Skip to content
Fritz Gerald ZephirinMenu

Observations · AI and Business Since 2022

Customer support

The first business function with documented, at-scale labor substitution. Klarna's assistant doing the work of 700 agents (2024) and Salesforce cutting support from 9,000 to ~5,000 roles (2025) made the new economics undeniable: AI handles the routine majority, humans handle exceptions and high-value moments.

21 observations · 5 key sources · 20222026

What drove it

GPT-4-class models via API, Salesforce Agentforce, purpose-built support platforms (Intercom Fin, Decagon, Sierra), and the Brynjolfsson field study showing novice agents gained the most from AI assistance.

The evolution

2022

The first obvious application: ChatGPT could already answer most tier-one questions. The open question was who would productize it.

2023

Claude and GPT-4 reached production via API, Klarna shipped one of the first ChatGPT plugins, and every support platform started building.

2024

Klarna's '700 agents' disclosure made the new economics public, Salesforce launched Agentforce, and GPT-4o's real-time voice opened phone support.

2025

From assistant to agent: Salesforce cut support from 9,000 to ~5,000 roles, and Stanford documented employment decline among young workers in exposed support occupations.

2026

AI-first support became the default posture as the layoff wave continued; reliability gains like GPT-5.5's halved hallucination rate pushed containment higher.

  • Q1–Q2 2026

    The AI layoff wave: payroll becomes capex

    The first sustained period in which profitable companies explicitly tied workforce reductions to AI adoption and capex funding.

  • April–May 2026

    GPT-5.5: OpenAI's flagship becomes an agent

    OpenAI's flagship line moved decisively from chat toward computer-operating agents — with reliability, not capability, as the headline metric.

  • June 15, 2026

    Salesforce buys Fin for $3.6B on a 76% resolution claim

    The largest consolidation yet in AI customer support, and the number at the centre of it deserves reading carefully. A resolution rate measures volume that never reached a human. It does not measure whether the customer's problem was solved, whether they came back three days later, or whether they simply gave up. Those are different questions, and only the first one is easy to instrument — which is precisely why it is the one that gets quoted.

  • June 17, 2026

    DeepL buys Mixhalo, moving translation toward live interpretation

    Machine translation has been encroaching on written work for a decade. Live simultaneous interpretation has held out, because it runs in real time, under social pressure, with no edit pass and no second attempt — the conditions under which tacit skill matters most. Buying a live-event audio platform is a statement about which direction the frontier is moving.

  • June 29, 2026

    Concentrix cuts guidance while AI deals grow 400%

    Read the two halves of the release against each other. The AI product line is growing fast in percentage terms off a small base; the services business still sets the guidance, and the guidance came down. This is what the middle of a service-to-software transition looks like on an actual income statement — not a clean pivot, but a fast-compounding new line that cannot yet carry the company, disclosed in the same document as a cut.

Key sources

The papers, announcements, and reports that document this industry's shift. Each also lives in the Library.

What businesses do differently

Support became a deflection-rate and resolution-quality discipline rather than a headcount discipline. Companies route by complexity, measure AI containment honestly, and rebuild knowledge bases as machine-readable sources of truth.

What this means for founders and leaders

Support economics changed permanently — plan for AI-first support from day one, with human escalation as a designed experience rather than a fallback. For BPO and CX businesses, the move up the value chain (quality, judgment, integration) is existential.