Our engineers write about MCP security, MCP gateways, agent evaluation, AI-governed software delivery and agent architecture. Articles open on LinkedIn.
How should we choose the right one for agentic AI? As AI agents move into production, an interesting architectural question is emerging: do we really need a large reasoning LLM for every decision an agent makes?
Anand NatarajanRead on LinkedIn ↗CLAUDE.md should be the navigation and behavioral contract, not the entire knowledge base. Business context, architecture, specifications, ADRs (Architecture Decision Records), rules and skills should each hold the information they are best suited to provide.
Kamal RajamathivananRead on LinkedIn ↗How the MCP Gateway enforces enterprise controls through one control plane.
Kamal RajamathivananRead on LinkedIn ↗How identity, delegation, capability registries and approval workflows govern MCP capabilities, with different handling for GUI users, applications and AI agents.
Kamal RajamathivananRead on LinkedIn ↗Why MCP security should shift from protecting endpoints to governing capabilities, with trust requirements that vary by caller.
Kamal RajamathivananRead on LinkedIn ↗MCP could become the standard way agents reach software, a stronger foundation than REST for discovery, execution and governance.
Kamal RajamathivananRead on LinkedIn ↗Where standard API gateways fall short for production agents, and how MCP-aware gateways add tool routing, caching, session handling and tool-level authorization.
Anand NatarajanRead on LinkedIn ↗Moving agent identity handling (CMID and DCR) into the MCP gateway to cut overhead, limit identity sprawl and enforce policy consistently.
Anand NatarajanRead on LinkedIn ↗Rubric-based qualitative scoring, aligning LLM-as-a-judge results with thresholds, and retrieval and metadata considerations.
Anand NatarajanRead on LinkedIn ↗Evaluation strategies for different agent types, from structured outputs like SQL generation to planner-executor models and LLM-as-a-judge.
Anand NatarajanRead on LinkedIn ↗Deterministic enterprise logic versus agent reliability, and the evaluation needs of integration architectures from hybrid to planner-executor.
Anand NatarajanRead on LinkedIn ↗Tell us where your operational load is heaviest. We'll show you what Operational Intelligence captures there, and what it lets you stop doing by hand.
Share a few details and we will follow up in 24 hours.
Your message is on its way to the MajorDomo team. We will follow up within 24 hours.
