AI cutting the need for Economy of Repetition?

I became a serious supporter of Economy of Repetition and Economy of Trust (learn once > trust > use everywhere > in all roles) when we rolled out BankID services back in the early - 90s. The one-time pass codes were frequently used, secure and trusted in banking – and the habit and trust could serve all identification and signing needs in society-at-large.

Now I see that business and personal wallets will receive, forward and verify not only identification but all sorts of verifiable credentials (sets of these needed in business and life events). With ever shorter attention span in these hectic times the Economy of Repetition and Trust in this all-encompassing space will get an even exponentially more valuable cognitive load cutter…

Wallets will also enable employing and empowering AI-agents to do legally binding actions. Not only find data and educate us like LLMs are doing so eminently already. Then my question:

Will LLMs and later wallet-carrying AI-agents take away much of the need for Economy of Repetion?

Edited ChatGPT’s new wievpoints:  they’ll reduce the need for repetition in some layers of the economy—but they won’t eliminate it. They’ll relocate it.

1) What “economy of repetition” actually does

Repetition isn’t accidental—it serves three core economic functions:

A. Cost amortization

• High fixed cost → low marginal cost (software, media, manufacturing templates)

• Repetition spreads the cost

B. Trust & standardization

• Repetition creates predictability (brands, processes, credentials)
• “Same thing again” = lower risk

C. Coordination efficiency

• Shared routines reduce cognitive load
• Society runs on repeated patterns (payments, contracts, workflows)

2) What wallet-carrying AI agents change

Agents with:

• their own wallets
• ability to transact
• ability to act continuously
• personalization capability

…introduce near-zero marginal cognition and execution.

That breaks a key assumption behind repetition:

Humans are scarce and expensive → reuse outputs.

Now:

Cognition + execution becomes cheap and continuous.

3) Where repetition declines

A. Mass-produced “good enough” outputs

• Marketing copy
• Basic software components
• Customer service scripts
• Templates

Agents can generate on-demand, personalized variants instead of reusing one version.

👉 Result:
Shift from “repeat the same thing” → “generate per instance”

B. Intermediation layers

Agents can:

• compare prices instantly
• negotiate
• execute transactions

So repetitive roles like:

• brokers
• middle-layer coordination
• routine procurement

…get compressed.

C. Behavioral routines

Humans prefer repeat actions because of:

• memory limits
• friction
• habit

Agents remove that:

• auto-payments
• auto-decisions
• auto-optimization

👉 Less repetition at the human behavior layer

4) Where repetition persists (or grows)

A. Infrastructure & protocols

Agents need stable systems:

• payment rails
• identity systems
• APIs
• legal frameworks

These rely heavily on standardization → repetition

👉 Repetition moves down the stack

B. Training & optimization loops

AI systems improve via:

• repeated interactions
• feedback loops
• reinforcement

👉 Massive increase in machine-level repetition

C. Trust anchors

Even with agents, humans still care about:

• brand
• reputation
• guarantees

Repetition still signals:

• reliability
• accountability

D. Economies of scale don’t disappear

Physical world still matters:

• manufacturing
• logistics
• energy

These still benefit from repetition and scale.

5) What actually changes: the location of repetition

Think of a stack:

6) New equilibrium: “dynamic repetition”

We don’t eliminate repetition—we get:

Repetition of patterns, not instances

Instead of:

• same product repeated

We get:

• same logic repeated
• same model reused
• same protocols reused

👉 Example:

• Not one fixed insurance product
• But millions of dynamically generated policies using the same underlying model

7) Second-order effect: collapse of “average user”

Repetition exists partly because we design for averages.

Agents enable:

• per-user optimization
• per-transaction customization

👉 The “average case” becomes less relevant
👉 Repetition as “one-size-fits-all” declines

8) The constraint that keeps repetition alive

Even with agents, three constraints remain:

1. Trust

People still prefer known, repeated structures

2. Coordination

Society needs shared conventions

3. Compute + energy costs

Even if low, they’re not zero → reuse still matters

Bottom line

Wallet-controlled AI agents will:

• ❌ Reduce repetition in:

  • outputs
  • human routines
  • templated services

• ✅ Preserve or increase repetition in:

  • infrastructure
  • protocols
  • learning loops

• 🔁 Transform repetition into:
  “repeatable systems generating non-repetitive outcomes”

So - we have to go very practical. Take business wallets and AI in use - and empower AI-agents where needed. Using is Learning.