Intel / video

Total Recall: Agent Memory and Harness Engineering — Ignacio Martinez, Oracle

Source: youtube.com
Author: AI Engineer
Date: Published Sep 16, 2026

Context

An agent's memory design is something its builders can change even when they use a model whose trained weights remain fixed. The published description of the Total Recall workshop frames this as engineering: designing the software around the model to make its work more repeatable.

Its concrete memory problem is several agents writing at once. A file-based system lacks transactional consistency, so the description says it needs a way to prevent their changes from colliding, which makes storage coordination part of agent design. The description also highlights promoting a successful workflow into a reusable, improved . Both ideas place improvements in persistence and procedure within the builder's control, without requiring a newly trained model. This rests on the video's description only; no transcript or demonstration was reviewed.

Key takeaways · AI-distilled

Ignacio Martinez (Oracle) frames a harness as a model wrapped in seven engineering layers: the model itself is 'the frozen part of the reasoning' (you rent the weights, you don't control them), and harness engineering turns that nondeterministic core into repeatable output.
On agent memory storage, files suit a model's instincts and are cheap to append to but lack transactional consistency; for coordinating parallel agents so they don't clobber each other's work, the room's answer reinvents what databases solved: work trees.
The talk borrows 'Umwelt' (Jakob von Uexkull's term for an organism perceiving reality only through its own senses) to describe an agent's : its effective 'lens' on the world is exactly whatever institutional knowledge someone remembered to write down.
A 'hysteresis variable' is proposed as a harness parameter: it encodes how much patience the system extends to a model before deciding to give up on a task, rather than retrying or escalating indefinitely.
'Skill promotion' is described as compressing a workflow that once took four hours of manual work into a faster, reusable version, then retiring the original slow path once the one is proven.

Terms in this piece · Glossary

agent harness — The scaffolding around a model that turns it into a working agent — the loop, the tools it can call, and the rules for when to stop.
agent skill — A reusable instruction file that teaches an agent how to do one job well — the procedure, the tools, and what counts as done.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
distillation — Training a small, cheap model to imitate a big one's outputs, keeping much of the capability at a fraction of the cost.

Read the source www.youtube.com