Olean businesses · Practical AI · Prompts and workflows

Our First Business Release: Practical AI for Olean Businesses

A hands-on companion to CNERD’s first business email: original campaign images, five copyable prompts, a review workflow, and an offline business-listing code example.

By Michael Kintner · Published October 7, 2026 · 9 min read

A busy business does not need another technology project without a clear purpose. It needs a better way to finish the work already on the desk: a customer reply, a proposal, a marketing message, or the notes from yesterday’s meeting.

Our first business-focused email introduced CNERD’s practical AI approach to the Olean community. This article turns that introduction into a small exercise you can try, review, and improve before making a larger commitment.

This post builds on the original business release email. Its campaign artwork is reused below; the prompts and code examples in this article are expanded teaching examples.

Original Olean business release artwork showing local business branding and practical AI collaboration — Original lead artwork from the first business release email.

Start with one piece of everyday work

The release describes support for business communication, routine office work, automation, custom tools, local AI, and training. Start with a task you can describe clearly and a result a person can check.

Everyday task	A useful first AI exercise	Human review
Customer communication	Draft a reply from confirmed notes	Check the answer, tone, and any commitments
Marketing	Prepare a short message about an actual service	Verify claims, prices, dates, and the next step
Office work	Turn meeting notes into a checklist	Confirm owners, deadlines, and missing decisions
Repeated handoffs	Map the steps before automating anything	Identify exceptions and approval points

CNERD campaign artwork showing a robot and practical AI services for Olean businesses — Original services artwork from the business release email.

A useful first goal is a reviewable draft, not a promise that AI will run the business for you. Work on a small example, check the result, and decide whether it makes that task easier.

Try this: review what a customer can find online

The email’s hands-on exercise starts with a business’s online presence. You can follow the same idea by asking a search-enabled assistant to investigate public information and show the evidence behind each finding.

Use an assistant with web search available. Replace the placeholders, then inspect its linked sources yourself. If search is unavailable, provide the pages or screenshots you want reviewed and ask it to work only from those materials.

PROMPT

Prompt 1 — Review local business visibility

Act as a careful reviewer of public business information.

Business: [BUSINESS NAME]
Location: [CITY, STATE]
Website: [WEBSITE URL]
Main services: [SERVICES]

Find the official website and public pages that appear to describe this business. Try several service-and-location searches a customer might use. Record the search terms, review date, and source URLs.

Check whether the business name, address, phone, hours, and service descriptions are clear and consistent. Distinguish a confirmed mismatch from information you could not access. Do not invent reviews, rankings, traffic figures, or a Google Business Profile you cannot verify.

Return: confirmed findings, unanswered questions, and five practical next steps. For each step, identify the evidence and the person who should review it. Make no changes to accounts or listings.

Turn observations into a plan someone can finish

Save the source URL beside each observation. “I could not verify the opening hours” is different from “the hours are wrong.” That distinction helps your team fix the right problem.

PROMPT

Prompt 2 — Build a seven-day action plan

Use only the verified findings and open questions below:
[PASTE YOUR REVIEW NOTES AND SOURCE URLS]

Create a small seven-day plan. Separate corrections we can make now from questions that need confirmation. For each task, include an owner, the source to check, the proposed change, and a simple completion check.

Prioritize unclear contact details, confusing service descriptions, and broken customer paths. Do not promise search-ranking improvements or invent deadlines. Leave unknown owners and dates marked “to confirm.”

WORKFLOW

One verified improvement to a customer’s path from discovery to contact

Before you start: Your confirmed business details · Public source links or supplied screenshots · A person who can approve changes

Tools: A search-enabled assistant or supplied source material · A simple review worksheet · Your existing website and listing accounts

01Choose a customer question
Pick one question, such as which services you offer or how a customer should contact you. Keep the review small enough to finish.
```
Which public page should answer this customer question, and what would a clear answer contain?
```
Expected result: One question and a short list of pages to inspect.
Human review: Check this result before moving forward.
02Collect and check evidence
Use Prompt 1, open the cited pages, and record what is confirmed. Mark inaccessible or ambiguous information as an open question.
```
Show the source for each finding and label anything you cannot verify.
```
Expected result: A worksheet with confirmed observations and source URLs.
Human review: Check this result before moving forward.
03Approve one change
Use Prompt 2 to choose one useful correction. Have the responsible person check the wording and the underlying business information.
```
Propose one small change using only our confirmed facts. Identify what needs approval.
```
Expected result: An approved correction and a named person to make it.
Human review: Check this result before moving forward.
04Recheck the customer path
After an authorized person makes the change, revisit the page and test the contact link or service information. Record what changed.
```
Compare the before-and-after evidence. Does the page now answer the customer’s question?
```
Expected result: A completed, checked improvement rather than an unverified suggestion.
Human review: Check this result before moving forward.

A small code example: compare your own listing details

You do not need code for the visibility exercise. If you like repeatable checks, the example below compares details you have already collected. It reads a local JSON file and reports missing or inconsistent fields; it does not search Google, scrape listings, or change any account.

Replace every sample business detail with information you have checked. The sample phone number and .example website are fictional. Put your confirmed reference first, then add the details from each page you reviewed. A matching report confirms only agreement with that reference, not that a listing is current or complete.

[
  {
    "source": "Checked reference — replace with your confirmed details",
    "name": "YOUR BUSINESS NAME",
    "address": "YOUR STREET ADDRESS, Olean, NY",
    "phone": "716-555-0100",
    "website": "https://your-business.example"
  },
  {
    "source": "A listing you reviewed — paste what it actually shows",
    "name": "YOUR BUSINESS NAME",
    "address": "",
    "phone": "(716) 555-0100",
    "website": "https://www.your-business.example"
  }
]

↓Download the business-listing worksheet (JSON)An editable sample file; replace its fictional details before using it.

Save the following as listing_audit.py. It uses Python’s standard library and performs no network requests.

#!/usr/bin/env python3
"""Compare a business's own supplied listing details; no web requests."""
import json
import re
import sys
from urllib.parse import urlsplit

FIELDS = ("name", "address", "phone", "website")


def normalize(field, value):
    text = " ".join(str(value or "").split())
    if field == "phone":
        digits = re.sub(r"\D", "", text)
        return digits[1:] if len(digits) == 11 and digits.startswith("1") else digits
    if field == "website":
        host = (urlsplit(text).hostname or "").casefold()
        return host.removeprefix("www.")
    return text.casefold()


def audit(records):
    if not isinstance(records, list) or not records:
        raise ValueError("Provide a non-empty JSON list of listing records.")
    if any(not isinstance(row, dict) for row in records):
        raise ValueError("Each listing record must be a JSON object.")
    # The first record is the reference you have checked yourself.
    reference = records[0]
    issues = []
    for row in records:
        source = str(row.get("source") or "Unnamed source")
        for field in FIELDS:
            value = normalize(field, row.get(field))
            expected = normalize(field, reference.get(field))
            if not value:
                issues.append({"source": source, "field": field, "issue": "Missing or unusable value"})
            elif expected and value != expected:
                issues.append({"source": source, "field": field, "issue": "Differs from checked reference"})
    return issues


if __name__ == "__main__":
    if len(sys.argv) != 2:
        raise SystemExit("Usage: python3 listing_audit.py business-listings.json")
    with open(sys.argv[1], encoding="utf-8") as file:
        records = json.load(file)
    print(json.dumps(audit(records), indent=2, ensure_ascii=False))

↓Download the listing-check script (TXT)Save the downloaded text as listing_audit.py to run the example.

python3 listing_audit.py business-listings.json

The supplied sample reports the second record’s missing address. Different phone formatting and a www prefix are normalized for comparison. Review every reported issue against its source before editing a public page; differences can be legitimate.

PROMPT

Prompt 3 — Explain a listing-check report

Here is the report from our local listing check:
[PASTE THE REPORT]

Here are the source pages and confirmed business details:
[PASTE THE EVIDENCE]

Explain each item in plain language. Separate a formatting difference, a confirmed factual mismatch, and a question that needs verification. Suggest the smallest correction for confirmed issues. Do not infer search rankings or treat this report as a complete audit.

Use the same approach for everyday communication

A good prompt supplies facts, describes the audience, and names the review step. The two examples below apply that pattern to customer communication and a useful marketing message. Use non-confidential sample notes while learning.

PROMPT

Prompt 4 — Draft a customer follow-up

Draft a warm, concise follow-up to a customer using these confirmed notes:
[PASTE NOTES]

Purpose of the message: [PURPOSE]
Next step the customer can take: [CONFIRMED NEXT STEP]

Keep the message under 150 words. Do not add prices, availability, delivery promises, discounts, or commitments that are absent from the notes. Put unanswered questions in a separate “Needs confirmation” list, outside the customer-facing draft. A person will review the message before sending.

PROMPT

Prompt 5 — Turn one real service into a helpful post

Write a short educational post for local customers about this service:
[SERVICE AND CONFIRMED DETAILS]

Audience: [WHO THE SERVICE HELPS]
A common question: [QUESTION]
Approved contact or booking link: [LINK]

Explain the service in familiar language, include one clearly labeled hypothetical example, and end with a useful next step. Do not invent customer stories, testimonials, savings, credentials, or guarantees. Return the draft plus a checklist of factual claims for us to verify before publication.

Before you use or share an AI result

Replace every placeholderUse checked business information and the right source links.Open the sourcesConfirm that the cited material supports each finding.Review any promisesCheck prices, availability, timelines, and commitments with the responsible person.Keep sensitive information out of practice examplesChoose public or non-confidential material appropriate for the tool you are using.Approve the final actionHave a person review the result before sending messages or changing public information.

Take the next small step with CNERD

The business release invites you to start with a free 20-minute AI consultation. Bring one task that takes too much effort, or one customer path you want to improve. We can use that concrete example to discuss a practical next step.

Explore CNERD and book a consultation. Prefer a direct conversation? Call Mike at (716) 970-4000.

For more exercises, get the Olean AI Business Growth Prompt Guide linked in the original email.

Source: the first business release email. The campaign images are original email assets; the worksheets, code, and expanded prompts above are teaching examples prepared for this article.