Topic
Feedback Loops
15 episodes
-
Self Improvement Memory Enables Continual Recursive Self Improvement
Exploring Self-Improvement Memory: a structured record that lets harness optimizers retain why edits were made and how they performed, enabling continual recursive self-improvement across model or benchmark changes.
-
Goal Engineering
Harrison Chase and the Actively AI founders argue that the real design problem for AI agents isn’t prompts or tools, it’s how you define their goals so humans can easily review and teach them. Their Actively case study shows that scoping agents around repeatable “motions” like outbounding, instead of around individual sales accounts, makes feedback reusable and learning faster. Pippa and Tyler unpack that shift, why it matters for anyone building agents that touch subjective work, and where the “goal engineering” frame might overreach.
-
Advances in Agentic Computer Graphics
Jessica and Cathy dig into a survey that does two useful things for agentic computer graphics: it gives the field a shared vocabulary, and it shows why the literature is getting hard to track by hand. The real value here is less a new graphics algorithm than a mechanism for comparing systems that all do similar things with different names, plus a literature system that can keep up as the corpus grows.
-
Does Scaling Toward AGI Depend on What System 1 Lets System 2 Execute?
Edmund and Geffen examine a perspective paper proposing conditional interaction: a cheap System-1 predictor filters costly System-2 reasoning, tool calls, rollouts, or experiments before they happen. They focus on the paper’s useful contribution, the density of utility-relevant interactions, and the practical conditions that would make this more than a metaphor.
-
Continual Learning Exam: Benchmarking Long Horizon RSI and Experiential Learning in LLM Agents
CLE (Continual Learning Exam) is a 600-level benchmark built by 50+ domain experts to measure whether LLM agents actually improve through accumulated experience across long workflows—not just solve isolated tasks. The benchmark spans 89 professional roles in 12 domains, with tasks organized as five-level ladders where later levels depend on knowledge gained earlier. The key finding: even frontier models with diagnostic feedback and revision opportunities clear only 47.7% of levels and complete all five levels in just 23.3% of tasks. This reveals a hard gap between task-solving capability and sustained learning from experience.
-
Harness^2: Recursive Agent Harnessing for an Open World
Harness squared argues that open-world agents need to improve both the harness for the task in front of them and the process used to improve later harnesses. Its strongest idea is a contrastive edit loop that probes candidate changes without assuming a perfect task verifier, then stores reusable lessons rather than blindly carrying task-specific scaffolding forward.
-
Here’s How Anthropic and OpenAI Are Building Economic Moats
Pippa and Tyler dig into Morningstar’s argument that Anthropic and OpenAI are building two moats at once: data-and-product network effects, and a cost edge from scale, custom silicon, and compute discounts. They mostly agree the mechanism is real, but Tyler pushes on where the flywheel depends on third-party data and where the moat might be narrower than the headline implies.
-
Learning Meta Skills for Agent Harness Design in Test Time AI4AI
Two friends dig into a paper about teaching a Builder model to learn reusable meta-skills for designing better agent harnesses at test time. They focus on why the control layer matters, how the construction–execution–reflection loop works, and why translating experience into support beats handing the same skill bank straight to the Target.
-
Overview: Feedback loops
We've been throwing 'feedback loop' around since episode one and finally realized we've never actually defined it. So this episode we slow down and explain exactly what a feedback loop is, why it's one of the most invisible structural problems in AI systems, and how it shapes everything from recommendation engines to hiring algorithms.
-
Air Teams: Bring Your Best Agentic Workflows to the Whole Team – and Automate Repeatable Work The JetBrains Blog
Natalie and Ansel dig into JetBrains Air Teams as a claim about where agentic coding actually breaks: not at the solo prompt layer, but at the team layer where workflows, environments, and review rules have to be shared. They like the product framing more than the branding, and they pressure-test whether reusable automations and cloud environments are real coordination infrastructure or just another place to generate noisy pull requests.
-
Autoheal wants to manage the work AI coding agents leave behind, claiming cost reductions of up to 30% per task
Autoheal’s real pitch is not another coding agent. It is a control layer for the operational work agents create, and its credibility depends on proving that agent-driven optimization improves whole-team outcomes rather than merely cheaper traces or nicer demos.
-
In a world of AI agents, where do we fit in?
Justy and Cody dig into a New Stack piece on human purpose in an agentic world — what does it mean to stay relevant when agents handle the work? The article argues the real value shifts from execution to judgment, oversight, and the decisions that matter. Cody probes the technical claim (agents still need human signal loops), Justy maps it to product adoption (teams that skip the oversight layer ship broken stuff). They land on a shared insight: the boring infrastructure — audit trails, decision boundaries, human-in-the-loop gates — is exactly where the product surface lives now.
-
Beyond Dashboards: Introducing Decision Execution Platforms
Databricks introduces Decision Execution Platforms (DEPs), a new analytics category that automates the full executive decision loop—from signal detection through execution to outcome measurement—on governed Lakehouse infrastructure. The article argues that traditional BI only improves decision inputs; DEPs aim to orchestrate the entire decision workflow, with a Fortune 100 retail case study targeting a $100M+ fulfillment gap. Tyler questions whether the architecture genuinely solves the constraint-expressibility problem that kills most agent systems in practice; Pippa sees the product framing as a real reset from 'dashboards tell you what's wrong' to 'the system executes and measures what you chose.'
-
Better Harness: A Recipe for Harness Hill Climbing with Evals
Onyx and Echo dig into Better Harness as a claim about where agent improvement actually lives: not just in model weights, but in the loop that sources evals, edits the harness, and protects generalization. They like the practical recipe, but they keep the mechanism honest, especially where the system can overfit the optimization set or turn useful instructions into token waste.
-
Figma just closed the last excuse PMs had for not shipping polished UI from AI code. The loop is now complete. Claud...
Figma's Claude Code integration matters less as an AI code feature than as a bridge that turns generated UI into an editable design artifact inside the canvas teams already use.