LanePal Demo Tutorial

1. Before You Start

• Open the demo URL in Chrome on a large screen (external monitor recommended)
• Check that the connection status shows "Connected" in the top-right corner
• For the inbound demo: have a second phone ready to call +1 (626) 842-9780
• For the outbound demo: Andy (+1 425-230-7002) and Hawk (+1 630-639-8711) are pre-loaded as test contacts
• Pick an active agent from the toggle buttons (Andy Cell or Alt Phone) before starting a call

2. Inbound Demo — Step by Step

Making the Call

1. Open dashboard.lanepal.ai and confirm the status badge says Connected
2. Have someone call +1 (626) 842-9780 from their phone
3. The selected agent's phone rings — answer it
4. Both parties are now bridged in a Chime Meeting (they hear each other normally)
5. The dashboard shows the call as Ringing, then Transcribing

Caller Script — What to Say

The caller should talk naturally about freight needs. The AI listens for cities, dates, equipment types, commodities, and capacity.

"Hi, I'm looking for a reefer load from Seattle to Portland, sometime after April 15th."

"Do you have any flatbed capacity going from Boise to Spokane? I need about 30,000 pounds moved."

"I've got a truck sitting in Yakima and I need a load going anywhere in Oregon. Dry van, full truckload."

"We need to move frozen seafood from Bellingham to Eugene by April 20th. What do you have?"

"I'm checking rates on the Seattle to Boise lane. What's your rate per mile for a dry van?"

What Appears on Screen

1. Live Transcript — appears within 1-2 seconds of speaking, labeled Agent/Customer
2. AI Suggestion — within ~1.5 seconds of hearing a city name or freight keyword: tag pills show the extracted params (From: Seattle, To: Portland, Reefer), a summary says "Found 7 loads from Seattle to Portland...", and a table shows carrier name, rate per mile, total rate, distance, weight, commodity, equipment, and broker credit score
3. Classification — after 4 transcript segments, GPT-4o-mini classifies the call (dispatch, inquiry, booking, complaint)

Follow-Up Phrases That Trigger New Suggestions

The AI re-analyzes whenever new intent is detected. The agent can steer the conversation:

"Let me check on that for you... So you need a reefer from Seattle to Portland?"
— AI picks up the confirmation and re-queries

"What about Tacoma instead of Seattle? We might have more options."
— AI detects the new origin city and queries again

"And you said you need refrigerated? What temperature?"
— AI adds reefer_temp_f to the query

"Are you looking for full truckload or would a partial work?"
— AI switches between FTL/LTL/Partial

Manual Query (Fallback)

If the proactive AI doesn't trigger, or you want to show a specific query, use the "Ask AI" input box:

1. Click the "Ask AI" input box on the dashboard
2. Type a natural language question. Results appear with the generated SQL visible.

Example queries:

"Show me all available reefer loads from Seattle after April 20"

"Which brokers have credit scores above 80 and pay within 7 days?"

"Flatbed loads under $3/mile from Portland to Boise"

"What commodities are moving from Yakima this week?"

3. Outbound Demo — Step by Step

Adding Contacts

1. Open outbound-demo.lanepal.ai
2. Two test contacts are pre-loaded: Andy Barr and Hawk
3. To add a new contact (e.g. the investor's number): scroll to the bottom of the contact list, fill in Name, Company, Phone, and Notes, then click "+ Add to List"
4. The contact list persists in the browser — it survives page refreshes

Running a Dial Session

1. Click Start Session (the green pulsing button)
2. The system calls the first contact on the list
3. When they answer, the call area shows: contact name, company, notes, call timer, live transcript, and AI suggestions (once freight is mentioned)
4. When the call ends, set a disposition:
   • Interested — contact wants to proceed
   • Callback — call back later
   • Not Interested — no fit
   • No Answer — didn't pick up
5. If Auto-Dial is on, the system calls the next contact automatically after a configurable delay (1-10 seconds)
6. Session stats update in real time: calls made, connected, average duration, interested count

What to Say (Outbound Script)

Since you are calling them, you lead the conversation:

"Hi [Name], this is [Agent] from LanePal Dispatch. I'm checking in to see if you have any loads that need moving this week. What lanes are you looking at?"

Then let them respond naturally. The AI will pick up on any cities, equipment types, or dates they mention. Good follow-up questions:

"Are you looking at the Seattle to Portland lane, or more eastbound toward Spokane?"

"What equipment do you need — dry van, reefer, flatbed?"

"What kind of rate per mile are you looking for on that lane?"

"Is that a full truckload or would you consider a partial?"

"What commodity is it? We want to make sure we match you with the right carrier."

4. The DEBUG View — Architecture Visualization

1. On any demo page, click the "DEBUG" button in the bottom-right corner
2. A live SVG architecture diagram appears, showing every component of the LanePal pipeline
3. As events flow through the system, each component lights up in real time — you can see the caller audio move through Chime, KVS, Transcribe, the AI, and back to the dashboard
4. Click "Simulate Load" to generate simulated traffic
5. Use the speed controls: 1x (realistic pace) → 5x (busy day) → 20x (heavy load) → 50x (stress test)

What Each Speed Demonstrates

5. Scale Demo — Investor Pitch

1. Open scale-demo.lanepal.ai — the simulation starts automatically
2. Four companies are shown simultaneously: Cascade Freight (Seattle), Pacific Haulers (Portland), Idaho Express (Boise), Montana Logistics (Missoula)
3. The shared architecture diagram in the center lights up as events flow through
4. Use the speed controls: 1x (realistic), 5x (busy day), 10x (at scale)
5. Each tenant card shows their active calls, event log, and status

Key Talking Points

• "One phone number per client, not one per agent or per call."
• "The architecture diagram in the center is shared — no per-client infrastructure."
• "At 10x speed, you're looking at what a busy Monday morning looks like with 4 clients. Same Lambda, same backend, same AI."
• "Adding a fifth client takes 30 seconds and 3 CLI commands."

6. What the AI Understands

These words in the transcript trigger immediate AI analysis (no waiting for segment thresholds):

Natural Phrases That Work

• "under $3 a mile" or "what's your rate"
• "after April 15th" or "next week" or "before the 20th"
• "30,000 pounds" or "full truckload" or "partial"
• "frozen" or "34 degrees" or "keep it cold"
• "produce", "frozen foods", "lumber", "electronics"
• "reliable broker", "quick pay", "good credit"

7. Coming Soon: DAT Integration

When we integrate with DAT, the demos work exactly the same way. Here is what changes:

New Capabilities DAT Enables

• Market rate context: "The average rate on the Seattle to Portland lane this week is $2.85/mi. This load at $3.20/mi is above market."
• Load freshness: "This load was posted 2 hours ago" vs "This has been sitting for 3 days — broker may be flexible on rate."
• Deadhead calculation: When the caller says "My truck is in Yakima," the AI calculates deadhead miles to each load's pickup point.
• Broker reputation: Real DAT credit scores and days-to-pay, not simulated.
• Equipment matching: Full 60+ equipment type codes, not just the 5 primary types.

What Stays the Same

• The AI pipeline (transcribe → extract intent → query → summarize)
• The ~1.5 second response time
• The dashboard UI and outbound dialer
• The WebSocket real-time push
• The proactive AI trigger logic (keyword detection + intent extraction)

8. Troubleshooting