Leadership Demo | ARC Conversational AI

đź’¬ AI Text Bot Demo From product questions to safer shopping guidance

This demo is more than a showcase of a text conversation. It represents several weeks of testing, refining, and tightening the way ARC’s virtual assistants answer product questions, handle uncertain matches, and guide customers to the right next step. The result is a safer, more consistent, store-specific AI shopping experience that supports staff, helps customers, and creates a service advantage competitors are not currently offering.

Store-specific inventory Estimated guide pricing Product Finder fallback E-commerce pickup links Clickable store phone support Uniform assistant structure low stock numbers: <5



Why this matters

Product questions are one of the most common reasons customers call or message a liquor store. They want to know if an item is available, what size it comes in, how much it may cost, and whether there are similar options.

Early testing showed that a generic AI answer is not good enough for product availability. If the assistant is uncertain, it must not guess. It must protect the customer experience, protect the store, and offer a helpful next step.

That is the important shift: ARC’s assistants are no longer being treated as simple chatbots. They are being developed as controlled, store-level shopping support tools.

The leadership takeaway

The achievement is not that AI is perfect. The achievement is that ARC now has a repeatable model for using AI responsibly.

  • It avoids confident answers when the product match is uncertain.
  • It uses approved store-specific product data.
  • It gives customers safer language around price and availability.
  • It routes customers to Product Finder, e-commerce, or the store phone number when needed.
  • It creates a more consistent customer experience across ARC locations.
The development journey

From AI answering to AI accuracy control

The Woodford Reserve test at Guisachan became an important learning moment. It showed why product availability, pricing, size, and store-specific inventory cannot be left to assumption. From that point forward, the focus became clear: build a safer formula that gives useful answers without allowing the assistant to overpromise.




Tested Real product questions were used to expose weak points such as unclear matches, similar branded products, pack sizes, and store-specific stock differences.
Tightened The assistant prompt was refined to prevent guessing, avoid blending product rows, and use safer wording when certainty is not possible.
Standardized The improved structure can now be applied across ARC’s virtual assistant team with store-specific inventory, links, and phone numbers.


“The goal is not to make the AI sound confident. The goal is to make the AI useful, honest, and consistent — even when the answer is not perfectly certain.”

The current ARC model

The safer customer pathway

The updated assistant experience is designed to keep the customer moving forward without pretending the technology is flawless. If the assistant can answer confidently from approved product data, it does. If not, it gives the customer helpful options instead of guessing.




What customers get


  • Fast product guidance without waiting on hold
  • Estimated guide pricing instead of overpromised pricing
  • Clear stock language that explains availability can change
  • A helpful Product Finder link when the assistant may have missed the item
  • The store’s e-commerce link for the most up-to-date pickup availability
  • A clickable store phone number when they want to speak with a real assistant

What each store gets


  • Fewer repetitive product-check interruptions
  • A digital assistant that supports customers while staff are busy in-store
  • A store-specific Product Finder page that helps customers browse before calling or visiting
  • A safer answer structure that reduces the risk of inaccurate product claims
  • A consistent customer experience across ARC locations
  • A modern service advantage that competitors are not currently offering their customer base
Store Product Finder

Every store has its own customer-friendly safety net

Each ARC location now has a designated Store Product Finder page. This gives customers a simple way to browse the store’s product list, search for similar items, and continue exploring when the assistant cannot confidently identify one exact match.




The key fallback phrase

One of the most important improvements is the softer, more honest not-found response:

“I may have missed the item.”

This wording avoids making the assistant sound broken or blunt. It keeps the tone helpful, admits uncertainty, and immediately gives the customer a better next step through the Store Product Finder, e-commerce link, or direct phone support.

When Product Finder helps


  • When the customer uses a nickname, partial product name, or spelling variation
  • When several branded products may match the request
  • When the customer wants to browse similar options
  • When the assistant cannot safely confirm an exact match
  • When customers prefer to self-serve before calling the store
The store CSV process

Why the weekly store CSV matters

The AI assistant and Product Finder page are only as useful as the product data behind them. This is why the store CSV file matters. It is not extra admin for the sake of technology — it is the data foundation that allows each store to offer a modern, store-specific customer service experience.




What the store team provides


  • A clean product CSV in the agreed format
  • Product names and descriptions that can be understood by customers
  • Size, pack, category, SKU, price, and OnHand information
  • A reliable inventory snapshot that can be transformed into AI-ready knowledge

What that CSV powers


  • The store’s AI text assistant
  • The store’s Product Finder page
  • Safer product lookup responses
  • Estimated guide price answers
  • Customer-friendly fallback links
  • A consistent AI experience across ARC stores


“The CSV is not just a spreadsheet. It is the fuel that powers the store’s AI assistant, Product Finder, and customer shopping support system.”

Behind the scenes

How the system works

The process creates a repeatable operating model for each store. The customer sees a smooth conversation. Behind the scenes, the system depends on clean product data, prompt guardrails, and store-specific links.




1

Store prepares the CSV

The store provides the product file in the agreed format so the data can be processed cleanly.

2

CSV is transformed for AI use

The product file is cleaned and structured for the assistant, including product name, size, category, SKU, estimated guide price, and OnHand snapshot.

3

Product Finder is updated

The store’s Product Finder page gives customers a self-serve way to browse and search the store’s product range.

4

Assistant answers from approved data

The AI assistant uses the store-specific knowledge structure instead of guessing from general knowledge.

5

Customer receives safe next steps

The assistant provides an answer when confident, or directs the customer to Product Finder, e-commerce pickup, or the store phone number.

6

Store staff help promote adoption

The more customers know the service exists, the more it can reduce simple product-check friction and improve customer satisfaction.

Screenshot proof

Real text conversations, real operating improvements

The following screenshots can be used to show the assistant in action: answering product questions, handling uncertain matches safely, offering Product Finder support, directing customers to e-commerce, and providing a phone option when human confirmation is preferred.




Screenshot of ARC AI text assistant answering a product question

Screenshot 1: Customer asks a product question. The assistant responds with clear product guidance using the approved store-specific structure.

Screenshot of ARC AI text assistant giving estimated guide price and stock wording

Screenshot 2: Estimated guide price and stock snapshot language help the customer while avoiding overpromising.

Screenshot of ARC AI text assistant offering Product Finder fallback

Screenshot 3: The assistant uses the “I may have missed the item” style of response and directs the customer to the Store Product Finder.


Screenshot of ARC AI text assistant offering e-commerce pickup link

Screenshot 4: The customer is guided to the store’s e-commerce page for the most up-to-date pickup availability.

Screenshot of ARC AI text assistant offering clickable store phone support

Screenshot 5: The assistant gives the customer a direct phone option when they want to speak with store staff.

Screenshot of ARC AI text assistant showing a complete customer support pathway

Screenshot 6: A complete support pathway: AI answer, Product Finder, e-commerce, and real staff support when needed. For stock items 5 or less there is additional note for the customer to check product availability before coming down to the store for purchase.

Store team adoption

Why staff support matters

The technology becomes more valuable when customers know it exists. Store staff do not need to explain the technical side. They simply need to help customers become aware that each store now has additional digital support for product questions, browsing, and pickup guidance.




Mention it in-store

If a customer asks about checking stock later, staff can mention the store’s AI assistant and Product Finder page.

Use it on signage

Shelf talkers, counter cards, and website links can help customers discover the service without adding pressure to staff.

Promote it as support

This should be positioned as a support layer for the store team, not a replacement for human service.



The simple message for store managers

Your weekly CSV helps power your store’s AI assistant and Product Finder page. The better the product file, the better the customer experience. With staff support, this becomes a practical tool that can reduce simple product-check friction, help customers shop with confidence, and give ARC a customer service advantage in the market.

Final summary

Powered by people. Enhanced by AI.

ARC’s AI text assistants are designed to work alongside store teams. They help answer product questions, reduce simple friction, guide customers to Product Finder and e-commerce, and provide a direct path to real staff when needed.

The bigger win is consistency: every store can have its own assistant, its own Product Finder, its own e-commerce link, its own phone support, and the same safer answer structure behind the scenes.

For Example store-specific Product Finder page is for our Revelstock Website, e-commerce URL, is for the Harvy Store and phone number is for the Guisachan Village store.