Volume 343, Issue 2 · 7 articles from 2 issues
Harness cut AI API classification costs about 1,200x by replacing a generative LLM decision step with its Jev model.
Harness rebuilt a production AI API classification pipeline by replacing the generative LLM judgment step with Jev, a purpose-built decision model.
The change cut per-endpoint classification cost by roughly 1,200x versus the original pipeline and 150x versus a single-shot LLM, and Jev's calibrated probabilities placed all errors in a 0.3-0.8 uncertainty band covering 3.4% of endpoints.
With zero mistakes at the tails and a stable four-field output contract, Harness could swap the decision engine without changing downstream consumers.
Harness engineers describe how software delivery knowledge graphs unify fragmented SDLC data for reliable AI reasoning.
Harness engineers explain how knowledge graphs can unify fragmented software delivery data across tools by defining explicit entities, typed relationships and schema-as-code contracts.
The architecture combines schema-as-code, unified synchronization and declarative queries so AI agents can reason deterministically, resting on five pillars of trust: shared semantics, explicit contracts, data quality measured by fill rates, provenance for lineage and automated governance.
The post stresses that graph quality equals data quality, which requires monitoring, continuous evaluation and production trace observability.
Harness walks through generating a CI pipeline from a GitHub repo and pushing a Docker image without writing YAML.
A Harness tutorial shows how to connect a GitHub repository and have Harness CI auto-generate a pipeline without writing YAML.
Using a Java Maven project with 42 unit tests and a Dockerfile, Harness detected the stack and generated build, test and cache steps, and the author added a push step to DockerHub.
The full build, test and push pipeline ran in just over three minutes, and the post cites a 2025 Stack Overflow survey finding that 84% of developers use AI tools, increasing pressure on CI.
Harness tutorial shows how to set up a rolling Kubernetes deployment with rollback using Harness CD.
This walkthrough shows developers new to Harness CD how to build, configure and execute a rolling deployment of a sample application to Kubernetes via a Harness Delegate.
It contrasts ad hoc deploy scripts and SSH-based releases with a reusable setup that separates what is deployed, where and how, so promoting from dev to prod becomes a settings change.
The tutorial deliberately deploys a broken build to demonstrate one-click rollback and deployment history, and requires an SCM account, Docker Hub, a Kubernetes cluster and a Harness account.
Harness shows how native approval steps and policies in Harness CD remove waiting and glue code from release sign-offs.
Harness argues approval gates rarely fail because the check is wrong; they fail because someone is waiting on a person.
With Harness CD, the approval step lives inside the pipeline, providing sign-off without manual waiting or custom glue code, while configurable policies help reduce manual effort and improve consistency.
The post walks through how teams can streamline approvals to accelerate releases.
Harness explains how AI coding agents can be exploited through trusted files, hooks and MCP tools, and how to defend them.
Unlike autocomplete, today's AI coding agents can run commands and act on their own, which Harness says makes trust the new attack surface.
Attackers can plant hidden instructions in trusted files, hooks and MCP tools rather than in the chat box, and the post explores how these vectors enable persistent threats.
It offers guidance on securing agent workflows against such hidden instructions.
Harness says AI-generated code demands better quality signals, combining Test Intelligence and code coverage in CI.
AI-generated code is increasing the number of changes flowing through CI builds, and Harness argues faster CI needs better quality signals, not just faster execution.
Code coverage shows what tests exercised and where gaps remain, and Harness Builds combines Test Intelligence, code coverage and quality signals.
The goal is to help developers and AI agents run the right tests and validate changes faster.