Sensedia, Squadra and MIT Sloan study shows 74% of companies face unexpected AI costs and identifies governance as a critical factor for scaling the technology

‍Survey indicates that the competitive advantage shifts from AI adoption to its integration and governance at scale.

MALVERN, Pa. — August 13, 2026 — A study conducted by Sensedia and Squadra, in partnership with MIT Sloan Management Review Brazil, reveals: although 66.2% of executives point to AI as a business priority, only 7.4% of organizations have managed to broadly integrate it into corporate processes.

The mismatch highlights that the challenge has shifted from experimenting with AI to operationalizing it at scale. Consequently, 74% of organizations already face unexpected costs, technical redundancies, and compliance risks stemming from the decentralized and loosely governed use of AI tools.

The global survey combines a maturity study conducted during APIX 2026, an event dedicated to the API and integration ecosystem promoted by Sensedia, with interviews featuring leaders from companies, in addition to analyses produced by MIT Sloan Management Review and consultancies like Bain & Company, Gartner, McKinsey, and Deloitte.

The data indicates that gains in productivity, automation, and reduced response times do not always translate into proportional growth or competitive advantage. The differentiator is less about adopting AI itself and more about the ability to integrate it into the core architecture and processes of the business.

Scaling depends on data and governance

The findings also reveal limitations in companies' information infrastructure. For 54.4% of technology professionals, access to the corporate data required for AI applications is not yet structured. Among them, 31.6% rely on manual context insertion into prompts, while 22.8% use tools without access to corporate data.

The absence of structured governance mechanisms reinforces this scenario. According to the study, 41.4% of technical professionals work without consolidated risk management processes, while only 4.8% of executives claim to have complete governance models supported by dedicated platforms.

Furthermore, 80% of organizations do not yet have structured agentic architectures and 68% have not adopted the MCP (Model Context Protocol) in production environments, which are elements considered fundamental for the next generation of applications based on intelligent agents.

"Adopting AI is no longer the main challenge for companies. As intelligent agents begin executing critical processes, the need grows for an architecture capable of connecting data, systems, and models under a single layer of governance. Without this foundation, costs, operational risks, and the difficulty of scaling the technology increase," states Kleber Bacili, co-founder and CEO of Sensedia.

"Organizations have rushed to adopt AI, but adoption without proper architecture quickly creates a costly governance nightmare," said Lisa Arthur, Sensedia’s SVP North America & APAC and Global Chief Marketing Officer. "To turn AI experimentation into sustainable value, leaders must shift focus from isolated adoption to integrated orchestration. Connecting AI models, APIs, and enterprise data under a single, transparent governance layer is what enables companies to scale safely, control costs, and drive true business outcomes." 

From adoption to scale

The study indicates that companies are entering a new stage of maturity. Following the expansion of copilots and point automations, focus is shifting to scalable operations integrated into critical processes and driven by business outcomes.

In this scenario, enterprise architecture, integration, and governance become competitive differentiators. Companies maintaining isolated initiatives tend to capture only incremental efficiency gains, while organizations connecting AI to core processes can transform the technology into a sustainable competitive advantage.

The research also points to the evolution toward agentic AI ecosystems and multi-agent architectures, in which applications move beyond acting merely as assistants to executing tasks, making decisions, and autonomously interacting with corporate systems. Supporting this new model will require evolving toward "agent-ready" architectures capable of delivering secure, contextualized, and scalable access to APIs, services, and data.

APIs become corporate AI infrastructure

With intelligent agents increasingly integrated into operational workflows, APIs and integration platforms assume a strategic role in enterprise architecture.

The study concludes that APIs no longer function merely as integration mechanisms between systems, but now form the infrastructure responsible for providing context, traceability, security, and governance for AI applications.

"APIs become the layer connecting intelligence, processes, and decisions. In an agent-driven environment, they are essential for ensuring governance, observability, cost control, and scalability," states Marcilio Oliveira, co-founder of Sensedia.

Case studies show the path to scaling AI

The analyzed cases show that more mature organizations share a common trait: AI initiatives supported by robust integration, data, and governance architectures.

At Bradesco, one of the largest financial and banking groups in Brazil and Latin America, more than 600 AI applications in production generate an estimated $48 million annually in value. At Inter, a Brazilian financial institution that operates as a “super app,” integrating banking services, investments, insurance, cashback shopping, and even an international account into a single app, over 500 models support critical decisions regarding credit, fraud, and personalization.

MAPFRE, multinational insurance and financial services group, utilizes intelligent agents in customer service and claims journeys. Hcor, a hospital and emergency center specializing in various medical fields, is advancing its data interoperability architecture and AI-based clinical support. Meanwhile, Klabin, Brazil’s largest producer and leader/exporter of corrugated packaging paper, has adopted a multi-agent platform with centralized governance and unified control.

Competitiveness relies on AI orchestration

According to the research, by 2028, most organizations will need to replace isolated assistive experiences with end-to-end intelligent automation workflows. Additionally, Gartner projections indicate that companies adding AI to legacy architectures without structural modernization could face significant pressure on their margins.

"In the coming years, competitive advantage will not be determined by the number of AI models in operation, but by the ability to orchestrate agents, data, APIs, and systems within a secure, governed architecture. This infrastructure is what will allow turning experimentation into operational intelligence at scale," concludes Bacili.

For Rômulo Cioffi, Chief and Innovation Officer at Squadra, generative AI has already begun to commoditize a significant portion of software engineering, shifting the competitive edge from execution to decision architecture. "The companies that will lead are those capable of integrating intelligence directly into the operational flow, with intelligent agents capable of continuously observing, interpreting, deciding, acting, and learning. Software ceases to be the center, and decision-making becomes the infrastructure," he states.

The complete study is available at: https://www.sensedia.com/content/report-efficiency-doesnt-pay-the-bills-the-ai-paradox-in-organizations 

About Sensedia

‍Sensedia is the independent backbone that enterprises trust to connect and govern every agent, API, and integration across any stack. With deep roots in Brazil and a growing global presence in North and South America, Sensedia serves enterprises across financial services, telco, healthcare, and retail. ISO 27001 and ISO 27701 certified and PCI DSS compliant, Sensedia’s AI Gateway and AI Foundation Services extend its API management and integration expertise into the agentic era. Learn more at: sensedia.com 

About Squadra

‍Squadra is one of the largest technology consultancies in Brazil, acting as a digital partner for brands such as Inter, Vale, VLI, Unimed, Claro, and MRV. It combines integral design, applied AI, and technical excellence in development, operations, and data to build modern, intelligent, and scalable digital platforms. It brings together more than 700 tech talents, operating across Brazil and on projects that cross borders. With its proprietary AI platform, Genius, it boosts results and accelerates digital journeys with real impact. Learn more at: squadra.com.br

‍

MEDIA CONTACT

Press Contact: pr.sensedia@sensedia.com 

‍