D-1933 Agentic AI Clint Blog OG 2

AI-First: Agentic AI needs a new architecture

Last edited: October 14, 2025

At Cribl, we’ve talked a lot about epochs. A moment in time when there was a before and after. AI, and specifically agentic AI, is an epoch. The way we work is going to forever change. There have been many such events in our lifetimes: the PC, the Internet, and the smartphone. AI will change how we work forever. Prior to the PC, there were people whose jobs were literally titled “computer”. Spreadsheets were columns on paper updated by armies of “computers” who manually updated rows of columns manually with adding machines. Entire job classes were eliminated, and entire new industries were spawned by the personal computer.

unnamed.png

Scene from “The Apartment” (1960), directed by Billy Wilder, starring Jack Lemmon & Shirley MacLaine

Similarly, AI will change how we all work. Agents, autonomous systems that analyze, reason, act, and learn across your digital environment, will change jobs and workflows for all of us. Cribl is on a mission to unlock the value of all of your data, specifically targeted at our core persona of IT and Security professionals. Today, our core users serve as the glue between human-generated context like tickets, wiki documents, chat threads and telemetry data. Take an alert and use their “meat computer” (brain) to translate their learned context into queries for their telemetry system to find malicious actors and resolve system issues. In the coming years, agents will change their work from being that glue to being the conductor of autonomous agents who amplify them from expert to a true 10X investigator.

But here’s the catch: the world’s telemetry infrastructure wasn’t built for agents.

Legacy platforms are optimized for humans typing in terms, like searching in Google but for their telemetry. Analytical queries are bolted onto a system designed to find needles in haystacks. Agentic workloads are going to destroy the systems built with the assumptions of the 2000s and 2010s. They simply weren’t built for it, in terms of scale, performance, economics, or ease of use in working with the gnarly data that is logs, metrics, and traces.

Cribl is.

We’ve spent years building the world’s best data engine for IT and Security, designed for the volume, variety, and velocity of telemetry data, with the ability to deliver high-performance infrastructure instantly and affordably at scale. That means Cribl is the platform ready to support both human and agentic workloads with our agents, your agents, or anyone’s agents, all running efficiently, contextually, and safely.

The Problem: Data Graveyards in an Agentic World

Legacy tools treat telemetry as data to be read and visualized. They collect it, index it, and pray someone searches it later. That model worked when humans were the only ones consuming this data. But in an agentic world, machines are now analysts, and they don’t wait around for dashboards. They interrogate data continuously, correlate context across systems, and demand the ability to investigate dozens to hundreds of hypotheses quickly and performantly. It’s essential in their mission to surface critical insights, generating orders of magnitude larger workloads on telemetry infrastructure.

Legacy systems will collapse under the load. Most companies are already running their analytics infrastructure at 80% resource consumption and have allocated as much of their datacenters and cloud environments to observability and security as they’re able and willing. Every query burns compute. Every index drives cost. Every “maybe someday” log line sits in a pricey tomb. The very architectures built to store data now prevent anyone, human or machine, from actually using it.

Worse, analysts lack the full view of an incident. That only comes from telemetry data and data from human-generated sources like Slack, Jira, Confluence, or GitLab to connect what changed, when, and by whom as they’re diving in to figure out what’s going on. Without both machine and human-generated data, users not only have to contend with the glacial pace of analysis, but manually cobble together information across sources. That’s no way to power agents or get answers.

The Vision: Agentic Telemetry

Agentic Telemetry is an architecture that fuses human, machine, and AI-generated context into one unified data layer. It’s not just logs and metrics anymore. It’s code commits, change tickets, workflows, runbooks, agent activity, and more all stitched together automatically, ready for both human and machine reasoning. Cribl’s vision is to make it possible not just to move data, but to understand it. A data engine that automatically parses, structures, and normalizes telemetry on ingestion, using AI models trained on years of real-world data patterns. One that automates the data engineering toil that today takes entire teams of humans and months of manual effort. Deliver the same value in moments and focus your time and energy on impact, not infrastructure.

The future is here. We will stop spending human hours engineering data whose value is not yet known. The system will structure it for you as it flows in. Once data is structured, agent-ready, and queryable in real time, you can say goodbye to expensive full-text indexing and say hello to fast columnar analytics and meaningful answers to your and your agents’ toughest questions. 

This is what Cribl’s engine was built for: to unlock value from all your data. Agentic telemetry is a natural evolution making data instantly usable, at any scale, by both humans and agents.

Cribl’s Agent and Yours

Cribl.Cloud is already laying the foundation. It's the first AI-first architecture that delivers on the promise of being schema-agnostic by design, open and federated across data stores, and structured for AI agents to operate at scale. All of this while delivering maximal value for the dollar.

We’re going deeper and building for the agentic reality. Cribl’s own agent, currently in development, is trained to be vendor agnostic and understand any data, not just data locked into a proprietary silo. Federation must start by being programmatic and deterministic before adding an agent on top. We believe a close coupling of the semantic layer, with rich understanding of the data, along with the agent which semantically understands the data will drive massive productivity gains. However, as an open company we will meet customers where they are and where their data is. Our lakehouse architecture will provide the best price and performance for the dollar, but our agent will operate on data stored anywhere. It will combine human insights, AI reasoning, and machine telemetry into one cohesive system that can explain, predict, and act.

But we’re not keeping that power to ourselves. We know choice matters now more than ever. Through our implementation of a Model Context Protocol (MCP) server, any third-party agent can run safely on Cribl’s data engine foundation. MCP gives external agents context-rich, governed, performant access to telemetry data.

So whether it’s your in-house automation, a reasoning assistant, or the next-gen SOC copilot, they can all run on top of Cribl securely, efficiently, and at full context fidelity.

Why It Works

  • Built for Agents. Cribl’s architecture handles the 10x or more query volume agents will generate without falling over. Federated, elastic, and cost-optimized.

  • AI-Structured Data. Ingest once, parse automatically, normalize instantly. No more raw text drudgery or schema drift hell.

  • Composable Intelligence. Human, AI, and machine data fused in one layer: one place to query, one timeline, one truth.

  • Open Access via MCP. Let any agent (not just ours) operate on your data safely and contextually.

The Payoff

  • Agents that actually work, because they can access structured, normalized, real-world infused context on scalable, elastic infrastructure. 

  • Humans who see instant causality, because context is automatically fused, not manually stitched together.

  • Budgets that finally make sense, because you’re not feeding gigabytes into overpriced indexes that agents choke on anyway.

The Bottom Line

AI isn’t coming to replace humans; it’s coming to replace bad architecture.

Agentic Telemetry gives you the foundation to embrace this once-in-a-generation shift: open, federated, intelligent, and agent-ready by design.

Beyond being the world leader in telemetry pipelines, Cribl’s mission has always been to unlock value from all IT and Security data in service of whatever comes next. And what’s next is agentic.

The agentic era is coming for observability and security. And Cribl is the telemetry data engine built for you – and your agents.

Cribl, the Data Engine for IT and Security, empowers organizations to transform their data strategy. Customers use Cribl’s suite of products to collect, process, route, and analyze all IT and security data, delivering the flexibility, choice, and control required to adapt to their ever-changing needs.

We offer free training, certifications, and a free tier across our products. Our community Slack features Cribl engineers, partners, and customers who can answer your questions as you get started and continue to build and evolve. We also offer a variety of hands-on Sandboxes for those interested in how companies globally leverage our products for their data challenges.

More from the blog

get started

Choose how to get started

See

Cribl

See demos by use case, by yourself or with one of our team.

Try

Cribl

Get hands-on with a Sandbox or guided Cloud Trial.

Free

Cribl

Process up to 1TB/day, no license required.