PRACTICAL AI. EVERYDAY BUSINESS WORK.

Start with the work.
Improve the process.
Then add AI where it helps.

I look at how work gets done and where it is unclear, repetitive, slow, or prone to errors.

I improve the steps, then use AI to make the work faster, clearer, or more consistent—with repeatable instructions and human review.

See a real-world example ↓Explore my approach ↗

For consulting teams: better research briefs, meeting follow-ups, and repeatable deliverables.

ONE EVERYDAY TASKILLUSTRATIVE

A new customer inquiry arrives.

Research the companyFigure out the next stepWrite a responseUpdate the customer record

WITH A REUSABLE AI PROCESS

“Review this inquiry using our process. Research the customer and help me prepare my next step.”

✓

A useful first draft of the work.
I review it, decide, and take action.

Examples are anonymized to protect client and partner confidentiality. Process architecture, methodology, and non-confidential outcomes are represented accurately. All demo records are synthetic.

01 / HOW I WORKSTART WITH THE WORK

First, understand the job
that needs to get done.

I start with the people doing the work. We walk through their process together and, when useful, use a short narrated workflow walkthrough to spot friction and judgment calls.

What needs to improve?

Find the repetitive work, missing information, delays, or mistakes.

What comes in?

Identify the email, request, form, document, or conversation that starts the work.

What happens next?

Map the steps, who owns them, and where someone needs to make a decision.

What should come out?

Define the finished result and how we will know it is useful.

THEN, MAKE IT REPEATABLE

Give AI a clear set of instructions.

For each step, I define the input, what AI should do, and the output I need. I save those instructions with examples, test the results, and reuse the process as new work arrives.

Write the processSteps, approved information, examples, and review points.

Test it on real workCheck the answer, adjust the instructions, and try again.

Help the team use itShow people when to use it, what to check, and how to handle exceptions.

Measure and improveCompare time spent, corrections, missed steps, and actual use.

Sometimes, watch the work +

Sometimes the fastest way to understand a process is to see it happen. When useful, I ask the person doing the task for a short narrated screen walkthrough.

It can reveal system switching, information searches, repeated data entry, approval waits, judgment calls, workarounds, and where someone gets stuck.

Record the task → Identify friction → Improve the process → Test the change → Get feedback → Refine

Current state: The person briefly narrates how they do the work today.
Observe: Identify repeated actions, searches, handoffs, workarounds, decisions, and delays.
Improve: Simplify the steps. Decide what to standardize, support with AI, automate, or leave alone.
Test: Let the person try the revised workflow.
Feedback: If something is awkward or fails, another short walkthrough can show exactly where.
Refine: Update the workflow, instructions, or AI assistance using what we observed.

This can make discovery and feedback more efficient than relying entirely on abstract descriptions or repeated meetings.

For sensitive work, use approved recording and AI environments, and use synthetic or test data unless organizational policy explicitly permits the underlying information to be processed there.

What goes into the instruction document? +

Example: research a new customer

The document can be a Markdown file—a plain-text document with headings and steps. The file format is less important than clear, tested instructions.

Read the inquiry and the information already available.
Research the company using relevant sources; show where findings came from.
Summarize the customer, their request, and the potential fit.
Flag missing information and separate evidence from assumptions.
Recommend the next step using the company’s process.
Prepare the agreed output: a brief, response draft, record notes, and follow-up task.
Leave decisions, external messages, and record updates for review.

02 / SEE IT IN PRACTICEMY METHOD · SYNTHETIC EXAMPLE

Real work.
Practical AI help.

I use these methods in functional beverage manufacturing. They help me qualify leads, respond to customers, prepare for calls, and assemble quotes with less time spent researching, retyping, and starting from scratch.

Qualify a leadInquiry → customer brief

Reply thoughtfullyMissing details → tactful questions

Prepare for a callAccount history → quick briefing

Assemble a quoteRequirements → quote draft

CASE 01 / QUALIFYING & RESPONDING

A new inquiry arrives. Is it worth pursuing?

AI-assisted qualification

Input: The inquiry, customer details, and available company information.
AI helps produce: A company brief, recommended next step, and a tailored email, text, or call outline.

See the example: inquiry → research → response +

Select each step to see the input, AI output, and my review.

The advantage

I can assess and respond sooner, with less effort drafting from scratch. I review the facts, wording, and tone so the message fits the customer and preserves rapport.

How I set this up—and where the customer record fits +

A customer conversation inside an AI project

I keep the business instructions and approved examples in the project, then start a dedicated conversation for the customer. I provide the inquiry and ask AI to follow the relevant procedure.

I define the research steps, output format, follow-up timing, and message style so the same process can be reused.

I review and update the CRM

A CRM is the system holding the official customer and sales records. I use the reviewed output to create or enrich the organization and contact, add notes, and set a task. I create a deal when it meets the business’s criteria.

This hands-on method includes manual review and updates. Connecting systems to do those updates is a separate implementation step.

THE SAME METHOD, THROUGHOUT THE CUSTOMER JOURNEY

Keep the work moving after the first response.

CASE 02 / GETTING READY FOR A CALL

“Remind me where we are with this customer.”

Persistent customer context

I’m about to call a customer and need to get up to speed. I ask AI to review the available conversation history and prepare a short briefing.

The advantage: Less time piecing together old emails. I can focus on the conversation and the next decision.

SYNTHETIC EXAMPLE

INPUT: HISTORY + MY REQUEST

“We discussed an initial order. Final files are pending. I sent the requirements checklist. Brief me for today’s call.”

OUTPUT: A QUICK CALL BRIEF

Where we are: Interested; final files still needed.
What I promised: Requirements checklist—sent.
What to ask: When will the files be ready?
Next step: Confirm readiness before scheduling.

How I set up briefings and keep the customer history useful +

What I use today

I keep a dedicated customer conversation, saved business instructions, and a current CRM record. The briefing format asks: Where are we? What changed? What did we promise? What is missing? What should I ask or do next?

After a call, I add notes or a transcript and request an updated summary and follow-up task. When a thread gets too long, I check a handoff summary before starting the next one.

Chat memory can miss details. I verify important facts; the CRM or other authoritative business system remains the system of record.

How I would extend it for a larger team

Bring together approved CRM records, emails, call notes, meeting transcripts, quotes, files, decisions, objections, promises, and operational updates—with appropriate access controls.

First check what the organization already has. In a Microsoft environment, Copilot and capabilities such as Work IQ are options to evaluate for retrieving relevant context. Check licensing and configuration before building a custom system.

Microsoft: Work IQ overview ↗

CASE 03 / PREPARING A QUOTE

The details are scattered.
I need to prepare a quote.

Controlled generation

Input: Customer emails, beverage requirements, quantity, label, packaging, and approved pricing.
AI helps produce: An organized quote draft, missing-input questions, and a customer message.

The advantage: Less effort gathering details and assembling the quote. I check the numbers, assumptions, and tone before sending.

See the quote example and setup +

SYNTHETIC EXAMPLE

Requirements2,000 canned beveragesLabel / packagingApplied label / case packExample approved price$4.50 per unitCalculated subtotal$9,000.00Still to confirmFreight, taxes, and timing

How I set it up: Define the inputs, approved pricing rules, quote format, and review checks. AI organizes and explains the quote; approved pricing and calculation systems control the numbers. AI should not invent prices. These figures are synthetic.

A tactful request for missing details: “To complete the shipping estimate, could you share the delivery address and any unloading requirements?”

Possible connected step: With an approved freight-system integration, request a carrier estimate using the shipment details and retain its source, date, and conditions. This is an extension to evaluate; no live freight rate is retrieved in this example.

CASE 04 / STARTING THE WORK

Turn a customer’s “yes”
into a ready-to-start job.

Workflow orchestration

A customer says yes, but work cannot start yet.
Input: The accepted order and required files, checks, and approvals.
AI helps produce: A missing-items summary and next-action draft; the workflow tracks completion.

The advantage: The team knows what is holding up the job and who needs to act.

Try a simple readiness check +

SYNTHETIC EXAMPLE · LAUNCH #1047

Requirements and files✓ Complete

Internal review✓ Complete

Customer approvalWaiting

SchedulingBlocked

Owner: Account Manager

How I set it up: Define the readiness checklist, task owners, and what must be complete before scheduling. AI summarizes gaps; recorded approvals control progress.

Next action: obtain customer approval.

Illustration only. No customer record is changed.

CASE 05 / IMPROVING THE STEPS

Fix the sequence before speeding it up.

Process redesign

A required item was being discovered after payment. I moved the readiness check earlier and created customer guidance. AI can help explain missing requirements; the process determines when work advances.

Before

Acceptance → Invoice → Payment → Missing requirement discovered

Improved

Acceptance → Readiness check → Requirement completed → Invoice → Execution

03 / START SIMPLE. BUILD WHEN NEEDED.FIT THE COMPANY’S WAY OF WORKING

Useful AI does not have to
start with a software overhaul.

I would begin with one useful task in the tools the team already has. More connections and automation can follow when the benefit is clear.

HELP ONE PERSON

Improve a daily task.

Use clear instructions and examples in an approved AI tool. A person provides the input, reviews the result, and acts.

A practical starting point for research, writing, and meeting follow-up.

MAKE IT USEFUL TO A TEAM

Share the process.

Agree on the instructions, trusted information, and review points. Test with users and teach the team how to use it.

Measure time spent, corrections, missed steps, and adoption.

CONNECT WHAT IS WORTH CONNECTING

Reduce manual handoffs.

Evaluate connections between AI, documents, customer records, and tasks with the company’s technical team.

Add automation where access, approvals, and reliability are established.

HOW I WOULD APPLY THIS WITH A CONSULTING TEAM

Start with one deliverable people produce every week.

Before a meeting

Turn approved background material into a brief: what matters, what is missing, and what to ask.

After a meeting

Turn notes into a reviewed summary, decisions, assigned actions, and a draft follow-up.

Preparing a deliverable

Use an approved template and source material to prepare a draft the consultant checks and improves.

My pilot would include: tested instructions, an example output, a quick-start guide, and a feedback loop. Compare time to a usable draft, corrections required, and whether people keep using it.

FOR A MICROSOFT-BASED TEAM

Start with Copilot and the work people already do.

I would identify a specific task, check what Microsoft 365 Copilot can access in that environment, and test a small, useful solution with the team. If it needs repeatable actions or connections, I would evaluate Copilot Studio and Power Automate with the appropriate technical owners.

Tools and experience, in context +

What these examples are based on

Hands-on AI-assisted customer work in functional beverage manufacturing: ChatGPT, documented instructions, customer conversations, and Pipedrive records. My broader background includes Process Street workflow discovery and implementation, plus rapid prototyping with tools such as Lovable and Codex.

What I would evaluate for the company

Microsoft 365 Copilot, Copilot Studio, and Power Automate in a Microsoft environment; Claude or other approved AI tools where appropriate; and the company’s existing customer systems, such as Salesforce.

The choice follows the task, information access, team needs, and the support required to maintain it.

For a technical reader: how the process, systems, and AI fit together +

Business knowledge

What should guide the work?
Procedures, policies, approved information, decision rules, examples, and templates.

System of record

What is true right now?
Authoritative customer and opportunity information: stage, value, and status. A CRM such as Pipedrive or Salesforce is one example.

Workflow / process state

What needs to happen?
Current step, tasks, owners, due dates, approvals, dependencies, blockers, and handoffs. Process Street is one possible tool.

AI assistance

What can we understand or prepare?
Research, retrieval, summaries, drafts, classification, comparison with history, missing information, and recommended next actions.

Start simple. Add connections when needed.

01
What I use today
Approved instructions → AI workspace → my review. I maintain the customer record.
02
Connected workflow · proposed next step
Connect business knowledge, CRM records, email and documents, task tracking, and AI assistance.
03
Enterprise evolution · proposed direction
Governed knowledge + enterprise customer platform + approved sources + retrieval / memory → AI / agent → tracked work, human approvals, and an audit trail.

Before adding a database, retrieval layer, or custom system, check the capabilities already available in the organization.

AI can assist the work without becoming the source of truth for the business. It should operate within a process the organization can see, govern, and improve.

Security and data governance are part of solution design: classify the data, use approved environments, preserve existing permissions, and avoid moving sensitive information into unapproved tools.

04 / ABOUT BRADBUSINESS EXPERIENCE. PRACTICAL APPLICATION.

APPLIED AI & WORKFLOW CASE STUDIES

Brad Shook.

Understand the work.
Improve the process.
Help people use AI.

Connect on LinkedIn ↗

My career has crossed technology sales, business ownership, healthcare operations, B2B business development, workflow consulting, and applied AI.

I have been the person doing the research, writing the follow-up, updating the record, and figuring out what needs to happen next. That experience shapes how I approach AI: make the work easier to do and easier to follow.

I’m interested in AI enablement, solutions consulting, implementation, and process improvement—helping teams turn useful ideas into everyday practice.

What I would bring: process discovery, practical prototypes, reusable instructions, and hands-on team support.