Category Archives: Artifical Intelligence

Agentic AI: No track (yet), no timetable

Posted on 6 October 2026 by dselz

Agentic AI is rewiring value chains by taking over the steps inside business processes, leaving many companies without a usable operating template. When the map is empty, timetable tools like committees and pilots are slow and misfit. The priority is to put builders in charge of laying new track before managers optimize the schedule. Continue reading →

Posted in Artifical Intelligence, Entrepreneur Confidential, PracticalEconomics | Comments Off on Agentic AI: No track (yet), no timetable

Schweizerhalle created dead fish, 5 years later we had an Hazardous Incident Ordinance – we need the same for AI

Posted on 20 September 2026 by dselz

Safety regulation usually follows a visible incident, not foresight. AI failures often leave no smoke, no dead fish, no front-page photograph, so political attention lags. Switzerland could move now with two simple steps: require critical infrastructure operators to declare autonomous AI systems, and ensure agent escape or permission escalation is treated as a reportable security incident, not merely “misalignment.” Continue reading →

Posted in Artifical Intelligence, Business, Entrepreneur Confidential, PracticalEconomics | Comments Off on Schweizerhalle created dead fish, 5 years later we had an Hazardous Incident Ordinance – we need the same for AI

“What has been thought, cannot be un-thought” – we need AI certification fast

Posted on 13 September 2026 by dselz

Voluntary restraint by individual labs will not prevent harmful outcomes from frontier AI, because the real risk is created at deployment. We need an aviation-style certification regime for high-risk AI deployments: independent testing, continuous monitoring, mandatory incident reporting, and enforceable accountability. Without it, insurance and boards will increasingly refuse large-scale agentic systems, and regulators will lose authority by lacking technical capability. Continue reading →

Posted in Artifical Intelligence, Business, Politics, PracticalEconomics, Think Different | Comments Off on “What has been thought, cannot be un-thought” – we need AI certification fast

“No AI” – the “No Software” re-run in enterprise techhnology

Posted on 31 August 2026 by dselz

AI is currently extending the life of legacy enterprise software by smoothing over painful workflows and filling capability gaps. But once automation removes the human checker, data quality and accountability become hard constraints that cannot be bolted onto transaction-focused systems. As in the “No Software” era, the real shift may arrive later than predicted, then happen fast, and the winning slogan will be the one incumbents cannot say back. Continue reading →

Posted in Artifical Intelligence, Entrepreneur Confidential, Think Different | Comments Off on “No AI” – the “No Software” re-run in enterprise techhnology

AI won’t kill legacy software, it may actually keep it alive

Posted on 24 August 2026 by dselz

The 2026 selloff in software stocks looked like the start of a SaaS replacement cycle, until enterprise buyers did the opposite: they renewed. AI is increasingly acting like WD40 and Play-Doh for legacy systems, reducing friction and filling gaps just enough to make switching harder. For AI-first vendors, the wedge is no longer “the incumbent is bad”, it is owning end-to-end workflow, unique curated data, and governance designed in from day one. Continue reading →

Posted in Artifical Intelligence, Entrepreneur Confidential, PracticalEconomics | Comments Off on AI won’t kill legacy software, it may actually keep it alive

Flat screens and enterprise AI, the waiting game…

Posted on 17 August 2026 by dselz

Waiting made sense when flat screens got cheaper every year, because the product was finished the day you bought it. Enterprise AI is different: the real cost is the learning curve, process redesign, and governance muscle you only build by doing. Deferring that tuition does not remove the productivity dip, it just postpones it until competitors are already climbing out. Continue reading →

Posted in Artifical Intelligence, Entrepreneur Confidential, PracticalEconomics, Think Different | Comments Off on Flat screens and enterprise AI, the waiting game…

If AI is a bubble, the real question is: Who survives the bust

Posted on 9 August 2026 by dselz

AI investment is starting to look like the railroad boom of the 1880s: massive buildout, intense competition, and a likely shakeout. The critical question is not whether the bubble pops, but which assets and operators endure when compute depreciates faster than it can pay for itself. Continue reading →

Posted in Artifical Intelligence, Internet, PracticalEconomics, Society, Think Different | Comments Off on If AI is a bubble, the real question is: Who survives the bust

When to trust your Gut, and when not to

Posted on 27 July 2026 by dselz

Two early bets, on maps for local.ch and on retrieval augmented models at Squirro, were validated years later by the market, not by a pitch deck. The real lesson is that intuition is only reliable in environments with repeatable patterns and lots of feedback, and dangerously convincing everywhere else. Continue reading →

Posted in Artifical Intelligence, Entrepreneur Confidential, Think Different | Comments Off on When to trust your Gut, and when not to

Enterprise software won’t be built. It will be assembled

Posted on 7 July 2026 by dselz

Enterprise software is shifting from multi-year builds to fast assembly for a growing class of judgment-heavy processes. The winning pattern pairs a vertical GenAI judgment layer with a horizontal workflow engine that routes work, logs decisions, and makes governance auditable. Continue reading →

Posted in Artifical Intelligence, PracticalEconomics, Think Different | Comments Off on Enterprise software won’t be built. It will be assembled

Which AI Projects Survive When Token Subsidies End?

Posted on 6 July 2026 by dselz

Many enterprise AI pilots look ROI-positive only because model pricing is still effectively subsidized. As token costs shift and agentic workflows multiply usage, the winners will be AI projects priced and managed around measurable outcomes. Stress-test your business case at 3× today’s rates and demand cost-per-outcome, not cost-per-seat. Continue reading →

Posted in Artifical Intelligence, Business, PracticalEconomics, Think Different | Comments Off on Which AI Projects Survive When Token Subsidies End?