SUB: Reltio CEO Manish Sood explains why “agentic AI” will first transform data matching and merging, as well as how enterprises can prepare for a new era of real-time, trusted data.
By Corey Noles
By 2026, the way enterprises manage their data may look radically different. Instead of humans painstakingly cleaning, merging, and validating records, autonomous AI agents will handle the heavy lifting—deciding which sources to trust, resolving conflicts in real time, and surfacing insights before teams even know to ask.
To understand how enterprises should prepare for this shift, TechnologyAdvice spoke with Manish Sood, CEO and Founder of Reltio, a company at the forefront of data unification and governance. In this conversation, Sood shares where agentic AI will first make its mark, how it will reshape the day-to-day work of data teams, and what business leaders should be measuring as they move toward an AI-driven future.
eWeek: When you say “agentic AI” for enterprise data, what’s the first everyday workflow it will actually change in 2026—matching/merging, survivorship, enrichment, or stewardship?
Manish Sood: Matching and merging will be the first everyday workflow transformed by agentic AI in 2026. It’s where the complexity of today’s data environments intersects most urgently with the speed and trust requirements of AI-driven operations.
Why matching and merging? Because real-world data is messy—full of duplicates, inconsistencies, and ambiguity. And in agentic AI workflows, there’s no time for human intervention every time a record doesn’t match perfectly. If an AI agent is going to recommend a product, process a claim, or engage a customer in real-time, it must operate on a clean, consolidated profile.
With pretrained, LLM-powered matching models, we’re already seeing the shift from rules-based “is this the same entity?” checks to agentic workflows that resolve matches in real time, complete with confidence scoring, auditability, and compliance guardrails. As this matures, it becomes the intelligent “front door” to every downstream workflow—survivorship, enrichment, stewardship, and beyond.
Agentic AI doesn’t just need data. It needs trusted, context-rich, consolidated data in milliseconds. That starts with entity resolution. That’s why matching and merging comes first.
eWeek: Paint a quick “day in the life” of a data team in late 2026. What’s different for a data steward, an analytics lead, and a business owner when agents are in the loop?
Sood:
Data Steward: Starts the day reviewing match suggestions already resolved by an agent overnight—complete with audit trails. Instead of chasing duplicates, they fine-tune policies and coach the agents.
Analytics Lead: Gets alerts from agents about quality anomalies before models drift. Focus shifts from cleansing data to collaborating with business teams on trusted, reusable data products.
Business Owner: Opens a dashboard where every metric is explained by an agent: what changed, why it matters, and what to do next. No backlog. No bottlenecks. Just answers—powered by governed, agent-ready data.
What’s different? Agents don’t just assist—they act. Teams move from fixing data to unlocking value.
eWeek: How do you let agents create the single source of truth without losing trust?
Sood: You don’t lose trust by letting agents help create the source of truth—you lose it if they do it invisibly. The key is governed autonomy.
Agents can act autonomously within boundaries: resolving matches, flagging anomalies, and enriching from verified sources.
The result? Transparency builds trust. Stakeholders see not just the data, but why it was trusted—who approved it, what logic was applied, and when.
That’s how agents earn their role—not just in surfacing the truth, but sustaining it.
eWeek: What early wins should a CDO expect in the first 90 days of embracing agentic AI in enterprise data management?
Sood: In the first 90 days, a Chief Data Officer introducing agentic AI should be able to point to some very tangible wins. For example, one of the first things you’ll notice is how quickly AI agents can clear out the backlog in match resolution. What used to be repetitive work for data stewards now gets handled in a fraction of the time, freeing up those teams to focus on higher-impact priorities.
You’ll also gain immediate visibility into data quality. Prebuilt agents can identify issues by domain, by source, even by business impact—so you’re not just improving data, you’re prioritizing the fixes that matter most. That clarity helps leaders establish a fast baseline and build early momentum.
And because analysts can finally stop spending their time cleaning data, you start seeing faster time-to-insight. Trusted data products come together more quickly, confidence in dashboards goes up, and KPIs start to feel reliable again.
Most importantly, business teams see the results. Cleaner data. Faster answers. And proof that this isn’t just another tech initiative—it’s delivering real business value from day one.
eWeek: What KPIs prove it’s working by the end of 2026?
Sood: By the end of 2026, the impact of agentic AI on enterprise data management should be visible in a few clear KPIs—and they tell a compelling story.
First, you’ll notice a significant reduction in manual effort. The percentage of match or merge decisions needing human review will fall. Data stewards will spend significantly fewer hours on repetitive tasks, and teams will spend less time tracking down and resolving data quality issues.
Second, data activation improves across the board. Dashboards and data products get to value faster. More of the data feeding AI models and workflows will be trusted and up-to-date. And business users will notice—they’ll have an easier time finding the data they need, when they need it.
Third, trust becomes something you can measure. You’ll see better data quality scores around accuracy, completeness, and freshness. Lineage becomes more visible. Policy adherence improves. And the amount of rework caused by bad or outdated data drops significantly.
Finally, you’ll see business value accelerating. More teams—from sales to operations—will start using agents to inform decisions. More actions will be taken autonomously, without incident. And more new use cases will take off thanks to trusted, real-time data.
Bottom line? Less friction, more trust, faster execution—and all of it traceable.
eWeek: How does agentic AI change data governance and compliance conversations with legal and security?
Sood: Agentic AI changes the conversation with legal and security teams in a pretty fundamental way. Where there used to be fear—about losing control, or opening the door to risk—now there’s proof that control is actually stronger.
Traditionally, governance has meant policies written down in documents, enforced manually after the fact. It’s reactive, it’s audit-heavy, and it’s slow. With agentic AI, that flips. Now, AI agents enforce governance at runtime. They inherit masking and access controls. They only operate within pre-approved boundaries. And every action they take is logged—with full lineage and rationale.
So when legal or security asks, “Can we trust what the AI is doing?”—the answer isn’t just “yes.” It’s “here’s exactly what it did, why it did it, and how we know it was within policy.”
For those teams, this isn’t a leap of faith. It’s a measurable upgrade in control: real-time enforcement, full auditability, and built-in oversight.
Agentic governance isn’t a compromise. It’s provably safer.
eWeek: Rules, LLMs, and retrieval: how do they fit together? When should an agent rely on deterministic data management rules vs. use an LLM with retrieval (RAG) to make a decision?
Sood: This isn’t an either-or question—it’s about making sure every AI decision is grounded in the same trusted foundation. Agentic AI needs a real-time, context-rich enterprise data graph to act reliably.
Most enterprises have customer, product, supplier, and location data scattered across dozens of systems, including CRM, ERP, data lakes, and third-party feeds. That data is often siloed, inconsistent, and not AI-ready.
Reltio solves this by continuously ingesting data from all these sources—structured and unstructured—and then unifying it in real-time. It utilizes advanced matching, survivorship logic, and AI-powered enrichment to create accurate and trusted profiles across various domains.
But here’s what makes Reltio unique: it doesn’t just create a clean record. It captures the relationships and interactions between entities—so you don’t just know who a customer is, but who they’re connected to, what they bought, when they called support, and how they relate to a product, supplier, or location. That’s what we mean by a data graph.
This graph becomes a system of context for AI. It provides every agent—whether resolving a match, personalizing an offer, or approving a claim—with a real-time, trusted foundation to make informed decisions.
So instead of every agent operating in isolation, they’re all drawing from the same continuously updated, governed, and explainable data environment. That’s what makes agentic AI scalable, compliant, and aligned with your business logic.
So the question isn’t which technique to use—it’s where those techniques are anchored. Without a system of context, even the best AI becomes fragmented. With Reltio, it becomes trustworthy at scale.
eWeek: What’s the right build–buy–partner mix? Where should enterprises lean on platforms (like Reltio + hyperscalers), and where does custom make sense?
Sood: When it comes to AI and data, the most innovative enterprises don’t try to build everything—they focus on building what differentiates them.
So what’s the right mix?
Buy the foundation. Data unification, governance, and real-time infrastructure are not DIY territory. You don’t want to spend years trying to cobble together your own data backbone. That’s where platforms like Reltio Data Cloud™, combined with your preferred hyperscaler, come in. They give you enterprise-grade trust, scale, and security—out of the box.
Partner where speed matters. Prebuilt agents, vertical templates, and implementation services can accelerate your time to value—especially for high-volume workflows like governance, match resolution, or data enrichment. There’s no need to reinvent what others have already optimized.
Build at the edge. Custom agents make sense when you’re solving for something uniquely yours—your scoring logic, your customer journey, your IP. That’s where your AI should feel like you, not a generic copilot.
Rule of thumb:
Platform for trust and interoperability
Partner for speed
Custom for edge use cases
eWeek: Looking past 2026, what’s the responsible pace? One bold bet you’d make for 2027–2028—and one pitfall leaders should avoid this year.
Sood: By 2027–2028, enterprises won’t just be deploying a handful of AI agents—they’ll be orchestrating hundreds, each performing high-stakes, real-time work across sales, service, compliance, and operations. The bold bet? Most of these agents won’t be built from scratch. They’ll be hired—just like employees.
That future only works if there’s a trusted system of context they can plug into instantly. That’s where Reltio comes in.
We believe every enterprise will need an intelligent data graph, not just as a source of truth, but as a source of action. One that continuously unifies, governs, and serves trusted data across domains, and proactively guides agents with policy, history, and context in milliseconds.
The bold bet: by 2028, the enterprise data graph will be as essential to agentic AI as the network was to the internet. And Reltio will be the foundation that makes agentic AI enterprise-grade—safe, scalable, and aligned with how our customers’ businesses actually run.
In other words, AI doesn’t replace data strategy. It makes it existential. And Reltio is how you make it real.
But in 2025–2026? Go slower to go faster. Start with embedded governance, not just flashy copilots. Trust is the compounding asset.
The biggest pitfall to avoid now is not letting LLM hype bypass your data foundation. AI and agents are tools that need trusted information to work properly. If the data isn’t unified, governed, and explainable—agents will move fast… in the wrong direction.
The post How Will Agentic AI Change Enterprise Data Management in 2026 and Beyond? appeared first on eWEEK.
No Responses