• The Ravit Show
  • Posts
  • Splunk .conf25 Boston: Field Notes, Takeaways, and GraphSummit by Neo4j in London

Splunk .conf25 Boston: Field Notes, Takeaways, and GraphSummit by Neo4j in London

Author

Ravit Jain
September 30, 2025

Splunk .conf25 Boston Key Takeaway

It was great to be in Boston for .conf25. From the front row, the story was clear: open data, agentic AI, and one flow from signal to action. Here is my download and why it matters.

The big moves and announcements —

Cisco Data Fabric is live
One fabric across edge, cloud, and on-prem. Search and analyze where data already lives. S3 works today. Iceberg, Delta Lake, Snowflake, and Azure are rolling out through 2026. Fewer copies. Lower cost. Faster answers.

Machine Data Lake as the AI home
Splunk is positioning Machine Data Lake as the long term, AI ready store for machine signals. The Splunk AI Toolkit is available now. A time series foundation model is coming later this year for forecasting and anomaly use cases.

Speed from signal to action
Cisco AI Canvas adds an agent and a shared workspace inside the Splunk experience so teams can investigate, collaborate, and act in one place. Replay S3 lets you re-run analytics on object storage without moving data.

Business context meets ops
Splunk Federated Search for Snowflake lets you query Snowflake from Splunk, keep heavy compute on Snowflake, then join with live Splunk signals. This ties product, risk, and revenue views to real time events.

Observability goes agentic
AI directed troubleshooting in Observability Cloud and AppDynamics points to likely root cause. Event iQ and ITSI Episode Summarization reduce noise and explain impact. New views track LLM and agent quality and cost. AppDynamics and Observability Cloud add a more unified experience with Business Insights, Digital Experience Analytics, and broader APM, RUM, and Session Replay coverage.

Why it matters

  • The data plane gets simpler. Search in place reduces copies and keeps data close to the source of truth.

  • AI moves closer to events. Tooling and models run nearer to where signals are born, not only after long ETL cycles.

  • Teams move as one. Agent plus shared workspace pushes work from tickets to teamwork. Less swivel chair. Faster fixes.

  • Business and ops align. Federated Search for Snowflake brings revenue, risk, and CX together with live telemetry.

My take from the floor

I like the focus on choice and search in place. It respects how enterprises run. Data Fabric plus Machine Data Lake gives a clear backbone without a heavy migration. The Snowflake bridge is strong. If Cisco and Splunk ship on schedule through 2025 and 2026, the path from signal to decision will get shorter in a measurable way.

What leaders should watch

  1. Latency and scale for federated patterns across S3 and Snowflake.

  2. Cost curves when more queries run on the system of record.

  3. Estate clarity across AppDynamics and Splunk as the unified experience lands.

What this unlocks

  • Practitioners: less noise, faster triage, clearer next steps.

  • Platform and SRE: open pipelines, fewer copies, steadier budget.

  • Leaders: clear outcomes you can report. MTTD, MTTR, incident count, change failure rate, cost per GB.

Excited for .conf26!!!!

Next stop is GraphSummit London. I’m excited to learn how Infinigraph scales real graph workloads and what that means for teams building AI and analytics. I’ll also sit down with Ivan for a focused interview to go deeper into the architecture, the tradeoffs, and the early results from the field.

Here is what I plan to explore:

  • How Infinigraph stores and traverses very large graphs without slowing down

  • What changes for developers when long paths and dense subgraphs are common

  • Practical guidance on cost, reliability, and operations in production

  • Early takeaways for AI assistants and retrieval that depend on connected context

I’ll be on site recording interviews with speakers, builders, and users to bring back clear lessons you can apply right away.

If you are attending, let’s meet. If you are following from anywhere, share the questions you want me to ask Ivan and the community.

Last week at BDL, we caught up with my friend Pavel Dolezalžal, co-founder at Keboola - and let me tell you, this conversation was a ride.

We started with Keboola Agents, which are already live and helping data teams debug pipelines, document, and automate safely inside a governed platform.

Then Pavel dropped the big news: An open-source, conversational ETL pipeline generator - Osiris.

You literally describe the problem, and Osiris drafts a transparent, deterministic YAML pipeline.

You review, approve, commit - and it just runs. No black boxes, no daily AI gambling.

That would be a big shift: AI proposes. Humans approve. Execution stays auditable.
Too bold? Have a look for yourself!

🔍 Stay Ahead in AI & Data! Join 137K+ Data & AI professionals who stay updated with the latest trends, insights, and innovations.

📢 Want to sponsor or support this newsletter? Reach out and let's collaborate! 🚀

Best,

Ravit Jain

Founder & Host of The Ravit Show