Strategy & transformation6 MIN READ
What a polished agent demo does not tell you
A demonstration shows one possible path. A buying decision needs evidence about representative work, failure, authority, recovery and operation.
THE SOFTLANDER BLOG
April 2026
Softlander blog perspectives published in April 2026.
In April 2026
21 articlesArchitecture & engineering6 MIN READ
Idempotency for AI agents: preventing one instruction from becoming three actions
Retries, duplicated messages and uncertain tool outcomes can multiply an agent’s effects. Idempotency turns a logical action into one controlled business operation.
Cryptographic engineering6 MIN READ
Hybrid key exchange: what it protects and what it costs
Combining a classical and a post-quantum key exchange is now the default for TLS. Understanding why the combination is used—and what it adds to every handshake—makes the deployment decision straightforward.
Evaluation & assurance6 MIN READ
How to design an independent agent benchmark that remains useful
A durable agent benchmark needs a declared construct, protected cases, reproducible execution, uncertainty, governance and limits on the claims its scores can support.
Governance & privacy6 MIN READ
The responsibility map every AI partnership needs
A useful responsibility map connects each AI decision and artefact to a named accountable owner, a performing role, required contributors and evidence of completion.
Security engineering6 MIN READ
Secure boot is a key management decision
Teams treat secure boot as a firmware feature and discover years later that the hard part was never verification. It was deciding who holds the signing key, how it rotates, and what happens when it is lost.
Strategy & transformation6 MIN READ
How to prioritise an enterprise portfolio of agent opportunities
Rank agent opportunities by decision value, workflow evidence, feasibility, consequence and reuse—not by enthusiasm, executive volume or demo polish.
Evaluation & assurance6 MIN READ
How should we evaluate an agent that works for days rather than minutes?
Long-running agents accumulate state, dependencies and chances to drift. Evaluation must test checkpoints, recovery, changing conditions and complete outcomes across time.
Security engineering6 MIN READ
Secure code review that changes the code
A review that delivers a ranked findings report usually changes nothing. A review that delivers a patch, a test that fails before it and a pattern the team can apply changes the codebase.
Evaluation & assurance6 MIN READ
Design agent evaluations backwards from the decision
Start with the choice an owner must make, then derive claims, evidence, cases and measures. The result is smaller, sharper and easier to defend.
Governance & privacy7 MIN READ
How to build an evidence register for an AI system
An evidence register should connect each claim and requirement to versioned artefacts, review decisions, known gaps and the conditions under which evidence expires.
Governance & privacy6 MIN READ
Autonomy is a spectrum of actions, not a product label
Describe autonomy operation by operation. A multidimensional action map gives leaders a better basis for limits, oversight and release evidence than one marketing label.
KEEP EXPLORING
Take the thinking
into your next discussion.
Read the method, assumptions and questions to ask.