Practical decisions for AI and software
Technical guides to choose models, verify software and plan delivery.
Search every article
Search the full article text.
No posts match your search. Try another term or pick a topic.
Latest articles
New comparisons, practical guides and engineering lessons.
Vibe Coding Legal Risks in Austria: Launch Checks
Consent, tracking, renewal reminders and withdrawal flows: turn Austrian legal requirements into checks for your AI-built app.
Commissioning a digital museum souvenir: what should visitors have to do on site?
One visitor scans the code and steps away. Can they still collect? What to put in the supplier brief and how to test the answer in the gallery.
How many tokens does Claude Code’s Code Mode save?
One job used 55,705 fewer input tokens and took 22 seconds longer. Try the terminal flag and a two-file check with a known answer.
Can you train MiniMind from scratch in two hours on an RTX 3090?
The stage table says 2.31 hours. Both trainers default to two epochs. Run the one-epoch recipe and check whether your examples retain any answer tokens.
Is a PaperOrchestra manuscript ready to submit?
A generated CVPR table kept the 49% certified collision rate and dropped the 9% ordinary rate. See how to audit a draft before you submit it.
Can Claude Fable 5.2 resume your saved agent sessions?
A worker restarts between an invoice lookup and the answer. Test which parts of the conversation survive your database, model upgrade and rollback.
A 2026 AI agent development stack, tested on a frontend regression
One click exposes a bug that a first-page screenshot misses. Follow the evidence from a ChunkHound search through a Pi workflow to a repeatable browser check.
How much of your company photo archive can AI sort?
A camera can record where someone stood. A colleague may be the only person who knows which repair they photographed. How to test both cases on a working copy.
Alibaba Open Code Review returned no comments. Which files did it check?
One invoice, two retry keys. See why Alibaba’s CLI selected the code but skipped the test that changed the payment contract.
Essential guides
- AI Agent Pilot in 30/60/90 DaysA serious production rollout plan for Austrian SMEs: scope and de-risk, build and shadow, then limited production and handover. The hard parts are permissions, approvals, evals, and a clean handover, not the model.
- RAG vs Fine-Tuning vs Long-Context 2026A workload-based method for testing retrieval, supervised fine-tuning, long context, and hybrid architectures without relying on universal cost crossovers.
- The Vibe-Code Production-Readiness ChecklistA scannable first-pass checklist for AI-assisted apps before real users touch them. Ten evidence areas, plus risk-based launch triage and the limits of a generic checklist.
- What Software Maintenance Costs After Launch: A DACH SaaS BenchmarkObserved post-launch maintenance cost bands for DACH SaaS: year-one, year-two and year-three spend split across bug fixes, dependency upgrades, cloud, observability, security, compliance, support, product iteration and emergency work, with an interactive calculator.
- Paid Pilot vs PoC vs Design Partner: Which One Proves Demand?Decision guide for choosing the right evidence test: technical feasibility, customer learning or willingness to pay, with a seven-condition paid-pilot scorecard and contract brief.
- Fractional CTO vs Hiring in AustriaCompare the leadership mandate, authority, availability, team needs, employment cost inputs and current supplier terms before choosing a model.
Explore by topic
Find guidance for the problem you are working on.
AI and agents
Engineering, operating and evaluating AI systems, coding agents and model infrastructure.
Product and MVP
Validate demand, shape scope and turn uncertain product bets into testable releases.
Delivery and QA
Architecture, production readiness and quality practices for software that must keep working.
Leadership and teams
Fractional leadership, AI-enabled teams and the operating decisions behind delivery.
Business and regulation
Buying software, funding delivery and navigating governance, contracts and regulation.
Web3 and privacy
Blockchain infrastructure, cryptography and privacy technologies evaluated for production use.
Get the next field note
One concise email when we publish. No tracking pixels, and no inbox filler.
Facing one of these decisions right now? Talk it through with us →