Services

I design and ship end-to-end AI workflows for sales and operations. Think of me as a sales operator who can architect, prototype, and productionize small tools—fast—using LLMs, automations, and lightweight front ends.

01

AI Workflow Architecture & Integration

What I build

  • Event-driven pipelines: intake → validation → enrichment → decision → action (email, CRM update, ticket, doc).
  • Retrieval-augmented generation (RAG): your SOPs/specs turned into a searchable knowledge layer for reps and customers.
  • Quoting & configuration engines: rules + embeddings + calculators that output BOMs, PDFs, and CRM line items.
  • Agentic orchestration: task-specific mini-agents with guardrails, timeouts, and human-in-the-loop checkpoints.

Typical stack

LLMs: ChatGPT (GPT-5 family), Gemini, Claude (as needed)
Orchestration: Make.com, n8n, Zapier
Data: Google Sheets/Airtable/Supabase
Front ends: Bolt.new (Stackblitz), Webflow/Next.js (lightweight)
Comms: Gmail/Outlook, Twilio/WhatsApp, Slack/Teams
Files: Google Drive/Docs, PDFs with templating

Deliverables

  • Sequence diagrams & data contracts
  • Working scenarios/flows with error handling & retries
  • SOP + rollback plan + monitoring dashboard
02

LLM Systems Design ("Prompt-Ops" done right)

What I implement

  • Task decomposition and tool use (functions, webhooks, calculators)
  • Context packing: chunking, embeddings, ranking, and prompt templates
  • Output contracts: JSON schemas with strict validation
  • Guardrails: policy prompts, regex filters, PII redaction, fallbacks

Evaluation & Quality

Golden sets + adversarial test cases
Latency/accuracy trade-off tuning and model selection
Cost controls and caching strategies
03

Productized Tools & Calculators

Examples I've shipped

  • CoreFlex Guardrail Calculator: layout mapping → standardized BOM → priced quote PDF → CRM
  • Sponsorship Visualizer: upload a logo, render on product surfaces for outreach and proposals
  • Service Ticket Router: classify → assign → SLA timers → customer updates → field ops sync

What you get

A clean web UI (Bolt.new) your team can actually use
Reusable presets and a cloning plan for new clients/locations
Admin controls for pricing, copy, and rules—no dev required
04

Sales Ops Automations

Lead capture → dedupe → enrichment → routing → first-touch
Pipeline hygiene, aging alerts, and "stuck" detectors
Quote-to-cash handoffs with docs, statuses, and notifications
Reporting: weekly rollups to Slack/Email with the metrics that matter

Platforms

Go High Level, HubSpot/Sheets, Google Workspace, Notion, Slack

05

Training & Enablement (Hands-On, Not Theory)

Workshops (customized to your team)

  • LLM foundations for operators: tokens, context, embeddings, RAG, guardrails
  • Prompt systems: roles, patterns, JSON outputs, evaluation loops
  • Build-a-flow: in 90 minutes we ship a real workflow your team keeps
  • AI for reps: inbox triage, research assistants, first-draft quotes (with approvals)

Enablement assets

SOPs, prompt libraries, test sets, and "break-glass" recovery docs

06

Reliability, Security, and Ops

Error handling, retries, and dead-letter queues (where supported)
PII hygiene, least-privilege API keys, and access review checklists
Audit trails: who did what, when, and with which prompt/version
Light monitoring: run logs, failure alerts, latency/cost dashboards

How I Work (Short and Real)

Architecture Sprint (1 week)

Map the flow, define data contracts, pick models/tools, spec failure modes.

Build Sprint (1–2 weeks)

Ship the smallest valuable slice to production with a UI and SOP.

Stabilize & Scale

Evaluate, tune, add automations or agents, then hand off with docs and training.

You'll see: working software, clear before/after metrics, and zero black boxes.

Where I'm a Fit

  • You have messy, manual processes in sales/ops and want a working fix fast.
  • You're okay starting small and measuring real outcomes.
  • You want your team trained to own the system after handoff.

Example Outcomes

  • Quotes: days → hours, standardized BOMs, higher first-call close rate
  • Tickets: 10+ hrs/week returned to the sales team, clean SLAs, happier customers
  • Outreach: personalized visuals → higher reply and approval rates

Ready to architect and ship an AI workflow for your team?

Start with an architecture sprint