Ask the Analyst

The data & analytics category's center of gravity moved in 24 months from 'where is my data stored?' to 'can agents reliably reason and act on my data?'. Every serious vendor retrofitted, re-architected, or got acquired around agentic reasoning.

PrePost GenAI thesis (synthesizing 2024-2026 category moves) · 2026

Gartner estimates over 40% of agentic AI projects will be canceled by end of 2027 due to escalating costs, unclear business value, or inadequate risk controls. Of thousands of vendors claiming agentic capability, Gartner estimates only ~130 ship real multi-step reasoning.

Gartner · Jun 25, 2025

Anthropic launched the Model Context Protocol (MCP) on Nov 25, 2024 - an open standard for connecting models to tools and data context. By late 2025 the ecosystem reported thousands of servers and broad SDK adoption; MCP was donated to a Linux Foundation agentic-AI effort in Dec 2025. It is the most important integration standard for agents since REST.

Anthropic · Nov 25, 2024

Astrix Security analyzed 5,200+ open-source MCP server implementations: 88% require credentials, but 53% rely on long-lived static secrets (API keys / PATs) and only 8.5% use OAuth - with 79% of API keys passed via plain environment variables. A live, unsolved governance gap in the agentic stack.

Astrix Security - State of MCP Server Security 2025 · Sep 2025

Google announced the Agent2Agent (A2A) protocol Apr 9, 2025 and donated it to the Linux Foundation Jun 23, 2025 - a complement to MCP for agent-to-agent interop. Standards consolidation accelerated through 2025.

Google Cloud / Linux Foundation · Apr-Jun 2025

Gartner revived the Metadata Management Magic Quadrant in Nov 2025 after a ~5-year pause, declaring that D&A leaders 'must improve metadata management maturity or risk AI failure' - recasting catalogs as 'AI failure insurance'.

Gartner · Nov 2025

EvoluteIQ raised $53M in a growth round led by Baird Capital (Sep 2025) to scale its agentic AI platform. Founded 2019; ~125 people with ~80-85% in India; US is ~60-65% of business, Europe ~25-30%. Funds earmarked for global expansion, R&D, and acquisitions.

Baird Capital / EvoluteIQ / YourStory / Tech.eu · Sep 17, 2025

EvoluteIQ's EIQ platform automates end-to-end business workflows via a proprietary 'Agentic Mesh Architecture' plus an AI Workbench, spanning banking & financial services, insurance, healthcare, telecom and manufacturing. Existing connectors + engine + visual workflow canvas + auth stack form ~60% of a 'mid-market Palantir alternative' stack.

EvoluteIQ product positioning + PrePost GenAI stack-coverage judgment · 2025-2026

EvoluteIQ's nearest-fit whitespace is an opinionated mid-market stack: ontology + semantic/context layer + agent builder + pre-packaged connectors, MCP-native, at transparent mid-market pricing - the space Palantir leaves uncovered below its enterprise-only TCV floor. EIQ's sub-scores below are the author's research estimates; adjust the EIQ sliders to test sensitivity.

PrePost GenAI (founder thesis) - DISCLOSED author estimate · 2026

Databricks announced a >$4B Series L at a $134B valuation (Dec 16, 2025), led by Insight Partners, Fidelity and JP Morgan Asset Management. It crossed a $4.8B revenue run-rate in Q3 (>55% YoY), with >$1B run-rate each from data warehousing and AI products, and positive free cash flow over the prior 12 months.

Databricks Newsroom / CNBC / PR Newswire · Dec 16, 2025

Databricks ships Agent Bricks for multi-agent systems on enterprise data, alongside Mosaic AI, MLflow 3.0, Lakebase (system-of-record, via the Neon acquisition) and Databricks Apps as the UX layer - a full agent stack sitting on the Lakehouse where the data already lives.

Databricks product · 2025-2026

Databricks Unity Catalog (open-sourced under Apache 2.0) provides governance, lineage and Lakehouse Federation across the platform; combined with Tabular (Iceberg) it consolidates open-table governance under one control plane.

Databricks / Unity Catalog · 2024-2025

Snowflake ships Cortex Agents, Snowflake Intelligence, and Openflow (NiFi-based ingestion from the Datavolo acquisition), pushing reasoning and agentic workflows directly onto the AI Data Cloud where governed data sits.

Snowflake product · 2025

Snowflake Horizon provides governance, and Semantic Views give a native semantic layer; Snowflake acquired Select Star (Nov 24, 2025) for cross-platform lineage, signaling that catalog/lineage federation is strategic and unsolved.

Snowflake / Select Star · Nov 24, 2025

Snowflake signed multi-year deals with Anthropic ($200M, Dec 3, 2025) and OpenAI ($200M, Feb 2, 2026) and shipped OneLake/MCP interoperability - a wide model and partner ecosystem feeding its data cloud.

Snowflake newsroom · Dec 2025 - Feb 2026

Palantir US commercial revenue grew +137% YoY in Q4 2025, accelerating every quarter (71%→93%→121%→137%). FY26 guidance ~$7.18-7.20B (+~61%). Palantir already shipped the ontology + object-action pattern (AIP) that competitors are now retrofitting.

Palantir earnings · Q4 2025

Palantir AIP + Ontology models objects, properties, links and actions with writeback, plus an FDE (forward-deployed engineer) delivery motion - the shipped 'object-action-workflow' substrate every lakehouse is now trying to add on top of open plumbing.

Palantir AIP · 2023-2026

Palantir's enterprise motion centers on large multi-million-dollar TCV deals and FDE-led delivery; the mid-market (1-3 verticals, transparent pricing, self-serve) is essentially uncovered - the structural opening for a 'mid-market Palantir alternative'.

PrePost GenAI analysis of Palantir GTM · 2026

Microsoft Fabric was ranked #1 in Forrester's Data Fabric Platforms Wave (Q4 2025); Microsoft pairs Fabric with Foundry Agent Service, Agent 365, Copilot Studio MCP GA and Purview - the most aggressive hyperscaler MCP evangelism.

Forrester Wave / Microsoft Ignite 2025 · Q4 2025

Microsoft Purview plus Agent 365 extends DSPM (data security posture management) to AI agents - governance and posture for an enterprise's agent fleet sitting on OneLake/Fabric data gravity.

Microsoft · 2025

Salesforce announced an $8B acquisition of Informatica (May 27, 2025), combining Einstein + CLAIRE + MuleSoft + Tableau with Informatica's catalog/MDM/ETL - the most consequential governance-layer acquisition of the decade. CLAIRE GPT, Copilot and CLAIRE Agents extend it into agentic data management.

Salesforce PR · May 27, 2025

Fivetran and dbt Labs COMPLETED their all-stock merger on Jun 1, 2026 (announced Oct 13, 2025). George Fraser is CEO, Tristan Handy President; the combined company serves 100,000+ data teams and positions as the data-infrastructure layer that makes agents trustworthy. Note: a merger announcement is not its close - this is the verified completion date.

Fivetran / dbt Labs press release · Jun 1, 2026

dbt's Apr-2026 benchmark found that for questions within a well-modeled semantic layer's scope, accuracy reached ~98.2-100% (Claude Sonnet 4.6 98.2%; GPT-5.3 Codex 100%) versus ~84.1-90.0% for raw text-to-SQL. The semantic layer is the accuracy unlock - NOT, as sometimes mis-stated, the lower number.

dbt Developer Blog - Semantic Layer vs Text-to-SQL 2026 · Apr 2026

dbt Core v2.0 (alpha) open-sources the dbt Fusion engine runtime under Apache 2.0; combined with Fivetran's 80-90% customer overlap with dbt, GitHub+dbt is the clearest distribution channel for an agentic data engineer that opens reviewable PRs.

dbt Labs / Fivetran · Jun 2026

Dataiku ships LLM Mesh (an enterprise gateway routing OpenAI/Anthropic/Mistral/Bedrock/Azure/Gemini with guardrails, cost and quotas), 'AI Agents with Dataiku' (Apr 24, 2025: Agent Connect, Safe Guard, Trace Explorer), visual Agents & Tools, and an Agent Management control tower - repositioned as a 'Universal AI platform' preventing AI sprawl.

Dataiku product · 2024-2026

Dataiku reported ~$350M ARR (Oct 2025, ~30% YoY) with an H1 2026 IPO target and 750+ customers (J&J, Toyota, GE, BNP, Novartis, LVMH). It added column-level lineage + LLM prompt/output lineage and MCP integration (DSS 14).

Dataiku · Oct 2025

Dataiku's verb shift: PRE you built flows (visual recipes + graphical Flow); POST you declare goals and agents + Flow execute and adapt (Cobuild turns NL business goals into auditable projects, pipelines, models and agents).

Dataiku positioning · 2025

DataRobot repositioned from AutoML pioneer to an 'Agent Workforce Platform': an LLM Gateway swapping ~70 models without new credentials, full LLM lineage, a guard library (PII/jailbreak/hate-speech + NVIDIA NeMo), Agent 'rehearsal mode', automated EU AI Act / NIST RMF compliance reports, and hard recursive-loop circuit breakers (the 'hidden AI tax').

DataRobot product · 2024-2025

DataRobot reported ~$225M ARR (2024, ~12.5% YoY - decelerating) off a stale $6.3B 2021 valuation, with an Agnostiq/Covalent acquisition (Feb 2025) for multi-cloud GPU orchestration and an SAP endorsement (Aug 2025).

DataRobot · 2024-2025

Alteryx (taken private at $4.4B by Clearlake + Insight, Mar 2024) shipped AiDIN - a GenAI engine across its Analytics Cloud - with AiDIN Copilot (GA Dec 16, 2025, Gemini-based) suggesting and auto-placing tools from natural language, plus Magic Reports/Documents and an AI Data Clearinghouse.

Alteryx · 2024-2025

Alteryx's verb shift: PRE analysts drag-dropped 260+ tools to a workflow; POST they describe the workflow in chat and AiDIN Copilot wires the canvas. 8,000+ customers; Highmark Health modernized 3,000 legacy workflows.

Alteryx · 2025

Atlan AI is an agentic data steward with an MCP server (Q1 2025); Atlan claims its agents prepare ~80% of context before a human steward reviews, with ~70% auto-acceptance, on an Iceberg-native Metadata Lakehouse knowledge graph.

Atlan · 2025

Atlan was named a first-time Leader in Gartner's revived Metadata Management MQ (Nov 2025) and was reported as the only vendor above average on all five use cases.

Gartner Metadata Management MQ · Nov 2025

Alation shipped an Agentic Platform + AI Agent SDK (Mar 2025) and acquired Numbers Station (May 2025) for NL→SQL; a 5× Gartner Metadata Leader reportedly prepping an IPO.

Alation · 2025

Collibra remains a Gartner governance Leader and shipped an AI Copilot plus Raito and Deasy Labs retrofits, moving toward enforcement (Spring '26), but had no first-party MCP server as of Apr 2026 - an ecosystem gap relative to MCP-native catalog peers.

Collibra · Apr 2026

Monte Carlo ships a Monitoring Agent + Troubleshooting Agent and an Agent Observability Platform; reported a $1.6B unicorn valuation, ~60% agent-suggestion acceptance at Roche, and a >80% incident-reduction claim.

Monte Carlo · 2025

Acceldata launched Agentic Data Management (GA Jul 31, 2025) with an xLake Reasoning Engine and 10+ agents (DQ, Drift, Cost, Lineage, Pipeline Health) plus MCP, citing 3T+ policy executions/month.

Acceldata · Jul 31, 2025

Anomalo applies unsupervised ML to data quality and was first to extend automated DQ to unstructured data; reported as the only observability peer in Gartner's Augmented Data Quality MQ.

Anomalo / Gartner · 2025

Bigeye shipped an AI Trust Platform and AI Guardian for runtime data-quality enforcement (GA Dec 10, 2025); $66-71M raised with no new round since 2021.

Bigeye · Dec 10, 2025

Starburst (Trino federated SQL + Icehouse) added an AI Agent, AIDA and MCP with auto-tagging - a federation-first fabric play across heterogeneous sources.

Starburst · 2025

Dremio (Apache Polaris co-creator) shipped a Lakehouse AI Agent, Autonomous Reflections and a Vector Lakehouse; reported Maersk running 5,000 users at 99.97% uptime.

Dremio · 2025

Denodo (data virtualization, a 6×-consecutive Gartner Data Integration Leader) added an Assistant, AI SDK, Agent Builder and DeepQuery - virtualization as an agent-queryable fabric.

Denodo / Gartner · 2025

Reltio shipped AgentFlow (agentic MDM - Resolver, Address Enricher) delivering context in milliseconds; reported a ~70% reduction in manual data stewardship.

Reltio · 2025

Ataccama ONE Agentic (GA Nov 6, 2025) unifies DQ + MDM + governance + catalog with a digital-steward AI agent; a 5× Gartner Augmented DQ Leader (#1 Vision in 2026).

Ataccama / Gartner · Nov 6, 2025

Matillion ships Maia - agentic data engineers built Bedrock-native; General Atlantic-backed, pushing self-healing pipelines that detect schema evolution and auto-generate corrective transforms.

Matillion · 2025

Prophecy shipped a Data Transformation Copilot (first for Databricks) reporting 160% NRR and 3.5× FY24 growth ($47M Series B extension, Jan 2025).

Prophecy · Jan 2025

Gable.ai does code-level field lineage and CI/CD data-contract enforcement (static analysis from app code), raising a $20M Series A (Mar 2025) co-led by Databricks Ventures - explicitly funding contract-native data quality.

Gable.ai · Mar 2025

The agent-first context layer (vendor-neutral, catalog-agnostic, MCP-native semantic federation across Snowflake/Databricks/Fabric/Postgres with evals, permission propagation and observability of which metric each agent answer used) is open whitespace - hyperscalers ship semantic layers tied to their own gravity, not neutrally. dbt's benchmark (semantic layer 98-100% vs text-to-SQL 84-90%) quantifies the accuracy stakes.

PrePost GenAI synthesis + dbt benchmark [#71] · 2026

An MCP-native data governance gateway (row/column masking + purpose-based access + agent-identity federation + multi-hop OBO token delegation across heterogeneous sources) is critical whitespace: Astrix shows 88% of MCP servers need creds, 53% static keys, only 8.5% OAuth - a live security bomb every MCP-adopting enterprise must defuse in 2026.

PrePost GenAI synthesis + Astrix [#4] · 2026

A mid-market 'Palantir alternative' - opinionated ontology + semantic layer + agent builder + pre-packaged connectors for 1-3 verticals at transparent pricing - is uncovered whitespace: Palantir's $10M+ TCV enterprise-only motion leaves the mid-market open. This is EvoluteIQ's closest-fit positioning given its existing connector/engine/visual-canvas stack.

PrePost GenAI analysis · 2026

An agentic data engineer that opens reviewable PRs to dbt/Airflow with full context (warehouse metadata + tickets/Slack + contract YAML + cost projections + lineage blast-radius) has a clear distribution channel: GitHub + dbt, with 80-90% Fivetran/dbt customer overlap [#72].

PrePost GenAI synthesis · 2026

Framework-agnostic agent observability with hard per-agent cost caps (OpenTelemetry/OpenInference + 20+ data-agent evals + Okta/Entra tie-in + SOC 2/HIPAA) is validated whitespace - Datadog-Metaplane and Snowflake-Observe (~$1B) prove $1B-scale outcomes for observability of the agentic stack.

PrePost GenAI synthesis · 2026

A governed 'dbt for unstructured' pipeline (contracts + incremental re-embedding + hybrid retrieval + per-tenant isolation + HITL labeling + MCP exposure) has clear demand: LlamaIndex's waitlist and a Databricks/KPMG strategic partnership signal pull for AI-ready unstructured pipelines.

PrePost GenAI synthesis · 2026

A federated, MCP-native, read-only catalog-of-catalogs above Unity/Polaris/Horizon/Dataplex/OneLake is real but contested whitespace - Snowflake's Select Star acquisition (Nov 2025) [#31] shows incumbents view cross-platform metadata as strategic; a new entrant needs a vertical or mid-market wedge to differentiate vs Atlan.

PrePost GenAI synthesis · 2026

Domain-specific data agents for regulated verticals (vertical ontology + curated benchmark + pre-integrated sources + HIPAA/SOX/PCI + certified eval sets) build durable valuation - Moveworks' $2.85B exit and Sierra's reported $10B show vertical depth compounds; strongest when paired with the mid-market Palantir-alternative play [#152].

PrePost GenAI synthesis · 2026

ServiceNow acquired data.world (May 7, 2025) for a Workflow Data Fabric, and acquired Moveworks for $2.85B (Mar 2025) - consolidation pulling agentic data capabilities into workflow platforms.

ServiceNow · 2025

Databricks acquired Tabular (~$2B, Iceberg) and Neon (~$1B, Postgres for AI agents, May 2025); Snowflake acquired Crunchy Data ($250M, Postgres) - the lakehouses bought operational Postgres to become the system of record for agents.

Databricks / Snowflake · 2025

A data.world knowledge-graph benchmark reported a ~3× accuracy lift (16.7% → 54.2% on GPT-4) when an LLM answered over a graph/semantic layer versus raw SQL - evidence that structured context beats schema-only prompting. Directional: single-vendor benchmark.

data.world · 2024