AI automation for small business

Find the leaks. 
Automate the fix. 
Hand over the keys. 

I come into your business, find the biggest time and money leaks, build the automations that fix them, then hand them over so your team owns them.

An AI workflow automation consultant and fractional AI operator for teams of 5 to 50. Practical automations and AI agents built on the tools you already use, with approvals, measurement and training baked in.

Live queue● Running
InboxAI agentApproval gateCRM · Books
“Do you have any openings Thursday afternoon?”
Availabilityconf 0.96Reading…
“Can you resend my last invoice?”
Billing · lookupconf 0.91Draft ready
“The tech never showed up and I took the day off.”
Complaintconf 0.58Escalated

Simulated run of the Email Triage blueprint · nothing is sent without approval

The internal AI hire, fractional

Operators who kill manual work are the new unfair advantage.

CEOs are hiring internal "AI transformation" people whose only job is to ship automations across sales, ops, finance and support. Box CEO Aaron Levie calls the emerging role the agent operator. Most 5–50 person businesses can't justify that hire full-time yet. BotsAndOps is the fractional version.

How the AI ops layer works

AI that works inside your business, with you in control.

01

Connect your systems

Inbox, CRM, scheduling, accounting and the spreadsheets your team already keeps, wired into one flow so every task starts with the right facts.

02

Measure before you trust

Each agent is tested against labeled examples of your real work. If a change makes it worse, it doesn't ship.

03

Let AI act, with approval

AI drafts and prepares, and your team approves in one click. Low-risk tasks earn more autonomy only when the numbers say so.

Your tool stack, connected

AI that works inside the tools you already pay for.

Pick a tool to see what we automate with it and which blueprints use it. No rip-and-replace: the AI ops layer sits between your systems and your team.

Make

Automation
MakeAI Ops layerYour team

What we automate

  • Watches inboxes, forms and apps for new work
  • Branches by confidence and risk
  • Retries and de-duplicates so nothing runs twice

Used in blueprints

Inbound Email Triage Agent with Approval Gate→
The Earned Autonomy method

AI starts supervised. It earns autonomy with data.

AI projects stall when nobody owns them, nobody measured the "before," and nobody planned for mistakes. Scroll to see how every workflow moves from Red to Green.

Autonomy level

Red: Human decides

Autonomy is earned with data. It's never assumed.

Stage 1 · Red

Human decides

AI researches, summarizes and drafts. A person makes every call. Every new workflow begins here, as does anything involving money, legal risk or an unhappy customer.

Move on when: a labeled test set exists and accuracy is measured.

Stage 2 · Yellow

Human approves

AI takes the action, but a person checks it before it reaches a customer or a system of record. Uncertain items go to their own review queue.

Watch: accuracy, approvals without edits, escalation rate.

Stage 3 · Green

AI runs, people spot-check

One task type at a time is promoted, only after it holds its target accuracy. A sample is still reviewed every month.

Demotion is normal: if accuracy slips, it goes back to Yellow.

Blueprints

See the actual systems, not just promises.

Reference builds with the architecture, stack, controls and what each one measures. Clearly labeled as samples. No invented client results.

01
Reference buildService & booking businesses

Inbound Email Triage Agent with Approval Gate

Every incoming email is read, labeled by intent, enriched with facts from your systems, and turned into a reply draft that a person signs off on. Auto-send stays off until the numbers justify it.

Make or n8n · OpenAI or Claude · Gmail / Outlook · QuickBooks Online · Google Sheets · promptfoo
See the architecture →
02
Reference buildAny industry

Ops Knowledge Brain: Cited Answers from Your SOPs

Your team asks questions in plain English and gets answers pulled only from your own policies and procedures, with a link to the source document under every answer.

n8n · Supabase pgvector · Claude or OpenAI · Google Drive / SharePoint
See the architecture →
03
Reference buildProperty, finance & back office

Document Intake Extractor (Invoices, Contracts, Forms)

Uploaded invoices, contracts and forms are read into structured fields with a confidence score per field, written to your tracker, and anything uncertain is flagged for a person.

n8n · Claude or OpenAI (vision) · OCR · Google Sheets / Airtable
See the architecture →
04
Reference buildProfessional & home services

60-Second Lead Intake Pipeline

A new web inquiry is captured, researched, scored, logged in your CRM, answered with a personalized first reply, and posted to your team in about a minute.

n8n or Zapier · OpenAI or Claude · HubSpot / GoHighLevel · Gmail · Slack · Tally / Typeform
See the architecture →
05
Reference buildAny industry

Ops KPI Dashboard with Weekly Written Brief

A live view of your core operating numbers plus a short weekly brief in plain English: what changed, what looks risky, and three suggested actions.

n8n · Google Sheets / Supabase · Claude or OpenAI · Looker Studio or a custom dashboard
See the architecture →
06
Reference buildAny industry

Multi-Agent Operations Advisory Team

Several specialized agents (research, strategy, marketing, operations) work through a business question in sequence and hand back a short executive brief with a 30-day plan.

n8n · Claude or OpenAI · Web search · Google Docs
See the architecture →
07
Reference buildProfessional services & B2B

Sales Pipeline and Forecast Board

All open deals on one board by stage, with live totals and a forecast weighted by your historical win rates, so you can see what's likely to close without maintaining a spreadsheet.

HubSpot or Pipedrive · n8n · Claude or OpenAI · Google Sheets
See the architecture →
08
Reference buildProfessional services & agencies

Staged Client Onboarding Checklist

A step-by-step onboarding flow where finishing one stage unlocks the next, each step has one owner, and both your team and the client can see progress.

Notion or Airtable · n8n · Gmail · DocuSign or PandaDoc
See the architecture →
09
Reference buildAny industry

SOP-Driven Task Drafts

A task board that shows the written procedure beside each task and drafts a first pass using both, so your team starts from a draft that follows your own process.

Notion or Airtable · Claude or OpenAI · n8n · Google Drive
See the architecture →

Drag or scroll to explore → · All blueprints

How an install runs

Five phases. Rules before the build.

Owners, risk levels and a failure plan are set before anything is automated, so AI lands on clean processes, not broken ones.

The full method
1

AlignWeeks 1–2

Owner interview, goals, a readiness check, an inventory of tools and unofficial AI use, and a shortlist of candidate workflows.

2

MapWeeks 2–3

Document the top three workflows step by step and measure today's baseline: time, cost, errors.

3

GovernWeeks 3–4

Name an owner for each workflow, set risk levels, decide what needs approval, and write the 'what if it breaks' plan before building.

4

DeployWeeks 4–8

Build against those rules and run side by side with the current process until quality targets are met.

5

OptimizeOngoing

Monthly review of results, exceptions and spend. Workflows move from Yellow to Green only when the data supports it.

Services

Audit. Build. Train. Operate.

Every engagement starts with finding the leaks. Most teams know within two weeks exactly where AI pays off, and where it doesn't.

Also available
Websites

Fast, modern websites and landing pages built to convert, with forms wired into your CRM.

Social media marketing guidance

Practical content systems and AI-assisted workflows to keep your channels consistent.

Custom app development

Internal tools, portals and lightweight apps when off-the-shelf software doesn't fit.

General tech help

A technical partner for the questions in between: tools, integrations, accounts and vendors.

FAQ

Straight answers

We already pay for ChatGPT or Copilot. Why would we need this?

Chat tools help individuals one prompt at a time. The bigger gains come when AI is wired into your actual workflows (inbox, CRM, scheduling, accounting) with clear owners, approvals and measurement. That's the part I build.

Will AI send things to our customers without anyone checking?

Not unless you decide it should. Everything starts supervised: AI drafts, your team approves. Specific low-risk tasks are only promoted to run on their own after they consistently hit an agreed accuracy level.

Do we need developers or new software?

Usually not. Most builds run on tools you already have or can subscribe to (Zapier, Make, n8n, your CRM and an AI provider), and they come with documentation your team can maintain.

Who owns what you build?

You do. Accounts are in your name, workflows are documented, and your team is trained before handoff.

Are the blueprints real client projects?

They're reference builds, clearly labeled. They show what I'd install and how it's controlled. They are not client case studies, and no results are implied.

How do we know if it worked?

We record a baseline before building (time per task, response times, error rates) and compare against it after launch, workflow by workflow, using your own numbers.

What size businesses do you work with?

Mostly owner-led teams of roughly 5 to 50 people: home services, professional services, healthcare and aesthetics practices, venues and agencies.

Start here

Find the three workflows worth automating first.

A free 20-minute call. No pitch deck. We look at where your team's hours go and whether AI is actually the right fix.

Ask the AI guide