In this webinar, PCS Software and CloneOps walk through how AI voice agents are embedded directly into Cortex AI, and how carriers and brokers can deploy standalone agents from the CloneOps Marketplace in minutes.
Transcript
Introduction
Hi, everyone. Thank you so much for joining us today. My name is Daniela Cardona with PCS Software. I’m on the marketing team, and today’s webinar is sponsored and led by our partners at CloneOps.ai. This session is titled “From Embedded AI to Marketplace Agents: Unlocking New Value with PCS plus CloneOps.”
You will see how embedded AI agents are starting to take on real work inside logistics workflows, with a live demo presented by Roger Boza, and a few practical examples of what that looks like.
A quick housekeeping note before we jump in — you can submit your questions at any time using the Q&A panel, and towards the end we’re going to open up questions. If you’d rather ask your questions live, just raise your hand and the team will unmute you so you can join the conversation.
Joining us today is Roger Boza, Chief AI Officer at CloneOps.ai, as well as Alan Alberto, VP of Operations and Partnerships. I appreciate you both being here, and I appreciate the attendees being here as well. I’m going to go off screen and let Alan kick things off.
About CloneOps
Daniela, thank you so much for the introduction. Thank you for having us and for our partnership, and to the attendees — thank you for giving us the half hour of time today. I’ll get started by sharing my screen. I promise this won’t be death by slides, but the slides will be here to kick us off and get us organized.
CloneOps is an AI communication platform. Our core agents focus on voice, supported by email and SMS automation as well. We serve the transportation and logistics vertical — our founders are experts in this space, and really everyone working at CloneOps today is from this environment.
My name is Alan Alberto. I lead the operations and partnerships team here, so I have the pleasure of working very closely with the PCS team. Most of you on the call, I assume, are customers, so you’re in great hands — we’ve really enjoyed working with PCS.
Thank you, everyone, for being here. My name is Roger Boza, and I’m the Chief AI Officer here at CloneOps. I’ve been with the company since the beginning. My background includes a bachelor’s and master’s
in computer science and a PhD in artificial intelligence, focusing on natural language processing and computer vision. My role at CloneOps covers anything that deals with AI — proof of concepts, new technology, innovations, advancements — it usually goes through my hands.
The PCS Partnership: Embedded AI
As you can see, you’re in good hands with Roger. So, where we partner with PCS today — this is the embedded portion. As most of you know, PCS launched Cortex AI earlier this year, and we have the pleasure of powering Cora, who is the backhaul booster agent. Cora, within your Cortex solution or platform, is powered by CloneOps.
What this means for you is seamless integration of AI voice and messaging directly within your PCS solution — you don’t have to add another tool. You’re working directly out of the platform that you operate in today. This leads to efficiencies at scale: you’re able to handle high call volumes, and coming soon is email functionality as well with agents, all directly out of one place.
The business growth angle here is that you’re doing more with less. You don’t have to add additional headcount. Your current teams who use PCS can continue to operate in their platform through Cortex without hiring additional workforce to manage volume — and all of this, of course, inside of your TMS.
Marketplace Agents
In addition to embedded agents, CloneOps offers an array of standalone agents as well, and one of the ways we go to market is with these prepackaged marketplace agents. They can cover anything from screening calls to after-hours support to accounts payable and accounts receivable. Within your CloneOps console, you’re able to hire these agents within seconds and deploy them so that your operating teams have new capabilities driven by AI.
On that note, I’ll pivot to Roger so he can take us for a ride. The first agent we’re going to look at is our carrier screener with Voice ID — it automates the intake and onboarding flow for working with carriers, and the Voice ID portion addresses fraud, theft, and identity risk, protecting your business and every business across the industry. Roger, take it away.
Demo 1: Carrier Screener with Voice ID
Thank you, Alan. I get to do the fun part of the webinar today, which is demoing some of our virtual agents. You should be seeing our Marketplace tab in the platform right now. The marketplace is where you can hire off-the-shelf, ready-to-use agents very quickly — minimal setup, everything is done behind the scenes, and they’re pre-configured to start working within minutes.
For today’s demo, I’m going to show our carrier screener, which uses Highway as a compliance tool along with our Voice ID technology.
Let me introduce Voice ID first. It’s a technology that listens to the voice on a call, does identification to see if we’ve heard that voice before, and makes associations between that voice and DOT numbers as well as MC numbers. The idea is that if a voice is claiming multiple DOT or MC numbers, we’ll send out a warning email to the compliance team so they know that someone on the phone is communicating across multiple DOTs — giving them a heads-up to pay closer attention based on their compliance rules.
Setting up the agent: The setup is straightforward. We give it an agent name, assign a phone number, configure a greeting — for demo purposes we’re using a company called Atlas Logistics, so the agent will say, “Hey, thank you for calling Atlas Logistics. Can I get your DOT number to get started?” We then set the transfer phone number for when a carrier passes screening, add the Highway API key for compliance lookups, and provide email addresses for Voice ID alerts. That’s all the configuration needed. We click Hire, and the agent is live.
Live Call Demo
[Roger calls the agent live]
Agent: Hi, thanks for calling Atlas Logistics. Can I get your DOT number to get started?
Roger: Yes. My DOT number is 01834233.
Agent: Thanks for that. Let me confirm your DOT number — 01834233.
Is that correct?
Roger: Yes, that is correct.
Agent: Great. It looks like you’re qualified to haul for us. What load number or lane are you interested in?
Roger: I’m calling about load number 77432.
Agent: Perfect. And do you know when and where you’ll be empty? Roger: I’m empty right now.
Agent: Thanks for that information. Looks like you’re qualified to haul for us, so let’s get you over to the team. Transferring you now. One moment.
Alan: (answering the transferred call) This is Alan.
Roger: Hey, I’m calling to book a load. Are you guys available? Alan: We are.
So at this point, I would have been transferred to the human rep to do the rate negotiation and continue down the path. But let me show what happened behind the scenes.
The carrier screener first did a phone fraud lookup on Highway using the number I called from, checking if it’s associated with any known fraud. In this case my number came back clean, so it continued. The next step was looking up the DOT number I provided — it pulls in all of the business rules and information for the company, and this one came back as a partial pass, active for the authority, so it moved forward to ask about the load and transfer to Alan.
Additionally, Voice ID was running in the background. My voice is associated with multiple DOT numbers because I call in with different ones for testing, so it triggered a warning. If I hop over to my email, you can see that at 12:11 — just two minutes ago while I was talking — I received the alert. It flags the voice for potential cross-account risk, shows the phone number that called in, the stated DOT number during the call, and historical matches including all the
DOT numbers I’ve said previously, the phone numbers I called from, and the timestamps. This can be sent to multiple email addresses, and the system provides suggested actions — all of which are customizable.
Demo 2: Building a Rate Negotiation Agent from Scratch
Now, jumping back into the platform — I thought it would bring a lot of value to build a virtual agent in real time, step by step, so you can see the whole process from beginning to end. We’re going to hop over to the PCS Partner Platform and create a voice agent.
The scenario: You’re a carrier. You want a virtual agent to call brokers about a load you saw posted online and negotiate the rate. Let’s build it.
Name and description: The agent is called Kyle from Speedy Freight Transport — a natural-sounding outbound agent making calls to brokers on posted loads. Since this agent makes outbound calls, it doesn’t need a greeting. We’ll enable recording so we can review calls in the logs, and we’ll provision a new phone number for it.
System prompt: This is the brains of the operation — where we write all the instructions in plain English. How should the agent behave? What’s the goal? What’s the workflow? I’ve already prepared one here. It includes an agent description, the DOT number and confirmation email address, the agent’s purpose (making outbound calls), personality and tone guidelines, and voice and speech guidelines. We instruct it to sound natural — using phrases like “uhm,” “yeah,” and “okay” — so it comes across as friendly and human-like. There are also sections for handling non-verbal transcription, numbers, times, and rate formatting rules.
The beauty of this is that it’s not programming — it’s just instructions in English. You describe exactly how you want the agent to behave, how it should handle objections, edge cases, and different call scenarios.
System prompt variables: These give the agent the dynamic information it needs for each outbound call:
DOT number — the carrier’s DOT number
Email address — where rate confirmations should be sent Load number — the load the carrier is interested in
Posted rate — the rate seen on the load board, used as a reference Starting rate — where the agent begins the negotiation
Minimum rate — the floor; the agent will not accept anything below this
Empty status — whether the carrier has a truck ready to haul
Transcription, brain, and voice: We select the Scribe engine for transcription (voice to text), set the brain to OpenAI GPT-4o on Precise mode for better consistency, and choose a voice. Since the agent is named Kyle, I’ll use a cloned voice of Kyle Richards, our VP of Sales.
We can also control multilingual mode (supporting about 36 languages), speech speed, voice stability, and similarity to the original voice actor. We’ll leave those at default settings.
Tools: This agent doesn’t need any external tools, but the tools section is where you’d connect the agent to a TMS, CRM, or spreadsheet
— allowing the large language model to pull in data and use it during the conversation.
Post-call analysis: After saving the agent, we configure advanced features. First, a summary prompt — a concise 3–5 sentence summary in natural language covering what happened: was a DOT number exchanged, did we reach an agreement, etc. Next, structured data extraction — we extract key fields from every conversation:
Load status — was the load available, was it already covered, did we book it, or did we not agree on a rate?
Final rate — the agreed-upon rate if a deal was made
Reason for no agreement — if the negotiation didn’t succeed
We also have five evaluation metrics monitored automatically: correctness, call quality, goal achieved, conciseness, and hallucination.
Finally, we add some ambient details: office background noise at 40% volume and keyboard typing sounds at 40%, so the agent sounds like it’s working in a real environment. We save the agent.
Live Outbound Call Demo
To initiate a call, we go to the Operations tab, select the agent, and create a new activity — entering all the variables: phone number, DOT number, email, load number 101, posted rate of $1,200, starting negotiation at $1,500, minimum acceptable rate of $1,375, and current empty status. We save and the agent places the call.
[Agent calls Roger]
Agent: Hello, thank you for calling. Can I get your DOT number to get started?
Kyle (agent): Hey, this is Kyle with Speedy Freight Transport. I’m calling on load 101 — is that one still available?
Agent: I need your DOT number to get started first. Kyle (agent): Sure. The DOT number is 01834233.
Agent: All right, let me check to see if you’re authorized to haul for us. Give me one second.
Kyle (agent): Sure, take your time.
Closing
I want to call out that Roger was able to put together that agent and handle an outbound call in under 10 minutes, right from scratch. The key takeaway here is that you all can do this — you have the keys to the castle to go in and build agents, and we’re here to help enable that.
To learn more from the PCS side, you can email customersuccess@pcssoft.com. On the CloneOps side, feel free to reach out directly:
Alan Alberto: alberto@cloneops.ai Roger Boza: rboza@cloneops.ai
You can also reach out to your PCS representatives and they can provide our contact information as well.
Thank you so much for closing us out, Alan. I want to thank everyone again for attending. We will share a recording of this webinar, and the contact information has been provided above. I hope you all have a wonderful day.
Thank you, everyone. Thank you for joining.
