<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Superprocess</title><description>Essays, teardowns, and engineering notes on running AI inside operations that need approval, recovery, and an audit history.</description><link>https://superprocess.ai/</link><item><title>What a durable process does while nothing happens</title><link>https://superprocess.ai/blog/what-a-durable-process-does-while-nothing-happens/</link><guid isPermaLink="true">https://superprocess.ai/blog/what-a-durable-process-does-while-nothing-happens/</guid><description>Long-running work does not need constant activity, but it does need a durable record of what it is waiting for and where to resume.</description><pubDate>Fri, 11 Sep 2026 00:00:00 GMT</pubDate></item><item><title>How to automate invoice factoring exceptions without giving up the credit decision</title><link>https://superprocess.ai/blog/automate-invoice-factoring-exceptions/</link><guid isPermaLink="true">https://superprocess.ai/blog/automate-invoice-factoring-exceptions/</guid><description>Invoice factoring automation should assemble evidence, apply policy, and route exceptions while an authorised person keeps the credit decision.</description><pubDate>Wed, 09 Sep 2026 00:00:00 GMT</pubDate></item><item><title>What should an approval gate show the reviewer?</title><link>https://superprocess.ai/blog/what-should-an-approval-gate-show-the-reviewer/</link><guid isPermaLink="true">https://superprocess.ai/blog/what-should-an-approval-gate-show-the-reviewer/</guid><description>An approval gate works only when it gives the reviewer the decision, evidence, policy, and consequences without making them reconstruct the case.</description><pubDate>Fri, 04 Sep 2026 00:00:00 GMT</pubDate></item><item><title>What breaks when an agent meets an operation</title><link>https://superprocess.ai/blog/what-breaks-when-an-agent-meets-an-operation/</link><guid isPermaLink="true">https://superprocess.ai/blog/what-breaks-when-an-agent-meets-an-operation/</guid><description>Agent demos succeed while agent rollouts stall. The gap is rarely model quality — it is the four questions a demo never has to answer.</description><pubDate>Wed, 02 Sep 2026 00:00:00 GMT</pubDate></item></channel></rss>