Dr. Ryan Ries here, back again with this week’s “AI in Your Industry” series. Today we’re covering a bit of a wild card - SaaS, but more specifically, support ticket management.
While this is pretty specific, the use cases in this industry are applicable across the board.
I’ll be running this series for a few weeks so please still reply to this email and let me know what industries you want to hear about next.
Some of the next ones I’ll be covering are:
- Manufacturing
- Sports and Gaming
- Agriculture
- Education
- Energy, Oil & Gas
Also, quick side note, for any Denver-based readers, we have a cool event with CrowdStrike that we’re co-hosting in a couple of weeks on March 26th at Wynkoop. Spots are limited so if you’re in the area and interested in AWS / Security / Cloud Management, sign up!
Use Cases I’ve Personally Worked On
Steel Shire Design: Three Months to Ten Minutes
Steel Shire builds software for transmission companies managing linear infrastructure projects. Their clients were drowning in right-of-way documentation. Easement agreements, property surveys, usage rights, multi-line permissions, all of it filed away in boxes.
Hundreds of thousands of pages.
When a field question came up, analysts would manually comb through those boxes, sometimes for three months, to find a single answer.
I’m overwhelmed just thinking about this.
We built the Right of Way Assistant on Amazon Bedrock. It pulls from structured metadata in Amazon RDS and unstructured documents in S3, and lets analysts ask plain-language questions to get real answers in seconds.
The results of this project for Steel Shire were incredible. What used to take three months now takes about ten minutes. That's a 12,960x improvement in research speed.
What I love about this one is what it revealed beyond speed.
The system surfaces insights clients didn't even know they had, pipeline installation status, traceability data, compliance gaps. Federal regulators are demanding more auditable documentation trails, and suddenly Steel Shire's clients have them automatically. That changes the relationship between vendor and customer in a way that compounds over time.
Fexa: Turning Work Orders Into Conversations
Fexa builds facilities management software for retail and restaurant chains, companies like Tesla and Crate & Barrel.
Their challenge was a classic friction problem. Customers would submit work orders for maintenance issues, only to find out a technician wasn't needed. Simple fix, yet expensive dispatch.
We built an AI-powered chatbot using Amazon Bedrock, Anthropic Claude, and Amazon Kendra. The chatbot sits in front of the work order workflow and offers troubleshooting suggestions before a ticket ever gets created. The tricky design constraint here was that the data source had to be beginner-friendly. Not written for experienced HVAC technicians, but for a store manager who just noticed water dripping from a ceiling tile.
That constraint actually made the problem harder, and more interesting. The customer loved it. Fexa is now exploring using the same architecture for another workflow.
A Large Building Materials Distributor: AI-Powered Customer Support
One of the largest construction materials distributors in North America came to us needing to upgrade how their support engineers handled inbound queries.
They have hundreds of locations, a massive product catalog, and support staff fielding the same questions repeatedly while complex tickets wait.
We built a proof-of-concept support agent using Amazon Q for Business.
The agent handles questions about support incidents and can take action, including creating help desk tickets on behalf of the customer, based entirely on the conversation context.
The agent understands the context and acts without having to do any form-filling or re-explanation.
The success bar we set: answer at least 70% of a representative set of common support queries accurately, as judged by experienced support staff. It cleared that bar.
The architecture is now being extended to additional user interactions across the platform.
What I'm Watching in the Industry
This industry is moving fast because the potential ROI is huge, pretty easily realized, and very obvious.
Here are a few use cases in the industry I’m most interested in:
Intelligent Routing: As you’re probably aware, this is a big trend in the contact center space and with contact center solutions like Amazon Connect. In the support ticket world, the agent can:
- Complete the ticket for the user based on a conversational prompt (no actual form fill required).
- Triage the ticket with articles matching the user’s message and the company’s knowledge base.
- Use internal tools like Salesforce, Confluence, Jira, etc to enrich information.
- Then the agent can route the ticket to the correct support team.
We actually have this at CDW in the form of a little Sasquatch agent named Harold. I’m not going to pretend that Harold works perfectly, but he is very helpful in filling out HR and IT tickets, and the team behind Harold is working constantly to make him the best sasquatch agent out there.
Tying this all together
AI in this field is all about making trapped information accessible.
The right-of-way documents at Steel Shire were always there. The answers to Fexa's customer questions were always there. The support ticket data was always there. AI made existing information usable at a speed and scale that changed what was possible.
Here’s what I want you to take away from this:
Don't start by asking, "What can AI automate?"
Ask instead, "Where does my organization have information that no one can reach fast enough to act on?" That's usually your golden ticket.
The other thing I'll say: the projects that worked well all started with a scoped, well-defined problem. Not "transform our operations with AI," but "cut the time to answer this specific category of field question." Specificity is what lets you measure success and build confidence to go further.
If you’re interested in building out a use case for your manufacturing company, reach out to our sales team here.
Until next time,
Ryan
Now, time for this week's AI-generated image and the prompt I used to create it.
Create an image of a muppet who is stressed trying to handle answering as many support tickets as possible. The muppet is standing in front of a ton of angry customers. Off in the distance, you can see a glorious AI agent that would automate this process for the muppet.
