Harness has rolled out a refreshed user interface that centers on an AI‑agent orchestration model and introduces three supporting concepts: the upcoming Harness Software Factory, a Vibe Mode policy layer for AI‑generated code, and a Flex Pricing credit pool. The changes aim to shift DevOps pipelines from hand‑crafted scripts toward machine‑speed, specification‑driven automation, which directly impacts engineers responsible for delivery, security, cost, and runtime governance.
New UI and Software Factory Overview
The updated console presents a consolidated view of the four core AI agents that now drive software delivery, security testing, runtime security, and cost management. Through the Software Factory, teams can define specifications that the agents enforce, reducing the need for ad‑hoc token usage and providing a consistent control surface for AI‑augmented workflows. Vibe Mode adds a policy enforcement layer for citizen developers who generate code with AI tools, ensuring that generated artifacts comply with organizational standards before they enter the pipeline.
AI Agent Architecture and Knowledge Graph
At the heart of the platform is a knowledge graph that tracks 567 distinct entity types and processes roughly 29 billion context updates each month. This graph supplies the contextual data required for the four primary agents to delegate work to sub‑agents, enabling autonomous task completion without human‑written scripts. The architecture is designed to replace traditional script‑based orchestration with AI‑driven decision making, allowing the system to operate at “machine speed.”
Operational and Security Implications
Practitioners should note several practical effects:
- Script reduction: The platform’s goal is to eliminate custom scripting for workflow construction, which can simplify maintenance but also requires confidence in the AI agents’ correctness.
- Token consumption control: By enforcing specifications, the Software Factory can limit the number of AI tokens consumed during code generation and validation.
- Policy enforcement via Vibe Mode: Organizations gain a point of control over AI‑generated code, mitigating the risk of unchecked changes entering production.
- Automation levels: Harness envisions a spectrum from fully autonomous agents to human‑augmented assistance, meaning teams must decide the appropriate level of automation for each use case.
- Flex Pricing credit model: Instead of licensing individual modules, teams purchase a pool of credits that can be applied across the platform, affecting budgeting and cost tracking practices.
Related CloudNinjas coverage: DevOps.
What This Means For Practitioners
Engineers should start by mapping existing pipeline steps to the four core AI agents to identify where automation can replace scripts. Define clear specifications for code quality, security, and cost that the agents can enforce, and test Vibe Mode policies on a sandboxed set of AI‑generated changes. Monitor the knowledge‑graph update rate to gauge the freshness of context data, especially when scaling to higher PR volumes (e.g., 15‑20 PRs per week per developer). Finally, evaluate the Flex Pricing credit pool against projected usage to ensure cost predictability as the platform’s AI components become more central to delivery.