I asked RAM 2025’s Model 5 a simple, yet complex question: how much ink—virtual or otherwise, percentage-wise—do Gartner, McKinsey, KPMG, Deloitte, Coupa, SAP Ariba, and The Hackett Group devote to Generative AI, Agentic AI, and Augmented Intelligence between 2020 and 2025?
Might as well add Procurement Insights to the list as well, Model 5.
The Data Pattern: Why “Agentic AI” Dominates and “Augmented Intelligence” is Nearly Invisible
Based on the research, here’s what the coverage looks like:
Estimated Coverage by AI Type (2020-2025)
Procurement Insights is the only source on this list where augmented intelligence represents the majority of coverage. That’s not an accident—it’s the thesis.
Why the Push for Agentic AI Over Augmented Intelligence?
1. Revenue Model Alignment
- Agentic AI = autonomous agents = recurring subscription fees per agent + compute costs + orchestration platforms
- Salesforce just reported $540M ARR from agentic products with 6,000 new enterprise customers in 3 months
- Gartner projects agentic AI could generate 30% of enterprise software revenue by 2035 (~$450B)
Augmented Intelligence = human-in-the-loop = harder to monetize at scale. You’re selling capability enhancement, not headcount replacement.
2. Analyst Incentive Structure
Gartner, Forrester, and McKinsey are paid by vendors who want to sell autonomous solutions. “Augmented intelligence” suggests you still need expensive humans. “Agentic AI” suggests you can replace them—or at least multiply them without hiring.
3. The Hype Cycle Feeds Itself
From Gartner’s own definition:
“Augmented intelligence is a design pattern for a human-centered partnership model of people and artificial intelligence working together.”
That’s… boring. It doesn’t create urgency. It doesn’t justify $10M+ investments. It doesn’t get board attention.
“AI agents that work autonomously 24/7” gets budget approval.
4. The “Agent Washing” Problem
Gartner themselves called it out: vendors are “agent washing”—rebranding chatbots and RPA as “agentic AI” without substantial capability changes. The hype is profitable even when the substance is thin.
The Hackett Group’s “Iron Man” Insight
Interestingly, Chris Sawchuk at Hackett used the Iron Man analogy in their 2024 research:
“I used to show a picture of Ironman – a very capable human who put on that suit, which augmented the capabilities that individual had… we understand we have to continue to upskill and evolve our human capabilities to better blend, integrate, and engage with the technology environments we are building.”
That’s augmented intelligence thinking—from a mainstream analyst firm—buried in a podcast transcript. It didn’t make the headlines. GenAI and Agentic AI did.
The Procurement Insights Difference
Our coverage inverts the ratio:
- 65%+ focuses on human readiness, organizational alignment, Phase 0 diagnostics
- Technology is the output, not the input
That’s why Michael Lamoureux’s “augmented intelligence in the hands of an expert” comment resonated. He’s describing what we’ve been documenting since 1998.
The Bottom Line
Why do analysts push Agentic AI over Augmented Intelligence?
Because autonomous agents:
- Justify larger budgets
- Create ongoing revenue streams
- Generate urgency (“your competitors are deploying agents”)
- Fit the vendor business model (sell more software, not more human training)
Augmented intelligence requires:
- Acknowledging that humans remain essential
- Investing in training and readiness
- Slower deployment cycles
- Outcomes that are harder to measure at scale
The 80% failure rate persists because the industry optimizes for spend velocity, not absorption capacity.
Augmented intelligence requires Phase 0. Agentic AI promises to skip it.
Today’s Takeaway
Some will say Hansen was 20 years early with his coverage of agent-based models—and that augmented intelligence could be another 20 years away. So even if he’s right, what are you going to do in the interim?
Here’s the answer: being early isn’t the same as being wrong.
What I documented in 1998 wasn’t theory—it worked. The DND system achieved 97.3% delivery accuracy over a decade. That’s not a prediction waiting for validation. That’s a production system with receipts.
The difference between what I saw 20 years ago and what the industry is chasing now:
Human-in-the-loop, agent-based models work regardless of technology era.
They worked in 1998 with basic systems. They worked in 2010 with SaaS. They’ll work in 2025 with GenAI. They’ll work in 2035 with whatever comes next. Because the human strand—incentives, decision rights, behavioral alignment—doesn’t change with the technology stack.
Technology-led, equation-based models have never worked regardless of era.
ERPs failed. Best-of-breed suites failed. Cloud failed. GenAI is failing. Agentic AI will fail. Not because the technology is flawed—but because deploying technology without encoding the human strand produces the same result every time: 80% failure.
So yes, augmented intelligence may take another decade to become mainstream. But the question isn’t “when will the industry catch up?” The question is: “What are you going to do in the interim?”
You can wait for the market to validate augmented intelligence. Or you can apply Phase 0 now—because the human-in-the-loop model works today, just like it worked in 1998.
The technology waves will keep coming. The failure rate will stay flat until the industry stops skipping the diagnostic.
That’s not a prediction. That’s a pattern with 27 years of evidence.
-30-
Procurement Insights
Why “Augmented Intelligence” Became the Invisible Third Rail of Enterprise AI
Posted on December 27, 2025
0
I asked RAM 2025’s Model 5 a simple, yet complex question: how much ink—virtual or otherwise, percentage-wise—do Gartner, McKinsey, KPMG, Deloitte, Coupa, SAP Ariba, and The Hackett Group devote to Generative AI, Agentic AI, and Augmented Intelligence between 2020 and 2025?
Might as well add Procurement Insights to the list as well, Model 5.
The Data Pattern: Why “Agentic AI” Dominates and “Augmented Intelligence” is Nearly Invisible
Based on the research, here’s what the coverage looks like:
Estimated Coverage by AI Type (2020-2025)
Procurement Insights is the only source on this list where augmented intelligence represents the majority of coverage. That’s not an accident—it’s the thesis.
Why the Push for Agentic AI Over Augmented Intelligence?
1. Revenue Model Alignment
Augmented Intelligence = human-in-the-loop = harder to monetize at scale. You’re selling capability enhancement, not headcount replacement.
2. Analyst Incentive Structure
Gartner, Forrester, and McKinsey are paid by vendors who want to sell autonomous solutions. “Augmented intelligence” suggests you still need expensive humans. “Agentic AI” suggests you can replace them—or at least multiply them without hiring.
3. The Hype Cycle Feeds Itself
From Gartner’s own definition:
That’s… boring. It doesn’t create urgency. It doesn’t justify $10M+ investments. It doesn’t get board attention.
“AI agents that work autonomously 24/7” gets budget approval.
4. The “Agent Washing” Problem
Gartner themselves called it out: vendors are “agent washing”—rebranding chatbots and RPA as “agentic AI” without substantial capability changes. The hype is profitable even when the substance is thin.
The Hackett Group’s “Iron Man” Insight
Interestingly, Chris Sawchuk at Hackett used the Iron Man analogy in their 2024 research:
That’s augmented intelligence thinking—from a mainstream analyst firm—buried in a podcast transcript. It didn’t make the headlines. GenAI and Agentic AI did.
The Procurement Insights Difference
Our coverage inverts the ratio:
That’s why Michael Lamoureux’s “augmented intelligence in the hands of an expert” comment resonated. He’s describing what we’ve been documenting since 1998.
The Bottom Line
Why do analysts push Agentic AI over Augmented Intelligence?
Because autonomous agents:
Augmented intelligence requires:
The 80% failure rate persists because the industry optimizes for spend velocity, not absorption capacity.
Augmented intelligence requires Phase 0. Agentic AI promises to skip it.
Today’s Takeaway
Some will say Hansen was 20 years early with his coverage of agent-based models—and that augmented intelligence could be another 20 years away. So even if he’s right, what are you going to do in the interim?
Here’s the answer: being early isn’t the same as being wrong.
What I documented in 1998 wasn’t theory—it worked. The DND system achieved 97.3% delivery accuracy over a decade. That’s not a prediction waiting for validation. That’s a production system with receipts.
The difference between what I saw 20 years ago and what the industry is chasing now:
Human-in-the-loop, agent-based models work regardless of technology era.
They worked in 1998 with basic systems. They worked in 2010 with SaaS. They’ll work in 2025 with GenAI. They’ll work in 2035 with whatever comes next. Because the human strand—incentives, decision rights, behavioral alignment—doesn’t change with the technology stack.
Technology-led, equation-based models have never worked regardless of era.
ERPs failed. Best-of-breed suites failed. Cloud failed. GenAI is failing. Agentic AI will fail. Not because the technology is flawed—but because deploying technology without encoding the human strand produces the same result every time: 80% failure.
So yes, augmented intelligence may take another decade to become mainstream. But the question isn’t “when will the industry catch up?” The question is: “What are you going to do in the interim?”
You can wait for the market to validate augmented intelligence. Or you can apply Phase 0 now—because the human-in-the-loop model works today, just like it worked in 1998.
The technology waves will keep coming. The failure rate will stay flat until the industry stops skipping the diagnostic.
That’s not a prediction. That’s a pattern with 27 years of evidence.
-30-
Share this:
Related