Arbiters Are The New Aggregators

Hey y’all — many of the most resilient software businesses of the post-dotcom era have turned out to be aggregators that:

1. Have direct relationships with their users

2. Benefit from zero marginal costs in serving additional users

3. Control demand, reducing supplier power

Yes, aggregators like eBay and others existed before but, as I talked about two weeks ago, it wasn’t until the after the dotcom bubble burst that most of the resilient internet companies were built.

You could even say it’s been the “Aggregator Era” for two decades now.

But I believe we’ve either already moved into the next era, or soon will. The next era will be the era of arbiters.

Here’s what they are, and why they’re going to disrupt everything:

Arbiters Are The New Aggregators

What’s an Arbiter?

An arbiter is an AI intermediary that filters, transforms, or autonomously acts on information, replacing direct access with AI-curation.

Unlike aggregators, which centralize and surface information with the goal of maximizing engagement, arbiters filter, summarize, or act to minimize noise and optimize for the user’s preferences.

To make this a little more concrete:

Google is the ultimate aggregator — it ranks and displays an obscene amount of information that’s on the internet (all of its indexed web pages).

However, Perplexity is a Tier 1 arbiter (more on tiers below) that is disrupting Google because it intelligently filters information, summarizes trusted sources, and gives direct answers that eliminate the need for the user to click through to the links Google curates.

Characteristics of Arbiters

Arbiters will all have these three characteristics:

1) User-Centric Control

Arbiters are AI-driven gatekeepers that act on behalf of and exclusively in the user’s interest by filtering, modifying, or discarding inputs first.

Unlike aggregators, which optimize for engagement, arbiters optimize for relevance, efficiency, and cognitive load reduction.

2) AI-Native Judgement

Instead of just organizing or ranking and displaying content (like aggregators), arbiters interpret, modify, or autonomously act on it.

They decide what matters and how it’s presented, with the goal of removing friction, noise, and unwanted influences — effectively shaping what the user is exposed to.

3) Barrier to Direct Access

Arbiters disrupt traditional information delivery by mediating what reaches the user and in what form.

Arbiters fundamentally alter how businesses reach audiences, customers, or leads by severing their direct line of communication in at least one direction. Instead of fighting for user attention, businesses must now convince the arbiter to pass their message through, creating an ecosystem where access is mediated by AI.

Types of Arbiters

Not all arbiters will be created equal. Here are the different types:

Tier 1 — Information Arbiters

The most basic type of arbiters refine and compress information before presenting it to the user, eliminating noise and reducing cognitive load. Most of these have the primary function of filtering and summarizing information.

Instead of sifting through search results, emails, or news articles, the user receives AI-generated summaries tailored to their needs. This disrupts most traditional content models by replacing direct browsing and search with curated insights.

Good examples include:

• Perplexity (today)

• An AI-powered inbox manager that curates a daily email digest

Tier 2 — Action Arbiters

The next level goes beyond filtering and also takes actions on the user’s behalf.

It handles repetitive tasks without, or mostly without, user intervention. It can sort and respond to emails, schedule meetings, manage online orders, or even negotiate refunds within predefined parameters.

Good examples include:

• Inbox managers

• Shopping assistants

• Calendar managers

Tier 3 — Strategic Arbiters

At this level, an arbiter isn’t just executing tasks — it’s optimizing entire systems.

A Tier 3 arbiter manages finances, automates complex workflows, and makes strategic decisions based on guidance (rather than specific, predefined parameters) from the user, and can adapt its behavior and goals over time.

These replace manual planning with AI-driven oversight.

Good examples include:

• AI financial planners

• Autonomous project managers

• AI-driven hiring systems

Tier 4 — Autonomous Arbiters 

The highest level of an arbiter operates with full autonomy, continuously making decisions with minimal user input.

A Tier 4 arbiter could manage an entire business, handle long-term investment strategies, or run personal logistics without constant oversight.

This shifts AI from an assistant to an independent operator, fully realizing the vision of agents.

As of yet, there are no examples I’m aware of that qualify as a Tier 4 arbiter, but they’re coming.

Why Arbiters are a Big Deal

When you’re building a business it helps to have leverage. Aggregators have been among the highest leverage and, therefore, most resilient business models of the last 20+ years.

However, arbiters have higher leverage than aggregators.

If aggregators commoditized suppliers arbiters have the power to erase them from the equation entirely.

For example, Airbnb aggregates home listings by displaying them all in the same format — hosts need to optimize their listing for what performs well on Airbnb — but a travel-booking arbiter would just make the booking without the user even seeing the different options.

This is a fundamental shift for all parties involved:

• For the user, they get a home that meets their preferences without spending time on it

• For the host, they need to optimize their listing for what arbiters like

• For Airbnb, they’ll need to provide information in a way that arbiters prefer, otherwise the arbiter may go around them and book off-platform

The paradigm shift isn’t here yet at a mass-market level, but we’ll begin to see the impact of it soon. Specifically user preferences and business requirements will change:

• From Pull to Push → Users will no longer actively search, browse, or filter through content. Arbiters will deliver curated insights, decisions, or completed actions based on context and preferences.

• From Open Feeds to Gated Access → Raw content streams (search results, emails, news, social feeds) will fade, replaced by AI-selected summaries, interpretations, or direct conclusions — often not including full context.

• From Ads to Negotiations → Businesses won’t just compete for user attention. They’ll have to convince the arbiter to pass their message through, whether via direct integrations, AI-native formats, or sponsorships within AI outputs.

Basic arbiters, like Perplexity’s current product experience, are already out there and venture-funded startups are actively working on Tier 2 and Tier 3 ideas as well.

These feel like inevitable opportunities for big companies to emerge from — users prefer efficiency and results over raw information and, on the other side, business look for ways to automate processes and reduce costs.

As arbiters improve, people will begin to default to trusting them over manually sifting through options, or at least trust them enough to value them higher than the time they’d otherwise spend doing the task themselves.

This is the same behavior shift that happened as they once moved from visiting direct websites to trusting search engines to find the right ones, and from search engines to recommendation feed algorithms.

The next step is no search at all — just answers, actions, and results.

Time will tell how quickly that happens but, when it does, arbiters will become an existential shift where whoever controls the AI-controlled lens shapes the future of all digital-based interactions.

PS — if you’re building an arbiter, let me know! I’d love to hear about it.

How We Can Help

Advertise in my newsletter to get in front of 70,000+ founders.