the-formula-for-agentic-ai-valu…
That challenges the idea that governance slows AI down. Good governance can make greater autonomy possible because the boundaries are explicit.
Intelligence Is Not Authority
That distinction becomes critical as agentic systems improve. AI may become capable of interpreting a situation, making a recommendation, initiating an action, coordinating other systems, and completing an entire workflow.
But capability does not automatically justify authority. A better operating model asks:
What should AI be allowed to do here? And what must remain human?
The answer will vary by process. A low-risk, reversible action may justify broad autonomy. A high-consequence capital, regulatory, customer, or workforce decision may require human judgment even if the AI is technically capable of acting alone. That is why process redesign and governance cannot be separated.
The Economics Have Moved Beyond IT
BCG’s spending data makes the organizational shift even clearer. The report says AI spending has risen to approximately 3.3% of revenue, with about 2.8% occurring outside enterprise IT and only about 0.5% inside it. the-formula-for-agentic-ai-valu…
That means most AI investment is already moving into areas such as:
- Operating workflows
- Products
- Data
- Talent
- Transformation programs
- Governance
- Workforce enablement
AI may still look like a technology category. Economically, it is becoming an enterprise transformation category. Leadership cannot evaluate AI purely through the IT budget; the full operating cost—and the full value opportunity—lives across the business.
Redesign Requires Measurement
If an organization changes how work happens, leadership needs to know whether the redesign actually produced value.
BCG notes that future-built companies rigorously connect AI initiatives to business outcomes and the P&L. Among this group, 95% measure value through direct P&L impact or a proxy, while many less mature companies still measure directionally or not at all. the-formula-for-agentic-ai-valu…
That creates a more disciplined loop:
Redesign → Execute → Measure → Learn → Adjust
not:
Deploy → Adopt → Declare Success
Usage can increase while business value remains weak. Productivity can improve while costs simply move elsewhere. Automation can accelerate a process that should have been redesigned first. The operating model has to learn from outcomes.
The Enterprise Cortex Matters More as Models Commoditize
BCG introduces another useful concept later in the report: the enterprise cortex. As AI becomes embedded in decision-making and execution, the report warns that vendor lock-in can become something deeper—cognitive lock-in.
BCG argues that companies should retain control over proprietary knowledge, shared ontology, business rules, decision logic, and operating context. the-formula-for-agentic-ai-valu…
Models may change. Providers may change. The underlying intelligence may increasingly become interchangeable. But the organization’s context is not interchangeable. Neither are its:
- Business rules
- Decision history
- Operating knowledge
- Workflow structure
- Risk thresholds
- Learned exceptions
That may be where durable advantage increasingly lives.
The Real AI Transformation Is Organizational
Taken together, the BCG findings point toward a different model of AI transformation. Not:
Better AI → More Usage → More Value
but:
Strategic Clarity → Process Redesign → Explicit Authority → Human/AI Allocation → Governance → Execution → Measurement → Learning
The technology sits inside that system; it does not define it. And that may explain why some companies are getting much more value from AI than others: they are not simply deploying better tools, they are redesigning the enterprise around what those tools now make possible.
The Question for Leadership
The 70% finding gives leaders a useful diagnostic. Look across the current AI portfolio and ask:
How many initiatives are actually redesigning how the business works? And how many are simply adding AI to the workflow that already exists?
That distinction may become increasingly important. Because if most future AI value comes from reinventing processes rather than deploying better tools, then the next competitive advantage will not belong to the company with the most AI. It will belong to the company that is best at redesigning itself around it.