Timothy Wong

Topic dashboard

AI Analytics

Last refreshed July 10, 2026 · 21 concepts

AI Analytics

Analytics is becoming a streaming, agent-driven discipline, and the human-opening-a-dashboard workflow is no longer the centre of the system.

My take

Most data teams are still optimising for a human opening a dashboard, asking a question, and waiting. That workflow doesn’t go away, but it stops being primary. When the consumer is an agent running in a loop, the contract changes - agents want structured answers with provenance, latency, and cost characteristics that look nothing like BI as we built it.

The unit of work in this new world isn’t the dashboard, and it isn’t the data contract - contracts are a 2022 idea. It’s the context layer that makes data legible to agents: business context, org context, interaction history, governance, managed as a first-class platform. Whoever builds this gets compounding returns. Whoever skips it ends up with a graveyard of one-off RAG pipelines that each rediscover the same business logic.

This is a harness problem. Models keep getting better but stay noisy. What compounds is the system around them - context, observability, evals, human review. “Add an LLM to the BI tool” loses to platform rebuild.

What I’m watching:

  • Evals and harness design becoming the core craft - the teams that win will be the ones who can measure agent analytics quality the way we used to measure model accuracy, and iterate on the system around the model rather than the model itself.
  • Context-building paired with observability as a single discipline - you can’t curate context you can’t observe in production, and you can’t trust observability without the context layer underneath it.
  • Whether the dashboard layer goes AI-native fast enough to stay relevant, or gets routed around entirely.

Everything above the divider is mine. Everything below is auto-assembled daily from my knowledge base — individual links and summaries may be stale or off-target. Last refreshed: 2026-07-10.

What’s shifted recently

  • Agentic Analytics Bi Evolution (updated 2026-07-09)
    Agentic analytics is the structural shift in business intelligence from static dashboards and human-triggered queries to AI agents that autonomously plan, retrieve, reason across,… — source · source · source

  • Databricks Snowflake Agentic Data War (updated 2026-07-09)
    The competition between Databricks and Snowflake to become the dominant data platform infrastructure for agentic AI systems. — source · source · source

  • Enterprise Agentic AI Architecture (updated 2026-07-09)
    Enterprise agentic AI architecture is the set of engineering disciplines — orchestration topology, memory layering, context management, cost governance, and safety controls — that… — source · source · source

  • Production Agent Platform Layer (updated 2026-07-09)
    The production agent platform layer is the set of shared infrastructure that sits between model APIs and business applications in deployed multi-agent systems — covering memory an… — source · source · source

  • Agentic Data Infrastructure Substrate (updated 2026-07-08)
    An agentic data infrastructure substrate is a unified, centrally governed data platform designed to be the single source of truth for agent reasoning. — source · source · source

  • Enterprise Agent Integration Layer (updated 2026-07-07)
    The enterprise agent integration layer refers to the set of CLI tools, APIs, protocol adapters, and platform extensions that established enterprise software vendors are building t… — source · source · source

  • Sap Autonomous Enterprise 2026 (updated 2026-07-05)
    The SAP Autonomous Enterprise is SAP’s strategic platform announced at Sapphire 2026 in Orlando on May 12, 2026. — source · source · source

The ideas I keep coming back to

Currently active (last 30 days):

  • Agentic Analytics Bi Evolution — Agentic analytics is the structural shift in business intelligence from static dashboards and human-triggered queries to AI agents that autonomously plan, retrieve, reason across,…
  • Databricks Snowflake Agentic Data War — The competition between Databricks and Snowflake to become the dominant data platform infrastructure for agentic AI systems.
  • Enterprise Agentic AI Architecture — Enterprise agentic AI architecture is the set of engineering disciplines — orchestration topology, memory layering, context management, cost governance, and safety controls — that…
  • Production Agent Platform Layer — The production agent platform layer is the set of shared infrastructure that sits between model APIs and business applications in deployed multi-agent systems — covering memory an…
  • Agentic Data Infrastructure Substrate — An agentic data infrastructure substrate is a unified, centrally governed data platform designed to be the single source of truth for agent reasoning.
  • Enterprise Agent Integration Layer — The enterprise agent integration layer refers to the set of CLI tools, APIs, protocol adapters, and platform extensions that established enterprise software vendors are building t…
  • Sap Autonomous Enterprise 2026 — The SAP Autonomous Enterprise is SAP’s strategic platform announced at Sapphire 2026 in Orlando on May 12, 2026.
  • Enterprise Agent Implementation Labor — Enterprise agent implementation labor is the emerging category of work — and workers — created specifically by the gap between agent capabilities and production-ready enterprise d…
  • Apache Superset Vs Incumbent Bi — Apache Superset is an open-source, self-hosted business intelligence platform positioned against proprietary incumbents (Sisense, Looker, Tableau) on grounds of cost efficiency, c…

Established:

  • AI Bi Data Agent Interface Shift — The shift from static, chart-based business intelligence dashboards to autonomous AI agents that reason over data, iterate across queries, and execute decisions in natural languag…
  • Embedded Agentic Analytics Saas — Embedded agentic analytics is the pattern of delivering AI analytics capabilities — conversational business intelligence, AI-generated dashboards, and autonomous analytics agents…
  • Enterprise Agent Observability OTEL Tracing — Enterprise agent observability is the discipline of instrumenting every step of an autonomous agent’s execution — tool calls, retrieval operations, subagent invocations, and inter…
  • Stanford AI Index And Jagged Frontier — The jagged frontier describes AI capabilities as uneven across tasks: models simultaneously achieve near-human performance on structured benchmarks (WebArena 74.3%, OSWorld 66.3%)…
  • Local First Workflow Engine Deterministic Agents — Local-first workflow engines for AI agents are self-hosted, open-source platforms that execute multi-step agentic pipelines entirely within the operator’s own infrastructure — no…
  • Agent Control Plane Execution Layer — The agent control plane is the infrastructure layer that sits between an organization’s business systems and its AI execution layer, enforcing cross-cutting operational concerns —…
  • Streaming Multi Agent Observability — Streaming multi-agent observability is the discipline of instrumenting continuously-running, event-driven AI agents — agents that operate on persistent data streams rather than di…
  • Agent CI CD Trust Boundary Expansion — AI coding agents deployed in CI/CD pipelines inherit the trust model of interactive developer tools — where a human is present to validate actions — but operate in headless, autom…
  • Agent Roi Decision Intelligence Pricing — Agent ROI and decision-intelligence pricing are two sides of the same procurement problem: enterprises must justify agentic AI spend against metrics that automation-era frameworks…
  • AI Coding Acceleration Stack Order — AI coding agents do not accelerate all software work equally.
  • AI Native Banking Infrastructure — AI-native banking infrastructure refers to ground-up financial backend systems — ledger, payments orchestration, compliance, and AML — built from the start to expose AI agents as…

Who I’m watching

No people or companies crosslinked to this area yet.

Sources I’ve been drawing on