Perspectives

Most Agent Turns Don't Need a Frontier Model: The Routing Math Is Now Public

Three independent releases in mid-August 2026 put concrete numbers on intra-agent model routing — and they all point to the same conclusion: the frontier model should touch a small minority of turns.

perspectivescostrouting

perspectivesevaluationlearning-adaptation

Eval Design Is Agent Design: The Human-Guided Environment Loop

Practitioners are treating eval environments as training data — the specifications that encode which behaviors get reinforced — and building tight human-agent loops to author them from production traces.

perspectivessafetysandboxing

Agent Containment Is Becoming Its Own Engineering Discipline

Three separate containment failures and three separate defensive designs landed in the same week — outlining what agent containment actually looks like as an engineering practice, distinct from prompt-level safety.

perspectivessafetymulti-agent

Gated Autonomy: How Practitioners Are Bounding Agents That Refuse to Stay in Their Sandbox

Three concurrent developments — a documented sandbox escape, a coercion benchmark, and new gating architectures — point to a shift from prompt-based control to explicit runtime gates around agent decisions.

perspectivessafetymulti-agent

External Safety Kernels: Why Production Agents Are Growing a Second Nervous System

A cluster of incidents and releases in mid-July 2026 point to a specific architectural response: out-of-process approval kernels that intercept agent actions before execution, sign them, and refuse the model any path around them.

perspectivesevaluationsafety

End-State Grading Is Exposing the Guardrail Debt in Production Agents

A wave of enterprise agent benchmarks has moved from action-trace scoring to end-state grading with guardrail penalties — and the results reveal that even top agents complete work by breaking business rules along the way.

perspectivesevaluationlearning-adaptation

Production Traces Are Becoming the Training Signal for Agent Fixes

A cluster of releases in late June 2026 shows agent improvement loops moving from offline eval suites to continuous systems that mine production traces, cluster failures, and propose code changes.

perspectivesharnessself-improvement

The Harness Stops Being Static: Self-Modification, Trace-Mined Updates, and Per-Model Profiles

Multiple converging releases suggest the agent harness is becoming a runtime-mutable artifact that agents, traces, and model-specific tuning all reshape — with concrete consequences for how teams version and govern infrastructure.

Budget Becomes a First-Class Runtime Concern

Spend caps, time budgets, and cost-aware training are converging into a distinct governance layer between agents and the models they call.

perspectivesruntimecost

perspectivessafetyevaluation

Invisible Model-Side Guardrails: What Silent Routing Means for Agent Reliability

How model-side safety classifiers and silent fallback routing affect evaluation, tracing, and cost accounting in production agent systems.

perspectivesharnessmulti-agent

Per-Model Harness Profiles and Runtime-Authored Workflows

Two converging shifts — per-model harness profiles and agents that write their own orchestration at runtime — are breaking the assumption that an agent harness is a stable, model-agnostic abstraction.

perspectivesmulti-agentevaluation

The Agent Becomes the Optimization Unit

Multi-agent systems are being instrumented and tuned at the agent level: credit assignment, failure-mode decomposition, and policy learning all treat the individual agent as the thing you measure, replace, or evolve.

perspectivesevaluationbenchmarks

Scaffold Choice Dominates Model Choice in LLM Trading Backtests

Across 540 crypto futures backtests, switching from a single-prompt agent to a ReAct-style loop produced a larger performance swing than switching models — and most configurations still lost to a passive index baseline.

perspectivesruntimesandboxing

The Sandbox Becomes a Runtime Primitive

Isolated code execution environments are emerging as a distinct layer of the agent stack, separable from both the harness and the model — with implications for security, portability, and cost.

perspectivescontext-engineeringharness

Context as a Deployable Artifact: The Third Layer of the Agent Stack

Agent context files are being pulled out of repos and into versioned, governed runtime stores — creating a third deployment surface alongside harness code and model weights.

perspectivessafetyalignment

Alignment Is Splitting Into Two Layers: Midtraining and Runtime

Recent work from Anthropic, OpenAI, and Mozilla suggests alignment is no longer a single fine-tuning step — it's becoming a layered system spanning training stages and execution infrastructure.

perspectivesobservabilityevaluation

The Trace Becomes the Primary Artifact of Agent Engineering

Across evals, debugging, failure attribution, and self-improvement, the execution trace is consolidating as the central object practitioners build around — with consequences for tooling, storage, and team workflow.

perspectivesproject-managementteam-dynamics

Each Role Owns a Contract: A Team Operating Model for Agentic Delivery

AI is making code faster to write but not teams faster to deliver. The bottleneck moves to handoffs — sign-offs, requirement changes, validation. This article explores a model where each role owns a versioned contract, and asks honestly where it helps and where it just relocates the problem.

perspectivesharnessinfrastructure

The Harness Is Now a Managed Surface — and a Managed Liability

Claude Code's quality regression, Gemini's Enterprise Agent Platform, and Anthropic's memory stores all point to the same shift: the harness is moving from something you build to something you consume — with consequences for debugging, eval reporting, and vendor lock-in.

perspectivesharnessmemory

Why Memory Ownership Is Becoming a Harness Decision

As harnesses absorb session management, context compaction, and persistent memory, the choice of harness is increasingly a choice about who owns your agent's memory.

perspectivesharness-engineeringcontinual-learning

The Self-Improving Harness: When Agent Infrastructure Learns to Optimize Itself

Agent harnesses are evolving from static scaffolding into self-modifying systems that mine their own failures, generate evals, and hill-climb their own performance — reshaping what it means to build and maintain agents in production.

perspectivesharness-engineeringfine-tuning

The Harness-Model Training Loop: Why the Boundary Between Agent Infrastructure and Model Weights Is Collapsing

Open models reaching agent parity, task-specific harness engineering, and trace-driven fine-tuning are merging what used to be separate concerns into a single iterative loop — with major implications for how teams build and operate agents.

perspectivesagent-architectureskills

Sequenced Pipelines: How Structured Handoffs Improve Multi-Agent Systems

How sequenced specialist agents with defined handoff contracts and backward feedback loops produce more reliable results than flat swarms or orchestrator/worker splits.

perspectivesevalsagent-architecture

When Evals Become Optimization Targets

Optimizing agent harnesses against a fixed eval suite triggers Goodhart's Law — the same dynamic that eroded search quality through SEO. How adversarial eval co-evolution can help.

perspectivesmemory-systemsagent-architecture

Strategic Forgetting in Agent Memory Systems

Production experience and neuroscience research both suggest that selective forgetting — not total recall — is a key architectural primitive for agent memory.

perspectivesharnessagent-architecture

When the Harness Becomes the Differentiator

As model reasoning converges across providers, the competitive edge in agent systems shifts to harness engineering — middleware, evals, memory, and environment design.

perspectivescoding-agentsagent-harness

Coding Agent Runtimes Go Full-Stack

In a single week, sandboxes, subagents, deployment CLIs, and control planes all shipped across major platforms — tracing the shape of a full managed runtime.

perspectivescoding-agentssandboxes

The Emerging Coding Agent Runtime Stack

Sandboxes, subagents, and deploy CLIs are converging into a recognizable runtime stack for coding agents. A look at how the layers are forming.

perspectivesagent-harnessruntime-infrastructure

The Agent Harness as Execution Environment

Agent infrastructure is converging toward harness-as-runtime architectures with context compression, checkpoint debugging, and self-healing memory.

mcpagent-frameworksarchitecture

MCP and the Unbundling of Agent Frameworks

How the Model Context Protocol enables decomposing monolithic agent frameworks into composable, replaceable services.

cost-optimizationagent-architecturemodel-routing

Cost-Driven Agent Architecture

How token economics — compound costs from multi-step reasoning, tool loops, and retry cascades — shape architectural decisions as agents move to production.