ITProGuru
Dan Stolts started in IT in 1988. For over a decade of that career he wrote ITProGuru for the IT community. Now he is building Just In Time AI, and documenting that build here for the AI era.
From enterprise IT to AI systems -- built in public since 1988.
See Dan Stolts' current work at danstolts.com Hire Dan Stolts -- CTO, Chief AI Officer, VP Engineering
Recent posts
New AI-era lessons alongside the enterprise-IT archive that built this site's reputation. Browse the full archives to tell them apart by year, category, or tag.
- How do I find what is slowing down my AI workflow?Use AI to maintain an evidence-backed investigation of CPU, memory, storage, models, agent concurrency, and cloud limits—and verify the result.
- How do I verify a ticket is actually ready before a team picks it up?Read the complete authoritative ticket and run a six-layer, read-only readiness review before pickup: PASS only when every applicable layer clears, FAIL with every observed gap, ...
- What Windows security settings actually matter if you don't have a dedicated security hire?Five: disk encryption, disabling legacy SMBv1, switching sign-in to a device-bound PIN, turning on Controlled Folder Access, and requiring Network Level Authentication on Remote Desktop -- ...
- How I Actually Build Durable Memory for an AI AgentThe hands-on version of agent memory: the index-plus-files layout I actually use, the frontmatter fields that make a memory file findable and trustworthy, the recall-then-verify discipline ...
- How I Batch My AI Prompts So I'm Not Paying to Re-Read the Same Context Six TimesThe hands-on version of the single-prompt batching rule: the exact prompt shapes I use, the CLI flags and system-prompt guardrails I set so an agent stops ...
- Why My Agent's Context Window Fills Up Faster Than I ExpectThe practitioner version: why an agent session runs out of working memory long before you expect it to, what I actually watch to see it coming, ...
- What I Actually Check Before I Trust an Agent's Tool CallThe hands-on companion: the specific checks I run before I let an agent's tool call through, the naming discipline that keeps a permission gate honest, and ...
- How do I scope an AI review panel that can actually be wrong?Scope the panel with a frozen evidence revision, a distinct read-only reviewer, an explicit return shape, a forward test before authorization, and a separate accountable owner ...
- How to record a durable approval a merge cannot fake?A runnable approval gate should reject merge status, stale hashes, wrong approvers, expired decisions, silence, and tampered records.
- How Do I Turn a Rule Into an Enforced Control?A written rule becomes a useful control when an objective predicate runs on the required path, denies the prohibited state, records evidence, and is protected from ...
- How do I give an AI agent memory that survives a new session?A runnable Python example retrieves a current instruction and rejects changed content, broken correction links, cycles, due reviews, and expiration.
- How do you get productive with a coding agent fast?Three habits: write a brief before you write a prompt, scope every request to one task the agent can verify on its own, and read every ...
Browse 225+ topics