Grid by LimaCharlie is now in beta: Agentic SecOps for the stack you have

Grid is LimaCharlie's agentic AI layer for security teams that want AI operations running across their existing stack right now. Security providers and SOCs need access to AI capabilities without waiting for a migration window, a contract renewal, or a vendor to ship the features they need.

Every major security vendor is offering some version of AI. CrowdStrike has Charlotte AI. SentinelOne has Purple AI. Microsoft has Copilot for Security. Each claims performance gains within its own ecosystem, and each works within the telemetry its vendor controls. 

For the MSSP managing 80 customers across four different EDR configurations, or the enterprise with two years left on a CrowdStrike contract, vendor-scoped AI covers one slice of the environment while leaving the rest unaddressed. The result is a collection of advisory features, each focused on its own vendor's data, with no shared operational layer across the full environment.

Grid is LimaCharlie's answer to this structural problem. It brings governable, operational AI capabilities to your security stack as it exists today. Grid runs in parallel with your existing tools and requires nothing from your current vendors to do so.

On May 19th, Grid entered beta.

An agentic AI layer for your current security build

Grid connects directly to your telemetry sources: your EDR, SIEM, cloud logs, identity providers, and any other API-enabled output. Agentic AI operators inherit the same API access as human analysts, are governed by the same access controls, and their actions are logged and reversible. 

Your existing vendor contracts stay intact. Nothing about the current stack needs to change. Think of Grid as a parallel pipeline, not an integration project. You don't migrate to the platform, you connect it alongside what you already have.

Through Grid's agent builder interface, security teams can configure AI agents tailored to their environment and workflows. You control what each agent can see, what it can act on, and where human approval is required. All agents run on Claude Code, with every action logged and fully auditable.

You can inspect, fork, and modify what any operator does. You have full control of the operational abilities of AI in your security environment. There is no black box.

Bridging the cross-vendor orchestration gap

Gartner's 2026 Hype Cycle for Agentic AI identifies cross-vendor orchestration as a primary gap in the current market, noting that most agentic offerings "limit it to a single development environment or limited to one's own development platform." The same report introduces the term "agent washing" to describe legacy automation tools repackaged as AI agents, and observes that "most products in the market are AI assistants, not AI agents." 

Grid is built specifically for the environment those observations describe: heterogeneous stacks that per-vendor AI cannot span, and buyers who have been sold advisory features under an agentic label.

No rip-and-replace required

The dominant assumption in our industry is that agentic AI requires a clean, unified stack. Migrate your SIEM. Standardize your EDR. Remap your indexes. Then, once everything sits inside one vendor's ecosystem, you can fully access the AI capabilities your team needs.

For most security teams, following this path is either impossible or years away. MSSPs can't force customers onto a single EDR. Enterprises are locked into multi-year agreements with vendors that haven't shipped adequate AI capabilities yet. Both groups are being asked to wait for a migration window that may arrive too late.

Grid starts from a different premise: security teams that built excellent stacks shouldn't have to discard them to access the next generation of operations. The agentic layer runs in parallel with your current stack, pulling from the same raw sources, and acting through the same APIs. Your incumbent vendors do not need to cooperate or even know Grid is there.

AI forward deployed engineers

Most vendor created AI interactions are transactional. You prompt, you get output, you move on. The context disappears and the next session starts from zero.

Grid's AI Forward Deployed Engineers work differently. Users can create AI forward deployed engineers (FDE) that act as persistent operators assigned to own a specific task or outcome inside your environment. You create it once. It stays active, monitors progress,  creates agents, and manages the task.

Here's a concrete example. Say you need to continuously monitor privileged and service accounts for unusual behavior. Rather than prompting Grid each time you need eyes on these accounts, you stand up an FDE and point it at the problem. The FDE creates the worker agents that watch for anomalous access patterns. It checks in on them every few hours. If detections stop firing, if results look anomalous, or if something in the infrastructure is blocking the workflow, the FDE flags it and investigates. It owns the outcome, not just the task.

You can interact with your FDEs directly from a dashboard. Pull one up for a detailed troubleshooting session or ask it clarifying questions about its task. The FDE has full context on everything it has managed since its creation.

The difference between a one-time agent prompt and an FDE is the difference between asking someone to run an errand and hiring someone to manage a function. One gets you an output. The other gets you a system that keeps working.

What beta access includes

Beta participants can connect their existing data sources and run Grid's specialist operators against live telemetry right now. Grid runs on Claude Code, and customers connect their own subscription to power their operators. You pay your AI provider directly at cost. LimaCharlie does not add a surcharge.

For MSSPs, multi-tenancy is fully supported from the start. A single Grid instance can span your entire customer base. Provisioning a new tenant is an API call. Detection rulesets, sensor configurations, and operator permissions replicate programmatically across any number of customer environments.

Pricing during beta is ingest-based, with no per-alert or per-playbook compounding. At MSSP scale, per-alert pricing makes agentic AI economically unworkable. Grid's model doesn't compound with alert volume. 

If you're managing a heterogeneous customer base or sitting on a vendor contract that won't let you move yet, beta access is the right starting point. Get started with agentic operators now rather than waiting for the right procurement window.

Request beta access at limacharlie.io/grid. Connect your first data source in under twenty minutes.