Guardian uses GPU-accelerated reinforcement learning to predict long-haul freight delivery times across 786 corridors and 134 US & Mexico freight hubs — more accurately than the routing APIs powering most fleet software.
Integrate in minutes. Start accurate on day one. Get smarter every week.
Sign up free in 30 seconds. No credit card. 25 predictions to try it out. Key arrives instantly.
POST origin + destination. Optional: load type, routing mode, departure time. JSON in, prediction out. One endpoint, any language.
ETA with confidence range, leg-by-leg route, HOS stops, fuel cost, weather forecast, and IFTA mileage. Report actual times and it gets smarter on your lanes.
Head-to-head against industry-leading routing APIs across real freight route predictions with live weather and traffic conditions.
Guardian's prediction is closer to actual truck delivery time 90% of the time.
Average error: Guardian 56 min vs industry routing 258 min (78% more accurate).
Based on 60,000+ live comparisons against leading commercial routing APIs.
Standard routing apps calculate car driving time. Your truck isn't a car. Federal FMCSA regulations add hours that consumer routing completely ignores:
Example: Miami to New York — routing apps say 18 hours. Actual truck delivery: 30+ hours. That's 12 hours your customer is waiting and wondering.
Not just maps. Real freight intelligence that learns from every delivery.
RTX-powered reinforcement learning processes millions of route simulations per second. Gets smarter with every prediction.
Guardian doesn't just check the weather now — it forecasts conditions for when your truck will actually be there. Depart at 6am, arrive Chicago at 8pm? Guardian checks Chicago's 8pm forecast, not current conditions. NOAA alerts, OpenWeatherMap, and storm tracking applied per leg, per arrival window.
TomTom traffic incidents, road closures, and construction detected and factored in. Automatic rerouting suggestions.
Solo driver: 11-hour limit, mandatory 10-hour sleeper berth, 30-minute HOS breaks factored in. Team driver: no sleeper stop — truck runs continuously. Routing apps ignore all of this.
Every prediction returns a confidence score, on-time probability, and a best-case / worst-case window. High confidence tightens the window. Low confidence widens it. Know exactly how much buffer to give your customer.
Fuel cost uses live EIA weekly diesel prices by PADD region — the same U.S. government data the industry runs on, free, no API key. Gulf Coast vs. West Coast diesel is a $1.30/gal difference over 1,000 miles. Guardian knows which states you're crossing and prices accordingly. Toll costs returned separately.
Guardian learns shipper and receiver detention patterns from real outcomes. Flags facilities with above-average wait times so you can price it in before accepting the load.
Live CBP border wait times from the US Customs XML feed. Per-state security multipliers for Tamaulipas, Nuevo León, Chihuahua, and more. Laredo, El Paso, McAllen, San Diego crossings covered.
Pass load_type and Guardian routes differently — not just a label. Hazmat: 55 mph limits, mandatory inspection stops. Tanker: same hazmat rules plus liquid surge handling. Oversize: 25% slower, daylight-only with overnight stops added automatically. Reefer: temp check stops every 8 hours. Five real routing profiles: standard, reefer, hazmat, tanker, oversize.
No app, no login. Text your route and get back ETA, confidence, HOS stop name and location, border wait, weather warnings, and fuel cost — all in one message. Text HELP for FAQ commands, EMAIL you@email.com for a full formatted report. In English or Spanish. Built for drivers at a truck stop with no laptop.
Pass mode: "fuel_saver" and Guardian routes to minimize diesel cost using EIA regional prices — not just shortest distance. Also supports fastest, balanced, and shortest. One param, four optimization strategies.
Every prediction returns miles driven per state. Every fleet files a quarterly IFTA fuel tax return by state — Guardian does the math automatically with every route. No manual calculation.
Truck is mid-run and an incident hits. Call /api/v1/reroute and Guardian checks current conditions, decides if a route change saves time, and returns the recommendation with minutes saved.
Batch endpoint handles 50 routes per call. Drop-in compatible with Samsara, Geotab, and Motive via REST API. Your fleet software calls Guardian — your drivers never change workflows.
One endpoint. JSON in, JSON out. Integrate in minutes, not weeks.
// Request curl -X POST https://api.guardianfreight.ai/api/v1/predict \ -H "Authorization: Bearer gdn_your_key_here" \ -H "Content-Type: application/json" \ -d '{ "origin": "Louisville, KY", "destination": "Atlanta, GA", "departure_time": "2026-03-10T06:00:00", "load_type": "reefer", "mode": "fuel_saver", "team_driver": false }' // Response { "prediction": { "estimated_hours": 8.4, "best_case_hours": 7.6, "worst_case_hours": 9.8, "confidence": 0.91, "on_time_probability": 0.88, "fuel_cost_estimate": 284.50, "toll_cost": 12.75, "total_trip_cost": 297.25, "load_type": "Reefer", "states_traversed": ["KY", "TN", "GA"], "miles_per_state": { "KY": 138, "TN": 121, "GA": 111 }, "arrival_local": "2026-03-10T14:27:00-05:00" }, "route": { "summary": "Louisville, KY → Nashville, TN → Atlanta, GA", "hos_stops": [{ "after_leg": 1, "reason": "FMCSA mandatory 30-min break after 8 hours driving", "duration_min": 30, "recommended_stop": "Flying J Bowling Green", "stop_interstate": "I-65", "stop_mile_marker": 28, "stop_state": "KY" }] }, "detention": { "city": "Atlanta, GA", "avg_wait_min": 48, "sample_size": 23 }, "conditions": { "weather": "clear", "traffic": "moderate" }, "risks": [], "meta": { "query_time_ms": 138 } }
Prefer Python over raw HTTP? One install and you're talking to Guardian with clean objects instead of JSON dicts.
# Install the client library pip install guardian-freight # Guardian itself lives on our servers. # This just gives you a clean way to talk to it.
from guardian_freight import GuardianClient client = GuardianClient("gdn_your_key_here") # Single route route = client.predict( "Houston, TX", "Atlanta, GA", load_type="reefer", mode="fuel_saver", ) print(route.estimated_hours) # 8.4 print(route.fuel_cost_estimate) # 284.50 print(route.hos_stops) # Flying J Bowling Green, I-65 mi 28 print(route.miles_per_state) # {"KY": 138, "TN": 121, "GA": 111} # Batch — 50 routes in one call results = client.batch([ {"origin": "Dallas, TX", "destination": "Chicago, IL"}, {"origin": "Laredo, TX", "destination": "Monterrey, MX"}, {"origin": "Memphis, TN", "destination": "Miami, FL"}, ])
Full API docs (Swagger UI) · OpenAPI spec · REST endpoint works with any language
Drivers at a truck stop with no laptop can text a route and get a full prediction back in seconds. English or Spanish.
| Louisville KY to Jacksonville FL | Full route prediction |
| ARRIVED / 11.5 | Record actual delivery time |
| HELP | List all commands |
| CITIES | Coverage & hub list |
| PRICING | Plans & prices |
| EMAIL you@email.com | Full report by email (opt-in) |
| NOEMAIL | Stop email reports |
Works with hazmat, tanker, reefer, oversize in the message. En español también — ver página en español.
Every plan gets the same complete data. You pay for volume, not features.
HERE Maps charges extra for weather layers. Samsara charges extra for detention analytics. Guardian doesn’t.
Submit real delivery times and Guardian trains on your specific corridors. The more lanes you run, the more accurate it gets for your network.
Powered by real data — not guesses
Guardian specializes in what matters most to fleet dispatchers: interstate and cross-border corridor routing between 134 major US and Mexico freight hubs.
Every time you report a real delivery time, Guardian trains on that corridor. Run the same lanes every week and submit outcomes — Guardian gets more accurate for your specific routes over time. Your data. Your accuracy.
Works on any corridor — including routes outside our core network. Guardian routes through known hubs and sharpens on your actual outcomes.
Everything fleet managers and dispatchers ask before signing up.
In our benchmark against industry routing APIs across 60,000+ head-to-head route predictions, Guardian achieves a 90%+ win rate with an average error of ~56 minutes. Industry routing averages ~258 minutes of error on the same routes. Results are from live comparisons using real weather and traffic conditions across 134 US & Mexico freight hubs.
Google Maps and HERE are built for passenger cars. They don't know your truck has an 80,000 lb load, an 11-hour drive limit, a 30-minute mandatory rest stop every 8 hours, or that hazmat routes can't use certain interstates. Guardian is trained exclusively on commercial truck data and bakes FMCSA HOS regulations, load type restrictions, and detention time history into every prediction.
134 freight hubs across the US and Mexico — 124 US cities coast to coast plus 10 Mexico cities (Monterrey, Guadalajara, Mexico City, Nuevo Laredo, Matamoros, Reynosa, Saltillo, Ciudad Juárez, San Luis Potosí, Tijuana). 786 direct corridors total. For routes between hubs we cover, Guardian routes through the network. Use the /hubs endpoint for the full list.
No. Guardian is an API — it plugs into your existing dispatch software, TMS, or any app you already use. Drivers never touch it. If you want drivers to interact directly, they can text a route to Guardian's number and get a full prediction by SMS. No app download, no login, no training required.
Guardian routes through its hub network to reach any point. For example, a request from a small town near Dallas to a suburb of Chicago gets routed Dallas → St. Louis → Chicago and each leg is predicted individually. Confidence will be lower on unfamiliar corridors, but the neural network uses geographic interpolation to estimate reasonably. Submit actual delivery times with the outcome endpoint and Guardian trains on your specific lanes.
Two ways. First, Guardian trains continuously on trillions of GPU-simulated deliveries — it never stops learning from its own models. Second, every real delivery outcome you report via the /outcome endpoint feeds directly into the Q-table and CorrectionNet. Enterprise customers running the same lanes weekly see accuracy improve specifically on their corridors over time.
The free tier gives you 25 predictions total to evaluate Guardian — no credit card required. Full access to every feature: route prediction, HOS stops, fuel estimates, IFTA breakdown, outcome recording, SMS interface. Once you've seen it work, pick the plan that fits your volume.
Yes. Guardian covers 10 Mexico cities and includes real CBP border wait times at commercial crossings (Laredo, El Paso, McAllen, San Diego) updated every 30 minutes. Mexico state security factors are applied per corridor. The SMS interface works in Spanish — drivers can text in Spanish and receive responses in Spanish.
Join the freight companies using AI that outperforms HERE Maps, the engine behind most fleet software.
Start Free — No Credit Card