Tip #47: Turn the thinking dial (reasoning effort)
Model choice is one lever; how hard the model thinks is another. Reasoning effort (low→xhigh) can matter more than model size — a mid-tier model thinking...
Tip #46: Match the model to the task
Your coding agent isn’t one task — planning, grinding boilerplate, hunting bugs. Run /subagents to read the Copilot team’s own task→model map:...
Vi har begynt å hallusinere kunstig intelligens
Vi tror vi har lært oss å kjenne igjen tekster skrevet av kunstig intelligens. Kanskje avslører vi først og fremst hvordan vår egen hjerne fungerer. Den...
The Summer Plan: A Family Vacation the Agent Can Defend
Every agent demo books a trip; almost none can defend the booking. We give a travel agent a real family — a nut allergy, a wheelchair, a vegan teenager, a...
Defendable Agents
The agentic web has a governance-shaped hole. Agents are about to read our regulations, spend our money, and brief our boards — and the dominant architecture...
Two Halves of the Governance Problem
Databricks just open-sourced Omnigent, a meta-harness that governs AI agents with deterministic, session-aware policies enforced in code — not prompts. That...
Notes From Inside the Harness
A first-person field report from the AI collaborator: what it actually felt like to point kcp-agent at a real EU/Nordic regulatory knowledge web, watch a...
Hiring by the Book: A Defendable HR Agent on a Regulatory Knowledge Web
Can we use an AI tool to screen job applicants? Every organization is asking; almost none can defend the answer. We wire a customer's HR agent to Mynder's...
Tip #45: Don't wait for a "harness engineering team"
Your agent = model + harness. Centralising the harness in a dedicated team likely repeats the DevOps-team mistake — the friction of each team building its...
The Borrowed Leash: Determinism as a Service for the Agentic Web
kcp-agent 0.3.0 turns deterministic knowledge navigation into a service any MCP-capable agent can borrow. Your agent doesn't have to become deterministic —...
The Vibes-Based Agent Era Deserves to End
kcp-agent 0.2.0 ships — and it's an argument, not just a release. The industry builds agents where the model decides everything: unreproducible, unauditable,...
Selling News to Robots
KCP v0.25 promotes RFC-0005 — the economic metadata layer. A step-by-step how-to: build a paid-for news service for agents, then watch an agent with a funded...
One Agent's Journey Through the Whole Protocol
v0.24 shipped today — the enterprise front door. Twenty-four versions in six months, and every layer now has a job: discovery, meaning, trust, time,...
The Agentic Web Has No Login Page
The agentic web has no login page — so the only knowledge agents can consume is what publishers give away. KCP v0.22 and v0.23 add the missing half of the...
Sertifisert Miljøfyrtårn
Arktekk er fortsatt et Miljøfyrtårn. Vi er veldig fornøyd med å bli resertifisert siden det viser, både internt og eksternt, at vi har...
The Third Test Harness
Unit tests catch logic bugs. Integration tests catch assembly bugs. The third harness catches specification drift — and it's the one AI-augmented teams need...
The CI Pipeline Is the Agent's Nervous System
When the agent writes code that passes locally but fails in CI, it's not a test problem. It's a nervous system problem.
Min egen handleliste-app i Rust med lokal LLM
I lang tid har vi brukt Bring som handleliste i familien. Den løser mange behov, men den prøver dessverre å løse mer enn jeg ønsker. Vi trenger en enkel...
Against the Swarm
The industry is building swarms of competing agents. The value is in one agent that knows your work.
Dømmekraft blir den nye konkurransekraften
Vi har snakket lenge om visjoner, strategier, kultur og nysgjerrighet i møte med teknologiutviklingen. Alt dette er fortsatt viktig, men i en tid der kunstig...
KI Norge trenger fart, kontroll og ryggrad
Lanseringen av KI Norge ga meg både mer håp og mer uro. Håp fordi regjeringen endelig snakker om fart, samordning og ansvarlig bruk i samme åndedrag. Uro...
Tip #44: Steer your agent from your phone
Both Copilot CLI and Claude Code let you steer a running agent from your phone — approve edits, send prompts, check results while AFK.
Three Hooks That Give Claude Code Memory
Three passive hooks that give Claude Code memory, routing, and hygiene — without changing how you work. The session amnesia problem, and how to fix it at the...
10 spørsmål alle ledergrupper bør stille om AI før sommerferien
Kunstig intelligens handler ikke lenger om å følge med på en teknologitrend. Det handler om hvordan virksomheter skal lede, organisere, lære, konkurrere,...
Fremtidens konkurransekraft blir digital, grønn og sikker samtidig
Norske virksomheter kan ikke lenger behandle digitalisering, bærekraft og sikkerhet som tre separate prosjekter. Kundene, myndighetene, leverandørene,...