Posts by Rene Zander
@Rene Zander
René Zander
Enterprise AI Architect | Production-Ready AI Agents for CRM & ERP
Posts by René Zander
The demo went well. The model answered correctly, the stakeholders nodded, someone said "let's get this into production." That was six months ago. The PoC still runs, the roadmap still lists it, and it is no closer to shipping than it was on demo day...
Every organization that lets an AI agent ship code has a sign-off step: a name on the release, a reviewer marking the pull request approved, a ticket moved to done. A human is accountable for that call, always. What most teams cannot answer is whethe...
Reliability in an AI agent is a harness property, not a model property.
The cleanest proof arrived at the bottom of the model-size ladder: a 688 MB model controlling a smart home, showcased by the model's own maker. The part worth studying is the 25...
The obvious way to improve a coding agent is to make it more capable: a stronger model, a wider context window, more tools, more room to act on its own. That is not where my problems come from. My agents seldom fail because they reason badly. They fa...
Half of the memory you are about to hand-write for your agent is already sitting in your codebase. The other half, no indexer will ever find.
Both gaps feel identical from the agent's side. It opens every session knowing nothing about your systems, ...
Your AI agent fails decisions for the same four reasons a bad manager does. A bigger model fixes none of them.
Not because the model is dumb. Because nothing in its loop forces it to widen its options, look for evidence it is wrong, or check itself ...
My AI agents could finish any task I handed them. Not one of them could tell me the task was a waste of a month.
That gap was never about model quality. It was about which layer I aimed them at. I had handed over execution: write the draft, run the ...
Knowledge is not flat. It has an address book, and the closest door comes first.
What ran and worked in your environment beats what you wrote down. What you wrote down beats what a teammate remembers. What a teammate remembers beats the top search r...
We are building AI agents with a fundamental architecture flaw.
A recent study tested six frontier models across 2,000+ sessions. Each agent was instructed to complete a specific process step before finishing. Every single model agreed. And every si...
Your task manager is the best agent memory you're not using.
Not because vector databases are bad. Because the store everyone builds for their agent starts rotting the day they stop feeding it. And the one knowledge base you feed every single day, y...
The AI agent used to be the star of every demo.
Now it's on the shutdown list. Not because the model got worse.
The most valuable asset in your AI program is in none of the quotes you ever signed.
A demo is a showroom. Good light, everything polis...
Build your AI skill once with your best model. Then run it on a model that costs a tenth as much until the next flagship ships. The output will not drop.
That sounds like a downgrade. It is not. It fixes the two things that make AI agents painful ri...
Some of the rules in your CLAUDE.md should not be rules at all.
Not because they are wrong. Because you have written the things you cannot afford to lose into a file the model reads once and then slowly forgets. A preference survives that. A constra...
I changed one model string in ten cron jobs last night. 4.7 to 4.8. Then I went to bed.
The benchmark threads can wait until morning. My agents can't. They fire whether I'm awake or not: a briefing at 6:30, follow-up drafts at 11:45, a sync at 4 AM ...
You added a skill last Tuesday. The agent hasn't called it once. Each new skill silently weakens the discovery odds of the ones you already have.
You assume it's a description problem. It isn't.
Everyone's pushing past 50 skills now. Vercel ships a...
TL;DR — why browserground, not the other 2B grounding models
You already know the hybrid-AI argument: don't pay frontier-vision rates for "where is the button?" There are three good 2B specialists for that job — UI-TARS, ShowUI, browserground. Here...
The first stage of AI work is prompting.
The last stage is removing the model from most of the workflow.
That sounds backwards.
It is not.
When a workflow is new, the LLM is useful because the work is still ambiguous. You are discovering what goo...
Forrestchang's andrej-karpathy-skillshttps://github.com/forrestchang/andrej-karpathy-skills CLAUDE.md is four rules aimed at the moment Claude is writing code. They work. What they don't cover is the moment Claude is running. Once a Claude-driven pip...
chevron_left