Bio

Jelani Richardson designs and evaluates conversational AI at Halyard Consulting. He builds prompt libraries, testing suites, and data workflows that make our chat and voice agents more accurate, inclusive, and reliable. Jelani bridges strategy and delivery by turning stakeholder goals into measurable prompt behaviors, then iterating quickly to ship improvements.

At Halyard, he also supports QA and data curation structuring knowledge bases, normalizing content, and instrumenting analytics so teams can see exactly what’s working and what needs refinement. His work helps clients adopt AI with confidence and accountability.

Focus Areas

Prompt Engineering Conversational AI (Chat & Voice) Evaluation & Red Teaming RAG & Knowledge Base Design Accessibility & Inclusive Language Data Curation & Annotation Bot Analytics & QA

Notable Work

• Prompt library for multilingual intake — Built reusable prompt patterns for service intake and FAQs; reduced handoff friction and improved first-response accuracy. Links: Services · Case Studies
• Evaluation harness for chat & voice agents — Created tests for accuracy, latency, and safety; established pass/fail gates before production releases.
• Knowledge base structuring — Organized source content into grounded, source-linked snippets to lower hallucination risk and speed retrieval.
• Red-team scenarios & escalation rules — Authored safety prompts and edge-case playbooks to route sensitive queries to humans.
• Analytics instrumentation — Mapped conversation funnels and drop-offs; surfaced insights to guide backlog priorities.

Approach

Jelani pairs Halyard’s Agile Methodology, short cycles, clear KPIs, constant feedback with an ethics-by-design lens. He focuses on measurable gains: fewer fall-throughs, faster resolution, and clearer answers backed by sources.