Almost every company says it's innovating with artificial intelligence. Few can explain where this innovation fits into their strategy, and that's where the problem lies.
Innovation with AI has become a mandatory phrase in any results presentation, annual plan, or corporate LinkedIn post. In just over three years, generative artificial intelligence has moved from the laboratory to the center of the innovation discourse of companies, in Brazil and around the world.
The curious thing is that the more the word appears, the less it seems to mean. A chatbot on the website, a pilot program with agents, a newly hired data team, and a complete overhaul of a business model all fall under the same label. And when everything is innovation, it becomes difficult to know if the company is actually building an advantage or just following the conversation.
In this article, we begin with an old question. (What does it mean to innovate?) To take a closer look at AI innovation: what the data shows about the results, why the hype confuses strategy, and how to decide where it's worth investing.
What does innovation mean, before discussing AI?
The discussion about innovation is much older than the language models themselves. The Oslo Manual, the OECD's reference for measuring innovation, defines business innovation as a new or improved product or process that differs significantly from what the company did before and that has been placed on the market or in use. Note what the definition does not require: global novelty, cutting-edge technology, or guaranteed success.
Strategy literature also distinguishes between types of innovation. Henderson and Clark, in 1990, discussed incremental, modular, architectural, and radical innovations, as they change the components of a product or the way they connect. Clayton Christensen described disruption as something else: solutions that are born in segments that leaders ignore and rise to challenge them.
We really liked how the Sandbox School of Strategy This discussion can be summarized as follows: innovation categories describe different natures of change, not quality grades. Disruption is not the “highest level” on a scale that begins with incremental change. This seemingly simple idea greatly helps in understanding what is happening with AI within companies.
Innovation with AI
The latest figures show very rapid adoption and a much slower return. The technology works, but adopting a tool and innovating with it are different processes, and most companies are still in the first phase.
- Widespread use: Nearly nine out of ten respondents in McKinsey's global survey report regular use of AI in at least one area of their business, and 44% already say that AI is scaling within their company.McKinsey, The State of AI, August 2026).
- Limited financial impact: In the same study, 37% say that AI has contributed positively to EBIT, and only about 6% fall into the group of “high performers,” who attribute a significant impact on the bottom line to AI. The proportion is the same as in 2025.
- Value perceived by few: Only 13% of CIOs surveyed by Gartner say their company has already achieved significant or transformational value with AI.Gartner, October 2026).
- Optimism ahead of the outcome: In BCG's AI Radar 2026, 82% of CEOs say they are more optimistic about the return on AI than a year ago, but only 6% of companies see real value.BCG, August 2026).
- The value is starting to appear: In September 2026, BCG indicated that almost half of companies already generate value with AI, compared to only 5% with substantial value in the previous year's survey.BCG, September 2026).
- Brazil is still in its early stages: The ICT Companies survey by Cetic.br showed that the use of AI in Brazilian companies increased from 13% to 17% between 2024 and 2025, reaching 50% in large companies and 15% in small companies.Cetic.br, June 2026).
- In Brazil, optimism comes before the test: 84% of executives at large Brazilian companies say that AI generates significant business value, above the global average of 76%.KPMG, October 2026But only 40.7% of companies are able to demonstrate financial results with generative AI (TEC.Institute and Peers, MIT Technology Review Brazil, August 2026).
Taken together, these numbers tell a well-known story for those who follow technology cycles: adoption happens much faster than companies' ability to transform innovation into results. When value begins to appear, it is concentrated in those who made clear strategic choices, not in those who launched the most pilot projects.
Only 40.7% of Brazilian companies are able to demonstrate financial results with generative AI.
TEC.Institute and Peers, MIT Technology Review Brazil, August 2026
Why AI hype confuses innovation strategy.
The first effect of hype is to transform technology into a ranking criterion. An initiative seems more innovative because it uses AI, and even more innovative if it uses agents, as if technical sophistication were synonymous with business ambition. It's the same misconception as treating disruption as the top of a ladder: the technological nature of a project says little about the value it creates.
The second effect is haste.. When the entire market announces AI innovation projects, the internal question ceases to be “what problem do we want to solve?” and becomes “what are we going to show?”. Pilot projects are born without an owner, without metrics, and without a path to production, consuming budget and attention and ending up as a pretty presentation at a board meeting.
82% of CEOs are more optimistic about the return on AI. Only 6% of companies see real value.
BCG, AI Radar 2026
The third effect is more subtle: AI becomes a shortcut to avoid making choices. We believe that innovation is, first and foremost, a decision about where to play and how to win, to use the language of Roger Martin and AG Lafley in... Playing to Win. A company that competes on efficiency and one that competes on experience should use AI in very different ways. When both buy the same tools, with the same rhetoric, neither is truly innovating.
This doesn't mean that hype is useless. Every major technological shift (the commercial internet, social media, mobile) has gone through a phase of exaggeration, and it was during this phase that many companies learned what worked.
The use of AI in Brazilian companies increased from 13% to 17% in one year. In large companies, it has already reached 50%.
Cetic.br, ICT Companies, June 2026
Three different ambitions for the same technology.
A useful way to organize innovation with AI is the Innovation Ambition Matrix, proposed by Bansi Nagji and Geoff Tuff in [publication name]. Harvard Business Review. She separates the initiatives by the type of ambition, not by the type of technology:
- Core: To improve what the company already does for its existing customers. In practice, AI can be used to reduce service time, automate internal processes, customize a CRM workflow, or qualify leads more accurately.
- Adjacent: Leveraging existing capabilities into new markets or new uses. For example, transforming the company's accumulated knowledge into a digital product with intelligent recommendations, or opening a new AI-mediated relationship channel.
- Transformational: To create something that doesn't yet exist for needs that the market doesn't yet meet. This is where new business models built from the ground up around data and AI come in.
In the original study, above-average performing companies divided their innovation resources into approximately 70% for core innovation, 20% for adjacent areas, and 10% for transformational bets, while returns followed an almost inverse proportion. The authors themselves make it clear that there is no universal formula: the right ratio depends on the sector, the competitive position, and the risk appetite of each company.
The value of the matrix for AI innovation lies elsewhere. It forces us to define, for each project, what kind of ambition it carries. An internal assistant that saves the team hours is a legitimate and valuable core innovation, and should be charged as such: short timeframe, measurable gain, rapid scaling. A new AI-based product is a transformational bet and needs a different benchmark, with time for learning and validation metrics, not immediate revenue.
How to decide where to invest in AI innovation
At follow55, when we help a company structure its AI innovation agenda, the conversation almost never starts with the tool itself. It begins with a few questions that seem basic, but which filter out a large portion of projects that shouldn't exist:
- What strategic decision does this project support? If the initiative is not connected to a clear choice of where the company wants to compete, it is a technology experiment, not innovation.
- What kind of ambition does he have? Core, adjacent, or transformational. The answer defines the timeframe, investment, and measurement method.
- Does the necessary data exist and is it reliable? Data quality is a recurring reason for abandoning AI projects. Without an organized foundation, the model only automates the mess.
- Who owns the outcome? Projects that remain solely within the technology realm rarely reach the operational stage. AI needs a business manager.
- How does the project connect to the rest of the ecosystem? An isolated solution solves a specific problem. Integrated with CRM, channels, and data, it changes the way the company interacts with the customer.
Frequently Asked Questions
What is innovation with AI?
Innovation with AI is the use of artificial intelligence to create or significantly improve a company's products, services, or processes, generating measurable value. Having an AI tool installed is not, in itself, innovation: what counts is the change it produces in the business.
Does all AI innovation have to be disruptive?
No. Most of the value generated by AI today comes from core innovations, such as process automation, customer service, and personalization. Disruption is a specific type of change, not a higher level of innovation.
Why do so many AI projects never get past the pilot stage?
Studies point to similar causes: lack of connection to strategy, low-quality data, absence of a business owner, and poorly defined metrics. Not surprisingly, only 13% of CIOs surveyed by Gartner in 2026 said they had obtained significant value from AI.
How do you measure the results of an AI initiative?
The metric depends on the ambition of the project. Core initiatives should show efficiency or revenue gains in the short term; transformational bets should first be evaluated through learning and validation with customers. Using the same yardstick for everything often kills good bets too early.
Where should you start with AI innovation in your company?
Focus on the business problem, not the tool. Mapping where the company wants to compete, organizing the data, and choosing a first core project with a clear owner and metrics is a safer path than launching several pilot projects simultaneously. It's also worth understanding how... consumer of the future The relationship with brands must change.
Ready to turn AI into strategy?
The important question isn't whether your company is innovating with AI, but rather where it has chosen to innovate, why it made that choice, and how it helps it win. Companies that answer these three questions well tend to need fewer pilot projects and reap greater rewards.
References
- OECD and Eurostat, Oslo Manual 2018: Guidelines for Collecting, Reporting and Using Data on Innovation
- Henderson, R. and Clark, K., Architectural Innovation, Administrative Science Quarterly, 1990
- Nagji, B. and Tuff, G., Managing Your Innovation Portfolio, Harvard Business Review, 2012
- Martin, R. and Lafley, AG, Playing to Win: How Strategy Really Works, 2013
- Sandbox School of Strategy, So, what does it mean to innovate?, 2026
- McKinsey, The State of AI in 2026: On the Road to ROI, 2026
- Gartner, Gartner Survey Finds Only 13% of CIOs Say Their Enterprise Has Realized Significant or Transformational Value from AI, 2026
- BCG, Why AI Pilots Rarely Deliver Real Business Value, 2026
- BCG, AI Is Starting to Pay Off, 2026
- Cetic.br, ICT for Businesses 2025, 2026
- KPMG, Global AI Pulse, Q2 2026 (via Brasil Inovador)
- TEC.Institute and Peers, From pilot to results: the profitability of AI., MIT Technology Review Brazil, 2026
Portuguese 
English