Services

Automation that runs your operation — built to last.

Five ways to take repetitive, high-volume work — or a whole fragmented system — off your team's plate, each one built on your own systems and accounts, accurate, secure, and still running months later.

5.0(17 of 18 rated)

Top Rated on Upwork — 22 clients, 31 contracts, 18 reviews

n8n, GPT-4o, Claude, AWS Bedrock, Microsoft Graph, Clio, LangGraph

What I build

Systems that pay for themselves — in hours saved, and tools retired.

Five practices, one discipline: take the repetitive, high-volume work — or the whole fragmented system it runs on — off a team's plate without ever putting accuracy or client confidentiality at risk.

01

RAG Assistants & AI Agents

Assistants trained on your private data and agents that take action — update the CRM, trigger workflows, reason across steps. Accurate retrieval, grounded answers, nothing invented.

  • A document ingestion pipeline you can re-run
  • Per-tenant or per-matter retrieval isolation
  • Citations on every answer, not just a confident guess
  • Defined action boundaries for anything agentic
Grounded retrievalPer-tenant isolationCited answersAgentic actions
answer

retrieve → reason → act

02

Document & Intake Automation

Automated intake, document generation, and structured-record pipelines — secure, structured, and built for the volume and accuracy document-heavy work demands.

IntakeRecordsBilling
document
03

AI Email & Workflow Automation

Context-aware drafting and routing that turns hours of repetitive email into minutes of review. Connected to your inbox and CRM, running 24/7.

Live workflowrunning 24/7
Inbox
AI agent
Draft
04

Custom Software & AI Operations Platforms

Full systems built from scratch when off-the-shelf tools can't do what your operation needs — client portals, internal dashboards, multi-tenant platforms, and the integrations that wire your existing tools into one place.

Connected systemsone platform
CRM
Work
Billing
Platform
Dashboard
05

Support Inbox & Voice Agents

Classify, draft and route customer email, chat, SMS and phone calls across every channel — with the urgent ones escalated to a person instead of buried behind the routine ones.

Live workflowrunning 24/7
Inbox
AI agent
Draft

Everything on the table

Five categories, fifteen ways to get the work off your plate.

Not every engagement needs a full custom platform. Some start with an hour on a call, some are one automated workflow, some are ongoing support on something already running. Pick the entry point that matches where you actually are.

AI strategy & discovery

Workflow implementation

Internal AI systems

Integrations

Ongoing support

Engagement

Three ways to start.

Every one of them begins with the same call, and none of them start with an invoice you didn't expect.

Automation audit

Free

I map the repetitive processes across your team, and tell you which are worth automating, which aren't, and what each would take — at no cost, before anything is decided. You keep the map whether or not you build with me.

  • Process + tooling walkthrough
  • Ranked shortlist with effort and impact
  • Free — no obligation to build
Most common

First build

Fixed price per build, starting at $250

One workflow, scoped to a clear deliverable, built on production tooling and tested against your real data. You watch it run end-to-end before we talk about anything else.

  • Starting at $250, scoped before anything is built
  • Wired into your inbox and CRM
  • Tested on your real, messy data

Ongoing partner

Hourly — $25/hr

For once systems are live: monitoring, tuning, and the next workflow when you're ready. This is the part most automation gets wrong — the months after launch.

  • $25/hr, billed for the work actually done
  • Fixes as edge cases surface
  • Expand when it's earning its keep
Not sure which one fits?

Check what's true for the process you're thinking of. This won't quote a price — just point you at the right starting point.

First build

Sounds contained enough to ship as one fixed-price build. The free audit works out exactly what starting at $250 covers here.

What moves the price
  • How many systems the build has to connect to
  • The volume of email, documents or records it runs on
  • What compliance or retention rules apply to the data
  • Whether anything is allowed to send without a human review
The guarantee

Tested before it ships, not after.

Every build runs against your real, messy data before it goes live. If it isn't accurate enough to trust, it doesn't ship — you're not paying for a system that quietly fails on the work that matters.

Not a fit if
  • You need something sent to clients fully unattended, with no human review
  • The process happens rarely enough that automating it costs more than doing it by hand
  • You can't give a build access to the real systems or data it needs to be tested against
  • You need a large support desk or bench — this is one engineer, not a team

How I work

From repetitive process to running system.

A clear, low-risk path from the first call to a system that runs your operation — and keeps running.

  1. 1.

    Discovery

    We map the repetitive process you want gone — the triggers, the tools, the edge cases — and agree on what 'done' looks like.

  2. 2.

    Design

    I design the workflow and the AI's guardrails: what it reads, how it decides, and where a human stays in the loop.

  3. 3.

    Build & test

    I build it on production-grade tooling, wire it into your inbox and CRM, and test it against your real, messy data — not a happy-path demo.

  4. 4.

    Run & refine

    It goes live running your operation. I monitor, tune, and stay on for the months that follow — so it's still working later, not just at launch.

Tools & stack

What the systems are actually built on.

Grouped by the job each one does. Everything here appears in work that shipped — it is a record of what I build with, not a list of logos.

Models & AI

Chosen per task, and swappable — no system here is welded to one provider.

  • Claude
  • OpenAI API
  • AWS Bedrock
  • Google Gemini
  • Hugging Face
  • PyTorch

Agents & orchestration

Where the logic lives: what runs when, what it is allowed to do, and where it stops for a person.

  • n8n
  • Make.com
  • LangChain
  • LangGraph
  • Webhooks

Retrieval & knowledge

How an assistant answers from your documents instead of from its own imagination.

  • pgvector
  • Pinecone
  • FAISS
  • Chroma

Backend & APIs

The services the workflows call, and the ones they expose.

  • Python
  • FastAPI
  • Node.js
  • PostgreSQL
  • Supabase

Apps & dashboards

Where a team actually sees and operates the system.

  • Next.js
  • React
  • TypeScript
  • Tailwind CSS
  • Stripe

Business systems

The tools a client already runs. The work is almost always meeting these where they are.

  • Microsoft Graph
  • Clio
  • Airtable
  • GoHighLevel
  • Gmail
  • Google Drive
  • Notion

Voice & messaging

For the workflows that reach a person by phone rather than by inbox.

  • Twilio
  • Vapi

Cloud & delivery

Where it runs, and how it gets there.

  • AWS ECS
  • Docker
  • Vercel
  • GitHub

Questions

The questions buyers actually ask.

Still unsure if your process is a fit? The fastest answer is a ten-minute call.

Book a call
  • Yes. Systems run in your environment with your accounts and keys — your data isn't sold or used to train public models. I build to the access and retention rules your work requires, with human review where accuracy matters.

  • A short call to see if a process is worth automating. If it is, I scope a first build with a clear deliverable — usually one workflow you can watch run end-to-end before we expand.

  • Most engagements are fixed-price per build, starting at $250, or $25/hr for ongoing work. You'll know the number before we start — no surprise invoices.

  • That's the whole point of how I build. It's tested on your real, messy data before launch, with a human in the loop where it counts, and monitored after — so the failure cases are handled, not discovered in production.

  • Any business running into the same wall: too much repetitive, high-stakes work for the team carrying it. The pattern that qualifies a business isn't its industry — it's a team drowning in repetitive email, intake, document, or CRM work. So far that's included law firms, marketing agencies, accounting and finance operations, and professional-services firms; the case studies on this site skew legal because that's where the most contracts have landed to date, not because it's the only fit.

  • All three are real options, and sometimes they're the right one. No-code tools work until the logic gets complex enough that you're debugging someone else's automation builder instead of your own process. An in-house hire makes sense once there's enough recurring work to justify a full-time seat. A bigger agency buys you a bigger team, usually at the cost of an account manager standing between you and whoever is actually doing the work. What I offer is the case in between: a system built for exactly your process, by the person who built it and who stays accountable for it — without the overhead of a full-time hire or the distance of an account team.

Losing hours to work an AI system could handle?

Tell me about the process. I'll tell you straight whether it's worth automating — no pitch, no obligation.

Book a free audit call