AI for small business

One useful workflow.
A clear plan to get there.

Start with the work that repeats: drafting replies, preparing paperwork, or organizing requests. Test the fit before committing to a bigger rollout.

The discovery call is free. Assessments start at $750, pilots start at $2,500, and support plans start at $300 a month. Fixed scope, agreed in writing before work begins. See pricing

Need IT help? Explore IT support and rates →

How I work

Three steps. A decision at each one.

01 · Understand

AI Workflow Assessment

I map the current process with you: who does the work, which tools and data it touches, and where time or errors accumulate.

  • Deliverable: a written workflow outline, candidate improvements, and a recommended first test.
  • Your decision: agree on the problem and whether AI is appropriate. A checklist or ordinary automation may be the better answer.
02 · Test

One-Workflow Pilot

Build a limited version around one task, using approved data and tools. Scope, pricing, software costs, access, and success criteria are agreed before implementation.

  • Deliverable: a working pilot, documented setup, and a review against the agreed criteria.
  • Your decision: continue, adjust, or stop. Test accuracy, effort, and failure cases before allowing broader use.
03 · Use

Training & Ongoing Support

Practice with the people who will use the workflow. Explain what the tool can do, where it can fail, and when to hand work back to a person.

  • Deliverable: hands-on training, a written reference, and an agreed support plan.
  • Your decision: set the check-in cadence, responsibilities, and support scope. Ongoing work and software subscriptions are confirmed separately.
Illustrative workflow example

From incoming question to reviewed reply.

Fictional sample data, not a customer result or a claim of measured time savings.

  1. Input

    A routine inquiry

    Sample inquiry: “Do you set up Wi-Fi for a small office? What do you need to know before a call?”

    Sample service information: Network and Wi-Fi setup is offered. Calls are by appointment.

  2. Draft

    AI prepares a reply

    Sample draft: “Yes, network and Wi-Fi setup is offered. Could you share how many devices need to connect and what issue you are trying to solve? A call can be arranged by appointment.”

    Flag for review: The inquiry does not include a device count or describe the current problem. The draft stays unsent.

  3. Human review

    A person checks & sends

    Confirm the service fits, check the details, and approve the wording. Keep the request for missing information; arrange an appointment only after checking availability. Then send through the normal process.

What a pilot would measure: drafting effort, corrections needed, and whether the final reply meets your standards.

Guardrails before rollout

Your team keeps the decisions.

AI can produce incorrect or misleading output, even when it sounds confident. Human review is part of the workflow, not an optional extra.

Privacy & approved access

Decide which data may be used, by whom, and in which tools. Review provider terms, data handling, retention settings, and permissions before connecting business systems. Don't enter passwords, financial details, or other sensitive information into unapproved tools.

Local and cloud tradeoffs

Local AI can reduce some external data sharing, but privacy depends on the full setup, including integrations, logs, backups, and remote access. Cloud tools have their own policies and costs. I explain the tradeoffs rather than promising absolute privacy.

Examples of what is possible

Four workflows, with you still approving the result.

These are examples of small-business workflows I can set up with you. They are illustrations, not past client results.

Draft a quote from an intake email

A customer emails what they need. A draft quote is prepared from that email and your pricing. You approve the price, scope, and wording before it is sent.

Invoice reminders you approve

When an invoice is overdue, a reminder is drafted in your voice. You approve it before it goes out.

A daily list of customer inquiries

New inquiries are summarized into one list: who wrote, what they need, and what is waiting. You decide what to handle first.

Meeting notes turned into tasks

Notes from a call become a short task list with owners and due dates. You review the list before the work is assigned.

Each example keeps a human approval step before a message, a price, or an assignment goes out.

Tool ownership, export options, licensing, and ongoing costs depend on the setup. I'll discuss them before you commit.

AI questions, answered

Questions about starting with AI

How can a small business start using AI?

Start with one repetitive task, not a company-wide overhaul. I begin with a free discovery call, then an AI Workflow Assessment that maps the process and recommends a first test. A scoped One-Workflow Pilot follows only if it makes sense, then hands-on training and an agreed support plan.

Is the AI discovery call free?

Yes. The discovery call is a free conversation about your business, not a commitment to a project. If you want to go further, the AI Workflow Assessment is a fixed price starting at $750, a One-Workflow Pilot is fixed scope starting at $2,500, and training and ongoing support is a monthly plan starting at $300 per month. Final scope and cost are agreed in writing before paid work begins.

Which tasks are good first AI projects?

Good starting points are repetitive tasks you can still review: drafting a quote from an intake email, invoice reminders you approve before they are sent, a daily list of customer inquiries, and meeting notes turned into tasks. A person checks the result before it is sent or assigned.

Does a person stay in control of the AI?

Yes. AI can produce incorrect or misleading output, so human review is built into every workflow. In a typical pilot the tool prepares a draft and flags missing details, and a staff member checks facts, prices, tone, and permissions before anything is sent.

How is business data and privacy handled?

Before any tools are connected, you decide which data may be used, by whom, and in which tools. Provider terms, data handling, retention settings, and permissions are reviewed first. Local AI can reduce some external data sharing, but privacy depends on the full setup, so the tradeoffs are explained rather than absolute privacy promised.

What if AI is not the right answer?

Then I say so. A checklist, ordinary software, or a simple process change is sometimes the better fit, and the assessment is designed to surface that before you spend money on a build.