Brix AI Agentic Talent Search

Measurable improvements across speed, funding, and recruiter confidence.

The agent-driven workflow I designed delivered meaningful results, with recruiters reporting faster decision-making and greater trust in the candidates surfaced.

Build an AI agentic talent search tool — aligning product growth with AI transformation.

The business goals were to:

1. Expand the user base by positioning Talent Search as a standalone AI agent, lowering adoption barriers for recruiters and agencies while creating a path toward the full ATS and EMS suite.
2. Reposition the platform as AI-native by embedding agentic workflows that improve recruiters' daily efficiency and strengthen its differentiation in the market.

How might we help recruiters efficiently define, discover, and evaluate the right candidates through an AI-driven workflow they can trust?

Through research into recruiter workflows, I identified two persistent pain points: defining what "ideal" looks like for a role is time-consuming and subjective, and existing search tools return results that require significant manual review before any action can be taken.

This created a bottleneck at both ends of the hiring process — before the search began and after results came back.

Connecting the mapping and search agents in a continuous learning loop

I designed the two agents to work as a feedback loop: the Mapping Agent structures recruiter intent, while the Search Agent learns from results to refine future matches — together forming a definition–discovery loop that gets more accurate with each hire.

HMW use AI to help recruiters define and refine Ideal Candidate Profiles?

Recruiters often struggle to articulate exactly what they're looking for — especially early in a role definition. I designed an AI agent that could take a brief, ambiguous job description and turn it into a structured, editable Ideal Candidate Profile (ICP).

My solution: recruiters input a rough job description, and the AI agent generates a refined ICP that can be reviewed, edited, and confirmed before the search begins. This shifted the burden of structure from the recruiter to the system.

Final Design

HMW make the first candidate ranking more accurate and aligned with intent?

Early search results were often misaligned with what recruiters actually wanted — not because the data was wrong, but because intent hadn't been fully captured upfront. A single pass wasn't enough.

My solution was a feedback mechanism that lets recruiters fine-tune results before a full search runs. By rating or adjusting the initial shortlist, the agent learns what "good" looks like for this specific role and recruiter — making the comprehensive search meaningfully more accurate.

Final Design

A scalable conversational design system for agent-driven interfaces.

I needed to define how AI agents communicate, act, and deliver results — across chat primitives, structured outputs, and feedback surfaces. Rather than designing one-off components, we built a system of reusable patterns that could scale across different agent types within the product.

Deeper integration and zero-loss data migration.

The immediate next steps focus on integrating Brix with the broader product suite — including standalone ATS and EMS design — with careful attention to data migration that preserves existing recruiter workflows and history.

Longer term, the feedback loop system opens up opportunities to personalize agent behavior per organization, learning from collective hiring patterns over time.