---
title: "Gartner D&A 2026: The Conversations We Should Be Having This Year"
url: "https://atlan.com/context-and-chaos/issue/gartner-danda-2026-the-conversations-we-should-be-having-this-year/"
description: "A field guide to the questions worth debating in the hallways, and a checklist you can keep on your phone."
keywords: "Gartner, Context Engineering, Data Strategy, AI Readiness"
---

> Atlan is hosting Context Conference, bringing together the leaders and builders at the frontier of giving AI the context it needs to understand their business. It runs online on October 28, 2026, from 11:00 AM to 2:00 PM ET. Atlan co-founder Prukalpa Sankar opens and closes the day. Leaders from AstraZeneca, BNY and Verizon share why they invest in context and what they get from it. Registrants get early access to The AI Context Gap, a new study from MIT Technology Review Insights. Register: https://atlan.com/context-conference/

A Context & Chaos field guide by Tathagata Das Sarma and Vivek Dubey, published February 26, 2026 (12 min read), ahead of the Gartner Data & Analytics Summit 2026: the questions worth debating in the hallways, a checklist to keep on your phone, and the sessions worth attending. It is built from gaps the authors see across data leaders: missing context, unresolved semantics, governance that arrives too late, unclear ownership, and underfunded change management.

**Authors.** Tathagata Das Sarma works at the intersection of AI, metadata, and data platforms. Vivek Dubey works on product, growth, and community at Atlan.

## Six questions to keep on hand

1. How are you ensuring models and agents see enough business context to make reliable decisions, and who owns that context supply chain?
2. What is one term or metric your leadership team still debates, and how many AI use cases depend on it today?
3. Where have you moved governance in your AI lifecycle: earlier, before models and agents reach production, or still only after agents have acted?
4. For your biggest AI initiative, how clear is the line between who owns the business outcome and who owns ongoing cost and drift? Are they the same person?
5. For your riskiest AI-assisted decision, who is accountable if it goes wrong, and do they have authority to pause or shut it down?
6. What is the hardest behavior change you are asking of teams this year, and which legacy process is most likely to pull you back?

## The context gap is the root problem

- Most organizations do not know whether their AI systems have enough context to decide reliably: "That is not a tooling problem. It is a context problem."
- Summits have shifted from generic data quality to AI-ready data, with active metadata as the foundation of an AI trust stack.
- Gartner surveys: most organizations still rely on manual metadata inventories; only a small fraction use machine learning to analyze metadata or power recommendations. Models doing pricing, triage, or risk scoring run on partial context.
- Failure pattern: a model optimizes a revenue metric finance and sales define differently; an assistant retrieves from an asset not certified in years; a governance rule has no traceable link to the pipelines, models, or prompts that should honor it.
- Agents act on it. A retention agent pulls a churn definition from one system, a discount policy from another, and triggers an offer with no human review; when they are misaligned it "gives away margin on the wrong customers."
- More policy tools that cannot see the same context produce more rules without alignment. Gartner analysts describe metadata as the matchmaker between user questions and data. The fix is a context layer: a live graph of how questions, people, policies, data assets, and models connect, powered by active metadata, acting as the routing brain between every AI question and the right governed data.
- Leader questions: what is our operating model for context, who maintains the context graph beneath every AI interaction, and how would we detect a context failure before it becomes a reputational or regulatory incident?

## Semantic debt compounds the gap

- Teams define "customer," "active account," or "churn" differently. It used to surface as conflicting dashboards; with AI it surfaces as inconsistent answers, recommendations, and policy behavior. "A dashboard shows a wrong number. An agent makes a wrong decision and executes it."
- Published glossaries create documentation, not alignment. Treat semantics like an API contract:
  - clear ownership and change processes for each key definition;
  - automated propagation of changes into schemas, metrics stores, policies, and prompts;
  - monitoring for semantic regressions, not just technical ones (know if someone quietly redefines "churn" in a feature pipeline before AI acts on it).
- Each governed semantic domain becomes a reusable context product; without it every AI use case reinvents the business and every agent carries its own version of the truth.
- Summit exercise: pick one high-stakes term your organization still debates and trace where it appears across features, retrieval rules, prompts, guardrails, and dashboards.

## Governance cannot be retrofitted onto AI

- Governance usually arrives after political commitment is high, cast as obstruction; controls are waived and exceptions proliferate ("governance by incident"). With opaque models, external APIs, unstructured data, and automated actions, retrofitting expands the risk surface.
- Governance by design needs a shared context layer: if tools cannot trace from a policy to the data, model, prompt, and action it covers, governance stays aspirational.
- Three entry points:
  - **Use case intake**: a light pre-build checklist of allowed and off-limits use cases, data sensitivity, and initial risk level.
  - **Data readiness assessments**: before funding, check lineage, quality, and semantic clarity. Ready means context products exist (definitions resolved, lineage traceable, policies linked).
  - **Model and agent design templates**: ownership, applicable policies, rollback plans, evidence logging; for agents, the scope of allowed actions, human-in-the-loop (a person approves before the agent acts) vs human-on-the-loop (the agent acts, a person monitors after), and the kill-switch or rollback mechanism.
- "If the Governance team first hears about a use case in a post-mortem, the organization is already behind."

## Someone has to own outcome, value, and drift

- The question for peers: who owns that model in year two? Models drift, prompts evolve, regulations change.
- Only about a third of data leaders systematically track ROI. Agentic AI's most valuable outcomes (incidents avoided, escalations prevented, manual rework eliminated) are invisible to traditional ROI frameworks.
- Three owners are needed: **Outcomes** (whose OKRs suffer if the model disappeared), **Value** (who proves value and decides to scale or retire, counting what did not happen), and **Drift and lifecycle** (who funds and runs drift monitoring, fairness reassessment, prompt updates, and retraining).
- Many organizations have quietly moved from human-in-the-loop to human-on-the-loop without deciding to, an unacknowledged governance gap. Gartner frames governance as a growth engine tied to trusted AI; ownership makes that operational.

## The change management risk nobody budgets for

- Scaled governance succeeds when reframed as a business enabler, with embedded domain ownership and automation and councils and sponsors with authority.
- Legacy lock-in risks: AI in approval flows designed for static reporting; AI expectations on fragmented tool stacks where each tool holds a partial view of assets, definitions, and policies; new behaviors without aligned incentives or capacity.
- The bottleneck has shifted from infrastructure to behavior; customers ask for success plans, change frameworks, and transformation scorecards, not just connectors and lineage.
- Reflect: will you retire approval committees that slow decisions but add little scrutiny? Do domain owners and stewards have time and authority to manage AI risk? Are you investing as much in training and communication as in platform licenses?

## Make the hallway conversations count

The value lies in ownership, context, and change questions, not tool questions. The organizations that win "will not be the ones with the flashiest agents, but the ones that quietly build a context layer and governance-by-design habit beneath every AI interaction."

## Sessions worth your time at Gartner D&A 2026

As listed on the page (March 2026). Full guide: [Sessions worth your time at Gartner D&A 2026](https://atlan.com/know/gartner-data-analytics-summit-2026-must-attend-sessions/).

| Theme | Session | Speaker(s) | When |
|---|---|---|---|
| Context gap | How to Build the Context Layer for Reliable AI Agents | Andres Garcia-Rodeja | Wed March 11, 10:30 AM |
| Context gap | Using Active Metadata to Support Data Agents for AI | Mark Beyer | Mon March 9, 4:30 PM |
| Context gap | Cargill's Reusable Framework for Evaluating Active Metadata Platforms | Kirsten Walter | Tue March 10, 3:00 PM |
| Semantic debt and data readiness | AI-Ready Data: Lessons Learned Become Practices to Follow | Roxane Edjlali | Mon March 9, 12:45 PM |
| Semantic debt and data readiness | Best Practices for Implementing and Maintaining a Data Catalog | Thornton (TJ) Craig | Tue March 10, 12:30 PM |
| Governance convergence | Crossroads Debate: Data & Analytics Governance vs. AI Governance | Andrew White vs. Lauren Kornutick, CIPM | Mon March 9, 2:30 PM |
| Governance convergence | Trust as the New Currency: A Paradigm Shift in Governance | Guido De Simoni | Tue March 10, 10:30 AM |
| Ownership and scaling AI | How Is Agentic AI Impacting and Disrupting Your Data Management Discipline - Now? | Ehtisham Zaidi | Mon March 9, 4:30 PM |
| Ownership and scaling AI | PPG's Formula: Scaling Data Literacy With AI | Robert Howden and Austin Kronz | Mon March 9, 2:05 PM |

## The Insight Index (recommended reads)

- [Something Big Is Happening](https://www.linkedin.com/pulse/something-big-happening-matt-shumer-so5he/) (Matt Shumer)
- [The 2028 Global Intelligence Crisis](https://www.citriniresearch.com/p/2028gic) (Citrini and Alap Shah)
- [Data and AI Pulse Check: Gartner 2025 D&A Summit Reflections](https://sanjmo.medium.com/data-and-ai-pulse-check-gartner-2025-d-a-summit-reflections-65d7e2d4a98f) (Sanjeev Mohan)
- [Data Engineering After AI](https://www.dataengineeringweekly.com/p/data-engineering-after-ai) (Ananth Packkildurai)
- [Five Levels Between Chaos and (Almost) AI-Ready Data](https://dataintelligenceplatform.substack.com/p/five-levels-between-chaos-and-almost) (Jean-Georges Perrin)
- [Data Governance Operating Model Blueprint](https://datagovernancefieldlibrary.substack.com/p/data-governance-operating-model-blueprint) (John Wernfeldt)
- [Why the Smartest People in Tech Are Quietly Panicking Right Now](https://medium.com/activated-thinker/why-the-smartest-people-in-tech-are-quietly-panicking-right-now-d2feb86e7e4b) (Shane Collins)
- [The Context Problem Nobody's Fixing in Talk to Data Initiatives (NL to SQL)](https://metadataweekly.substack.com/p/the-context-problem-nobodys-fixing) (Manoj Shanmugasundaram)

Originally published in the [Context & Chaos newsletter on Substack](https://metadataweekly.substack.com/p/gartner-d-and-a-2026-the-conversations). All issues: [https://atlan.com/context-and-chaos/](https://atlan.com/context-and-chaos/).