Organizations Are Deploying AI Agents Faster Than They Can Explain the Context Behind Their Decisions, New Study Finds

GlobeNewswire | Quantexa
Today at 8:00am UTC

LONDON and NEW YORK, Nov. 17, 2026 (GLOBE NEWSWIRE) -- Banks, insurers and government agencies are putting AI agents to work faster than they can explain the context behind their decisions, according to a new InfoBrief from IDC commissioned by Quantexa, the global data, analytics and AI software company pioneering Decision Intelligence. The InfoBrief, “The Context Gap: What Agentic AI Needs to Scale,” surveyed 366 technology leaders at large banks, insurers and government agencies in North America and the UK in September 2026. It found that agents are expected to take on nearly a third of suitable work within 12 months, yet only 8.7% of organizations have context broadly available or embedded in their decision-making and AI workflows.

Agents already run in 21.6% of the workflows organizations judge suitable for AI agents, roughly one in five. Respondents expect that to reach 32.1%, nearly one in three, within 12 months.

Without context, an agent makes decisions it cannot ground or explain. Banks, insurers and government agencies must be able to show regulators, auditors, customers and citizens not only what an AI agent decided, but why: which data, which entities and which relationships sit behind each decision, traced back to source. Meeting that standard takes a context layer that connects data across silos, resolves it into entities and relationships, and shows its work on every decision. The stakes are rising: IDC forecasts enterprises will spend $1,086 billion on AI in 2026 and $2.6 trillion by 2029 according to IDC’s Worldwide AI and Generative AI Spending Guide (August 2026 forecast).

The IDC study defines context as the business, operational and relationship information within an organization’s data that AI agents and AI-driven decisions need to be reliable, explainable and trusted. Quantexa provides Real-World Context, a connected view of the people, businesses, places and events that matter, and the relationships and networks between them. It is a crucial type of context that organizations need to ground their AI agents in trusted information.

The Building Blocks Exist. The Connected, Real-World Context Does Not.

Most organizations have invested in individual context capabilities, but few have joined them strategically into the connected layer that AI decisions depend on. About half report having implemented each of 11 capabilities in at least some use cases, including audit trails (56.2%), entity resolution (53.8%), and knowledge graphs (51.5%). Far fewer run them across the organization as part of an integrated data estate: only 23.6% run entity resolution broadly and 16.7% run knowledge graphs broadly. Just 8.7% have context broadly available or embedded, while 39.3% still confine it to selected use cases. Relationship and network context is the gap organizations cite most often. The IDC study concluded that the issue is not a lack of capabilities, but a lack of connected, reusable context at scale.

Key Findings

  • Organizations Think Their Data Is Ready. Their Own Answers Say Otherwise. While 73.8% rate the data feeding their AI as good or excellent, 86% say it needs some improvement, a lot of improvement or a complete overhaul. 43.4% say it needs a lot of improvement or a complete overhaul.
  • Adoption is outrunning the real-world context needed. Agents are expected in 32.1% of suitable workflows within 12 months, up from 21.6% today, while only 8.7% of organizations have context broadly available or embedded.
  • Lineage ranks ahead of model complexity: organizations cannot explain what they cannot trace. Lack of data lineage (35.5%) and lack of decision lineage (35.2%) lead the barriers to trusted AI decisions, ahead of model complexity (29.8%).
  • Relationships are the most-cited missing context. Relationships and networks, meaning how people, organizations, accounts and events connect, are cited by 27.0% of respondents as the missing context that most limits AI-driven decisions.
  • The challenge shifts from connecting data to understanding relationships. Among the least mature organizations, 45.4% cite integrating data across silos as a challenge and 18.5% cite identifying relationships and networks. Among the most mature, that flips to 30.9% and 37.4%.

Recommendations for Leaders

The IDC InfoBrief recommends that organizations match investment to their stage of maturity: build trusted data foundations first, scale proven use cases next, then extend context across internal and external data relationships. For chief data and AI officers, the research recommends making lineage of both data and decisions a design requirement and treating relationships and networks as a core context capability. For business leaders, it recommends defining the decisions agents will own with a human in the loop, along with the context they need, and measuring decision accuracy and trust rather than adoption alone.

"Organizations have invested in the building blocks of context, but those capabilities sit in pockets. Lack of data and decision lineage, not model complexity, is what most limits trust in AI decisions, and a real-world relationship view is the context organizations are missing most."
— Stewart Bond, Vice President, Data Intelligence and Integration Software, IDC

“Agents are helping to make decisions, and in some cases making them autonomously, before anyone can show how those decisions were reached. In banking, insurance and government, a decision you cannot explain is a liability. This is a Real-World Context problem: data is fragmented, entities are unresolved and the relationships between them stay hidden. The institutions that scale agentic AI will be the ones that build a context layer including capabilities to resolve their data into entities and relationships that lets them show their work on every decision.”
— Dan Onions, Global SVP, Data and AI Platform Solutions, Quantexa

Download the InfoBrief. The full report benchmarks 11 context capabilities, breaks results out for financial services and government, shows how challenges change as organizations mature, and gives chief data and AI officers and business leaders a stage-by-stage plan for building a context layer for trusted, explainable AI. Read it at https://www.quantexa.com/resources/idc-study-2026/

Join the webinar. Quantexa will host a webinar, “Scaling AI: How to Identify and Close the Context Gap,” on Tuesday, December 8, at 11:00 a.m. ET / 4:00 p.m. GMT, to discuss the findings and how organizations can close the gap between agent adoption and trusted, explainable decisions. Register at https://www.quantexa.com/events/register/enterprise-data-and-context-readiness-for-ai-webinar/

The IDC survey, conducted in September 2026, included 366 director-level and above decision makers and influencers in IT, AI, and data roles at organizations with 1,000+ employees that are piloting or using agentic AI. Respondents came from financial services (251, comprising 201 in banking and 50 in insurance) and government (115, covering federal and limited state government). By region, 251 were in North America and 115 in the UK. The margin of error at 95% confidence is about 5 points overall, 6 points for financial services and 9 points for government and the UK.

The InfoBrief was authored by Stewart Bond, vice president, Data Intelligence and Integration Software; Sam Abadir, research director, Risk, Financial Crime, and Compliance; and Ruthbea Yesner, group vice president, Worldwide Government, at IDC.

IDC InfoBrief, commissioned by: Quantexa, The Context Gap: What Agentic AI Needs to Scale, EUR155013926, November 2026

About Quantexa

Quantexa is a global data, analytics and AI software company pioneering Decision Intelligence. When the stakes are highest, the world's most complex organizations turn to Quantexa for clarity. The Quantexa platform brings fragmented data together to create Real-World Context, a reliable and connected view of the people, businesses, places, and events that matter most, so teams and AI can make mission-critical decisions with confidence. Spanning entity resolution, knowledge graphs, contextual analytics, and agentic AI, Quantexa is stack agnostic, secure by default, built for enterprise scale and explainable by design. Customers use Quantexa to protect, optimize, and grow by solving complex challenges across the enterprise, from modern data management and customer intelligence to KYC, financial and economic crime, risk, fraud, and security. Quantexa has been named a Decision Intelligence category and innovation leader by all the major industry analyst firms. Founded in 2016, Quantexa has more than 850 employees and tens of thousands of users globally, working with billions of data points around the world. For more information, visit www.quantexa.com or follow us on LinkedIn.

Media Inquiries

Michael Lane
Quantexa
michaellane@quantexa.com

Ben Davies
SourceCode Communications
Ben.davies@sourcecodecomms.com


Primary Logo