Agent-Native Second-Order Lens
Redesign infrastructure when a new kind of user becomes dominant.
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 95%
The Agent-Native Second-Order Lens examines what must change after a new technology introduces a fundamentally different user. In this episode, people delegate browsing, comparison, research, and transactions to agents, making the agent an increasingly important consumer of websites and applications. The framework starts with the triggering innovation, identifies activities transferred to the new actor, and compares that actor's needs with the assumptions built into current infrastructure. Teams then locate mismatches in identity, permissions, data access, navigation, pricing, verification, and transaction design. Those gaps reveal product requirements and startup opportunities that are easy to miss when attention remains fixed on the agent itself. The lens is repeatable for any innovation that changes who interacts with a system.
Origin
Extracted from Marketing Against The Grain
Core principles
- 01Major innovations create indirect infrastructure needs.
- 02Identify who or what becomes the new primary user.
- 03Study how that user consumes data and performs actions.
- 04Do not assume human-facing interfaces remain the main interface.
- 05Look for both incumbent reinvention and new-company opportunities.
How to run it
- 1
Identify the new actor
Define the technology-enabled user that may become a major participant in the system.
Pro tip Describe its capabilities, goals, constraints, and who authorizes it.
Watch out Do not assume the new actor behaves like a faster human.
- 2
Map transferred activities
List the browsing, analysis, communication, purchasing, or operational tasks moving from people to the new actor.
Pro tip Start with workflows already showing repeated adoption.
Watch out Speculative activities should be separated from observed behavior.
- 3
Compare interaction needs
Examine how the actor wants to authenticate, consume data, evaluate options, and execute actions.
Pro tip Include machine-readable data and verifiable permissions in the analysis.
Watch out A human-friendly page may still be hostile to reliable agent use.
- 4
Locate infrastructure gaps
Identify where current websites, social networks, APIs, and policies fail to serve or govern the new actor.
Pro tip Consider safety and accountability gaps alongside usability.
Watch out Optimizing access without authorization controls can create serious abuse risks.
- 5
Choose reinvention opportunities
Decide which gaps incumbents should address and which could support new products or platforms.
Pro tip Prioritize gaps tied to growing, high-value workflows.
Watch out Do not rebuild infrastructure before validating that the behavior is durable.
In the wild
If recruiting agents increasingly evaluate candidates, professional networks may need structured, verifiable ways for people to expose work samples, authorship, credentials, and permissions. A platform could provide machine-readable portfolios while preserving human presentation and privacy controls.
→ Candidates become easier to assess accurately, and agents receive more reliable evidence than scraped profile prose.
Common mistakes
Designing only for the human interface
Keeping all information trapped in visual pages ignores how agents retrieve, compare, and verify data.
Ignoring governance
Agent-native access without identity, consent, permissions, and accountability can amplify misuse.
Is it for you?
Best for
Product leaders, founders, and platform teams planning for widespread autonomous-agent adoption.
Not ideal for
Short-term feature prioritization where the new actor has little adoption or no measurable effect on current users.
From the transcript
“it creates a second-order effect where like you need all new infrastructure on the internet.”
“Websites need to work differently, social networks need to work differently.”
“Because you have a new user on the internet that's basically becomes the predominant user of most of the websites and apps, which is an…”
From the episode
This FREE AI Agent Does a Team’s Work in 35 Min (Manus AI)