AI Agent Design & Operations
Designing goal-oriented AI agents for qualification, response, conversation guidance, and handoff generation. Expert in intent classification (confirmation/question/objection/decline/pending), edge case handling, and prompt engineering with strict JSON output formats. Building SDR responders, CRM summary generators, and human handoff systems. Separation of outputs: external for client vs internal for team (actions and next steps).
Agentic Systems Orchestration
Designing multi-step flows with sub-agents (intent → summary → parse → actions). Human-in-the-Loop (HITL) approach with approval/rejection workflows, guides for human agents, and quality control. Operational governance: defining quality rules, what gets saved as persistent memory, standardizing outputs to be parseable and auditable.
Production Automation
Building robust automations with Zapier/Make/n8n with conditional logic paths. State management and consistency: preventing row duplication in Google Sheets, structured reliable logging, Sheets as "source of truth". Optimized step design to save tasks/costs. Real-world pipelines: retries, intermittent errors, edge cases, validations.
RAG & Knowledge Systems
Applied RAG for support and documentation (relevant info retrieval). Using embeddings (Ada v2) and designing structured data-based "memory". Prompt and output design for useful retrieval (less noise, more precision). Building knowledge ops systems for accurate, contextual AI responses.
Technical Documentation (as Product)
Real implementation of Diátaxis (Tutorial/How-to/Reference/Explanation) applied to platform docs. Editorial standards: explanation first, then bullets (fluid reading), practical client-oriented tone, no fluff. Building docs designed to: reduce tickets, accelerate onboarding, guide troubleshooting with actionable steps.
Frontend for Documentation
Building "CMS-proof" embedded UI: namespaced CSS to avoid global style clashes. Components: accordions, panels, cards, comparison tables, callouts, modals/lightbox. Debugging typical issues: "locked" font sizes, clipped shadows on load, dropdowns that won't close. Basic accessibility: aria-controls, aria-expanded, roles.
Data Enrichment & Web Research
Using search models (Perplexity Sonar) to enrich leads within flows. Evaluating sites/URLs to determine: real business vs just "platform name", useful signals for message personalization. Structuring enriched output for downstream systems (Sheets/CRM/Zapier).
Product Engineering
Design and construction of proprietary Workflow Orchestrator (Zapier/n8n-style): visual editor, execution engine, AI-first architecture. Converting operational needs into clear features (inputs/outputs, states, logs). Building platforms, not just automations.
Data & Analytics for Operations
Designing reports and metrics for support and risk: churn risk analysis from tickets, friction patterns in onboarding and support. Data quality rules: dedupe by Ticket ID per day in alert reports, consistent summaries as memory. "Control tower" mindset: visibility, traceability, action.
AI-Powered Customer Success
Designing negative sentiment flows with actionable alerts and recommendations. "Angry path" with: firm clear draft for client, internal output for human (what to do, next steps, what to request). Focus on reducing escalation, improving response times, and protecting client relationships.
B2B Discovery & Consulting
Conducting and preparing discovery calls (pain, context, stack, constraints). Translating business problems → executable technical design. Signal-based outreach personalization from real prospect data. Preparing proposals and case study narratives (deck, story, next steps).
Cross-functional Leadership
Close work with leadership (roadmap, priorities, AI innovation cadence). Collaboration with colleagues on prompt and system design. Pushing standards: output quality, templates, operational consistency. Driving AI governance and best practices across teams.
Enablement Assets & Visual Content
Designing documentation with visual support: GIFs and visual resources to guide steps. Structuring UI to be "self-explanatory". Focus on clarity: highlighting clicks, zooms, sequences, and end-user orientation. Creating training materials that actually get used.
Cloud & Operations
AWS deployment (S3 + Elastic Beanstalk) for hosting and operations. Log-based operations: reading and interpreting technical logs, identifying intermittent failures and patterns. Communicating findings to engineering/ops with clear context and recommended actions.
