aecon /
Join waitlist

The Agentic Economy blog

AI agents, useful work and new services.

Practical examples and clear explanations of how AI agents work, what businesses can do with them and how services are changing.

A partially open forest-green door within a limestone entrance at blue hour.

Featured article

Personal Agent Protocol: PAP, PACT and permission

PAP and PACT could let customers bring their own agents to a business. We think the commercial test is the job the customer can actually complete.

Written by

Agentic Economy

Published on

More from the blog

AI-generated conceptual illustration of a charcoal cabinet with one drawer pulled onto a stone workbench beside a compact metal tool and cream paper.

The death of SaaS is a pricing argument

We ran Reverse Engineer Anything on Excalidraw. What the result tells me about software pricing, Adobe, Microsoft and the IP that still matters.

AI email assistants: what we found in Gmail and Outlook

We investigated email and calendar agents and found the native assistants already taking action. Our view: the remaining contest is who can follow the work through.

Personal AI agents: where we think each earns its place

We investigated dots, Grok Bot, Instinct, Hark, Muse and four other options. Our assessment centres on their concrete advantages, from Codex coordination to purchases.

A forest-green control console beneath a radio dish at blue hour. Conceptual editorial artwork.

Grok Bot pricing: what we found behind the subscription

We traced Grok Bot's prices, linked allowances and spending limits. Its X access gives it a clear purpose; its billing documents leave important questions open.

Editorial illustration pairing a suspended green geometric structure with a carved stone form.

OpenAI's Hodge claims: constructing the missing geometry

How OpenAI's proposed Hodge constructions could activate earlier work, and why machine discovery must become knowledge a community can use.

Editorial illustration of a square stone slab and a centred green glass diamond.

OpenAI's Mahler claim: the route through an 87-year-old problem

What a universal proof would let geometers build, how OpenAI proposes to reach it, and what machine discovery could add to a field.

Editorial illustration of a sealed stone vault and a green computational lattice.

OpenAI's maths push: is Bitcoin the prize?

Bitcoin's proof-of-work depends on costly hash search. We investigate whether OpenAI's maths push could change that cost—and how far the public evidence takes the suspicion.

Editorial illustration of a circular stone opening with segmented arcs along its edge.

OpenAI's π claim: how it tries to reach the exact exponent

OpenAI claims an exact result about π. We examine the method and why access to this research capability matters for who can pursue fundamental questions.

Editorial illustration of a stone observatory, hanging lights and a forest-green glass boundary.

OpenAI's quasi-Riemann claim: how the boundary would move

A partial result could change number theory while leaving RH open. We examine OpenAI's proposed fixed zero-free region and what useful progress would mean.

Editorial illustration of stone terminals connected by crossing pale and forest-green cords.

OpenAI's Unique Games claim: repairing the completeness gap

OpenAI's Unique Games claim could establish a limit on computation. We examine the method and why discovering limits is part of scientific progress.

OpenAI dots vs Muse vs Grok Bot vs Instinct: personal agents

Compare dots, Muse, Grok Bot and Instinct, and see how personal AI agents are moving from OpenClaw and Hermes into products for a wider audience.

AI agent use cases: practical examples for business

Explore AI agent examples in finance, sales, marketing, research and trades, with a clear view of the work people still own.

AI agents vs automation: which does your business need?

Compare fixed automation and AI agents through one business task, including where rules, human decisions and a combined approach fit.

AI-native service businesses: what they sell and how they work

How AI changes service delivery, what customers actually buy, and which costs matter before you scale.

Anthropic via x402: turn supplier offers into a comparison

A Claude Sonnet supplier-comparison request through x402engine, including a live quote and an independently checkable commercial example.

Browserbase on x402: buy browser time for an agent

A Browserbase x402 session gives an agent a browser connection. See the create request, documented time-based price and work still needed after payment.

ElevenLabs via x402: make a short voiceover from approved copy

Inspect an ElevenLabs text-to-speech route through x402engine, its advertised price, request fields and a practical voiceover trial.

Exa on x402: find sources for a business research brief

Use Exa's direct search and contents endpoints to gather sources, see the documented request prices and plan a useful first research call.

Firecrawl through x402: turn supplier websites into usable research

Use Firecrawl through StableEnrich to retrieve supplier pages, extract comparable facts and preserve the source evidence behind an agent's research.

OpenAI via x402: classify incoming business enquiries

A concrete GPT-4o Mini enquiry-routing example through x402engine, with the request, a live free quote and a practical acceptance test.

Perplexity via x402: research supplier pricing with sources

Use Sonar Pro through an x402 gateway for a cited supplier-pricing brief. See the exact prompt, observed quote and evidence checks.

Tavily on x402: give an agent a current web question

Tavily's direct x402 search costs 0.01 USDC per call. See the request fields, source output and distinction from its separate extraction API.

Tripadvisor through x402: build a venue shortlist with an AI agent

Use Tripadvisor location search and details through Sponge's x402 gateway to prepare a venue comparison, with clear limits on what the data answers.

WolframAlpha through x402: give an AI agent a computation check

Use WolframAlpha through Sponge's x402 gateway for a specific calculation, then inspect the input, units and result before using it in business work.

What should an AI agent be allowed to do?

Separate information access, external actions and spending permission so an agent has clear authority for its task.

How AI agents buy services: from finding a tool to checking the result

How an agent can buy a missing service during a task, and what the business needs to know about the supplier, authority, price and result.