Skip to content
Lencioni & Associates
Abstract illustration: bold navy planes and a copper underline on a composed surface, standing for editorial writing rather than a literal page.

Writing

Insights

Notes on infrastructure, identity, AI adoption and IT operations.

  1. Abstract illustration: a sparse constellation of pale nodes linked into one shape against deep navy, with a single copper point among them.

    12 min read

    Configuration as an Asset

    Thirty years of configuring systems by hand left the one thing the business depends on in people's heads. Identity as code moves it into the system. Here is the cost model with every variable named, an Okta-to-Entra ID migration done two ways, and why as-code is the precondition for letting AI anywhere near the identity platform.

    IdentityIdentity as CodeZero TrustAI Operations

  2. Abstract illustration: pale nodes above forced through one copper gate before reaching ordered navy blocks below, standing for identity as the control plane every path must cross.

    4 min read

    Identity is the control plane for AI agents

    An AI agent authenticates, holds credentials and takes actions against your systems. That makes it a principal, and principals belong in your directory — not in a configuration file owned by whoever set the integration up.

    IdentityZero TrustAI Agents

  3. Abstract illustration: two bold navy fields joined as equal peers by a two-way copper bridge with a shield mark at the midpoint, standing for hybrid architecture rather than compromise.

    4 min read

    Hybrid AI isn't a compromise — it's the architecture

    Running models both in a vendor's cloud and on systems you control is usually described as a concession to compliance. For most mid-market organizations it is simply the correct design, and it should be chosen deliberately rather than settled for.

    Hybrid AIArchitecture

  4. Abstract illustration: five copper decision marks on a single path between an origin node and a destination, standing for the questions that must be answered before a pilot reaches production.

    4 min read

    Five questions to ask before your first AI pilot

    Most AI pilots that stall do not fail technically. They fail because nobody agreed in advance what data was in scope, who owned the result, or what would count as it having worked.

    AI AdoptionGovernance