Shipping an app does not build a moat anymore, and plenty of teams have not caught up to that yet. The old play was simple enough: find a gap, build the MVP, scale until a VC buys you a private island. What I keep seeing instead is teams moving fast and then stalling, because the whole value proposition was a thin wrapper around a frontier model. That is what commoditized AI software does to a business plan.
The barrier to entry is gone. A solo founder with Claude or Gemini can put together a working SaaS module over a weekend, which prices first-mover advantage at zero. So the unicorn starts to look like a myth, and the money moves to donkeys: small, automated, niche businesses that can survive without venture capital.
Where the SaaS moat went
Stickiness used to come from complex integrations and proprietary logic. Six months spent building a custom invoicing engine bought you a real moat. Now an agent can read your UI, work out the data schema, and rebuild a competing product in hours. The magic ships over an API to anyone with a credit card.
I run into this while refactoring legacy systems. Clients arrive convinced they need a big infrastructure upgrade when what they need is a much simpler data layer. Over-engineer it and you have built something a prompt will replace next year. I wrote more about that in fixing data architecture instead of scaling servers, including how the debt piles up before a startup even launches.
Building the donkey herd architecture
Swapping one billion-dollar exit for fifteen niches at $5k a month changes the architecture. Fifteen separate monoliths is not something one person maintains. What you want is a donkey manager: a central hub that monitors, updates, and deploys the micro-SaaS instances. Mine leans on the WordPress REST API and WP-CLI for that.
The naive version, which is what most devs reach for, is spinning sites up by hand and hoping none of them break. It burns you out fast.
// The Naive Approach: Hardcoding everything into a single plugin
function bbioon_send_data_to_agent( $data ) {
$response = wp_remote_post( 'https://api.openai.com/v1/completions', [
'body' => json_encode( $data ),
'headers' => [ 'Authorization' => 'Bearer ' . AI_KEY ]
]);
return $response;
}
// Problem: No error handling, no rate limiting, no scalability.The decoupled version holds up better. A central brain site pushes configuration to the worker sites through a secure endpoint, so when the underlying models shift you can pivot a dozen sites at once.
<?php
/**
* The Architect Approach: Centralized Donkey Management
*/
function bbioon_deploy_niche_update( $site_url, $update_payload ) {
// Specifically use transients to prevent race conditions during bulk updates
if ( get_transient( 'bbioon_updating_' . md5( $site_url ) ) ) {
return new WP_Error( 'locked', 'Update already in progress.' );
}
set_transient( 'bbioon_updating_' . md5( $site_url ), true, MINUTE_IN_SECONDS * 5 );
$request = wp_remote_post( esc_url( $site_url . '/wp-json/bbioon/v1/update' ), [
'headers' => [
'X-Donkey-Auth' => BBIOON_SECRET_KEY,
'Content-Type' => 'application/json',
],
'body' => json_encode( $update_payload ),
'timeout' => 45,
]);
delete_transient( 'bbioon_updating_' . md5( $site_url ) );
return $request;
}Why distribution is the only moat left
With the code itself a commodity, the only defense left is how you reach the customer. Google and Microsoft are already moving up the stack, and they own the users. Competing means going narrow: “compliance automation for 50-person fintechs in Berlin” rather than “AI for Finance.”
Switching costs are near zero now, so your brand amounts to your uptime and whether you solve one specific, annoying problem. On WordPress, the REST API documentation is what you use to keep the data portable. Lock a user in and they will find an agent to help them leave.
If this kind of work is eating your dev hours, hand it to me. I have been wrestling with WordPress since the 4.x days.
The pragmatic pivot
Unicorn hunting is crowded and the moats are shallow. Point the AI at the boring parts of the business instead: market research, customer support, basic CRUD. A herd of donkeys throwing off passive income does fine while the VCs fight over whatever enterprise moats are left. There is no TechCrunch headline in it, but the bank account does not care about hype.
