Currents
Where enterprise software is heading, from the people building it.
Agent Containment Without a Platform Team
Vendor guidance on containing AI agents assumes a platform team. What blast-radius control looks like when your whole engineering org is ten people.
Read post →
Cutting Maintenance to Fund AI Is How the AI Stops Paying Off
CIOs are funding AI by cutting maintenance and legacy contracts. Why the deferred estate sets the ceiling on what the AI investment can return.
The Bottleneck Moved to Review. Nobody Re-Staffed for It.
Code volume tripled and delivery barely moved. Generation was never the constraint. How to staff and sequence for review and verification capacity.
Scoping AI Work When the Expectation Is Already 100x
When the sponsor has already promised a 100x AI multiplier, the honest scope cannot contradict it in the room. Metric selection is the way out.
Before an Agent Touches Production Credentials
Most agent security failures are ordinary plumbing mistakes: secrets, egress, logging. The audit to run before an agent gets production credentials.
Fix the Foundation Before You Bolt On the Agent
Agents amplify whatever estate they sit on. The remediation that has to come first, and how to tell when a legacy estate is ready for an agent layer.
AI Archaeology on a ColdFusion Estate: Evidence Beats Plausibility
AI will read your ColdFusion estate and confidently tell you things that are not true. How evidence-grounded, AI-assisted modernization actually works.
Why Long-Running AI Agents Fail, and What the Harness Has to Do
For multi-hour autonomous work, the model is rarely the bottleneck. The harness that carries state across context windows is. What to look for.
Reading Model Releases Like a Buyer, Not a Fan
A new frontier model ships every few months. A framework for deciding which releases actually matter to your business and which are just news.
Why AI Agents Stall in Production. It Is Not the Model.
Most enterprises now run agents in production, but few see real ROI. The gap is not the model. It is the observability and guardrail layer around it.
ColdFusion Shipped a Native AI Framework. Should Your Legacy Estate Use It?
Adobe put native LLM integration, RAG, and MCP into ColdFusion 2025. Whether your legacy estate should use it depends on where that estate is headed.
The $123K Nobody Was Looking At: What AI-Accelerated Diagnosis Looks Like
A small medical practice, months of unexplained cash decline, and an answer found in hours. What AI actually contributed, and what it didn't.
How to Run an AI Pilot That Doesn't Waste the Budget
Most AI pilots fail before they start, by testing the wrong constraint. How to scope a pilot that produces a decision instead of a demo.
Gartner Says Agents Will Take $234 Billion in SaaS Spend. What Has to Be True First.
Agentic arbitrage is real, but it runs on integration and retained context most legacy estates do not have yet. What the Gartner number quietly assumes.
Claude Opus 5's Launch Weekend Is Part of the Release
Claude Opus 5 shipped Thursday; by Monday the status page showed elevated errors. How a production buyer reads a launch weekend, and what to do this week.
When Your Codebase Is Also Your Database: Untangling a 200GB Repo
Some legacy codebases double as archives: generated reports, binaries, and data living beside source. How to untangle one without a big-bang rewrite.
Choosing AWS vs Azure for a Legacy Windows and ColdFusion Estate
Where a legacy Windows estate should land is decided by ops ownership, licensing, and migration discipline, not by cloud feature comparisons.
What to Ask an AI Consultant Before You Sign
Seven questions that separate AI consultancies that ship production systems from the ones selling decks. Use them before the contract, not after.
Productionizing Someone Else's AI Proof of Concept
The PoC worked, the team moved on, and now it has to become real software. Why that means rewriting more than anyone expects, and how to plan for it.
Agentic Coding in Production: What Actually Holds Up
AI agents can now write real software. What separates the teams shipping production code with them from the teams generating expensive rework.
Rails and Temporal for Batch Workloads: The Pattern, Not the Hype
Legacy batch jobs fail silently and restart badly. Why durable workflow orchestration plus Rails is the pattern we reach for when batch has to be rebuilt.
Migrating a Production ColdFusion App to CF2025: A Field Checklist
Lessons from a real ColdFusion 2025 migration: licensing gotchas, the reporting stack surprise, testing before the DNS swing, and rollback posture.
ColdFusion in 2026: What Still Runs on It and Why That's Not the Scandal People Think
ColdFusion still runs insurance, government, and education systems in 2026. The risk is not the language. It is unmanaged end-of-life exposure.
Your API Keys Just Went Public on GitHub. The Next 24 Hours.
Making the repo private is not remediation. What actually needs to happen in the first day after credentials leak, in the order it needs to happen.
How We Think About ColdFusion Migration
ColdFusion migration is a judgment problem before it is a code problem. How we decide what to move, what to rewrite, and what to leave running.
Is Your Legacy App Worth Modernizing? A Decision Framework
Remediation, modernization, rewrite, or leave it alone. A working framework for deciding what a legacy application deserves, before anyone writes code.
What an AI Readiness Assessment Actually Tells You
An AI readiness assessment is not a maturity badge. It is a diagnostic that names your binding constraint before you spend real money finding it the hard way.
Have a problem worth solving?
Tell us what you are trying to build or modernize, and we will tell you honestly how we would approach it.