Our approach
We begin with a situation in someone’s life and ask what useful software could make easier. From there, we build the core experience, examine its limitations, and refine it through testing and review.
Alongside our products, we develop internal systems that help us organize AI-assisted work and evaluate proposed improvements.
A product needs a clear purpose. We identify who it is for, what they are trying to do, and where their current experience becomes difficult. That gives us a basis for deciding what belongs in the product—and what can wait.
We concentrate first on the core interaction. A useful first version should make its purpose understandable and give us something concrete to examine. Additional features need to support that purpose.
We use prototypes, technical checks, and appropriate evaluation to identify limitations and guide revisions. We distinguish between what we intend a product to do and what the available evidence shows it can do.
Development progress does not automatically establish readiness for public use. Before release, we consider the intended experience, reliability, privacy, support needs, and unresolved limitations.
Our own systems
BHV develops its own systems for organizing AI-assisted work and evaluating how that work gets done.
Foundry is our internal operating framework. It organizes responsibilities, priorities, information sources, and review requirements across our ventures, with people accountable for business decisions.
Named Agents are specialized AI agents with defined roles and responsibilities. Our approach allows each agent to identify limitations and propose improvements to its instructions, workflows, and supporting tools.
CAGE—Containment, Authority, Governance, and Evaluation—is our in-house system under development for evaluating and improving those specialized agents and reviewing all AI agent harnesses used across our ventures.
An AI agent harness is the set of instructions, tools, and controls that guide an agent’s work. CAGE is designed to test proposed improvements, check them against defined boundaries, and identify issues before changes are adopted.
An agent’s proposal begins a review process. Proposed changes must meet applicable testing, review, and approval requirements before adoption. The ability to seek improvement does not give an agent authority to expand its own permissions.
Get in touch
Have a question about Boggs Habibi Ventures or one of our products? Send us a message.