Proactive signals for operational risk.
Precursory is the reasoning layer for freight operations. It detects operational risk, aligns your teams, and acts while there's still time.
Risk in freight builds quietly: revenue left uncovered, margin that slips, service that breaks. The signals you need already exist. But they sit scattered across teams, and arrive a step too slow.
Powered by Nucleus, a reasoning model purpose-built for freight, trained on real operations data and fine-tuned by the operators who live the trade-offs every day.
We stand behind it: when Precursory misses expectations, we pay. Under development
// HOW IT WORKS
Detect. Align. Act.
You already have the signals. Precursory detects the risk, aligns every team around it, and acts while there's still time to change the outcome.
Pattern detection
Detecting the risks that matter, early enough to act.
Static averages and rules flag problems late, and repeat issues go unnoticed.
Learns patterns from history, flags risk early and ranks what needs attention.
Coordination
Aligning the right information and people on the same decision.
Information sits in separate systems, and handoffs between people drop context.
Keeps one shared record so everyone works from the same picture.
Intervention
Acting by deciding the fix and getting it done.
Operators decide each fix from scratch, so quality depends on who is on shift.
Proposes the fix; the operator confirms, and people or AI agents execute.
// MEET NUCLEUS
Reasoning model, purpose-built for freight. Trained to act like your best operator.
How do you teach a model to act like your best operator?
You start with real operations data, then let your planners, fleet managers and pricing specialists show it the trade-offs they make every day. That is how Nucleus learned to detect risk the way your best operator would notice it, align the right people and information the way they would, and act the way they would. The difference is that Nucleus never stops reasoning, weighing thousands of scenarios while there's still time to act.
And it is fine-tuned for your operations, so it stays with you and keeps learning as your team does.
// USE CASES
Three risks. One model.
Freight operations is a repeated coordination game, played across many teams at once. Risk builds in three places: revenue, margin and service.
Turn to see how Precursory detects, aligns and acts on each.
// MARGIN
Do you know exactly how much margin risk you're carrying — before you quote it?
on this lane
within 50 miles
on this load
With that much confirmed capacity in view, Precursory priced aggressively with the shipper — closer to what the market would actually bear — while still protecting margin, because the carrier side of the deal was no longer the risk. The surplus a rep couldn't see in the moment is exactly what let the price move.
Nucleus corrects the market baseline against real-time capacity — pulled from Motive and Samsara geolocation data — and load-specific constraints, so the price reflects the load that's actually being quoted. It reasons for margin directly, not the win, so the trade-off the business needs is already built into the number. And it weighs the market, capacity, and constraints as multiple scenarios at once, fast enough to send the quote to the shipper and lock the rate with a third-party carrier before the picture has had time to shift.
Emails, boards and APIs
Telematics, chats and emails
DAT, SONAR and in-network historicals
Reads and reconciles every source into one record
Reasons across demand, supply and price
Learns the brokerage’s own history
Best-fit truck for each load
Offer priced from market and network
Margin check before the broker confirms
Sets the guardrail on every match
Learns preferred carriers and past waits
Encodes each account’s requirements
Connects to the brokerage’s systems
// SERVICE
Do you know exactly which load is a risk to your performance — before it's too late to intervene?
pickup
made
despite the hour lost
The moment the pickup delay hit, Precursory modeled the cascade — not just to this delivery, but to the load already planned after it. It surfaced the fix to dispatch: hold the second seat-swap until after delivery, skip the fuel break since there was enough fuel to reach Joliet, all within HOS limits — then kept watching to confirm the next load stayed protected too.
Nucleus plays forward the scenarios of this coordination game and catches the delay while there's still a window to act. It connects the signals that today sit scattered — telematics, appointment windows, customer SLA terms, traffic, and weather — so a delay's downstream impact is visible the moment it starts, not after it's spread. It reasons across planners, fleet managers, and customer success together — routing efficiency, the driver relationship, the customer relationship — instead of any one function's goal alone, so the response reflects the full consequence. And it weighs every affected load as a parallel scenario in real time, fast enough that there's still a real choice to make: reroute, adjust, or notify the customer, before anything is actually late.
Mastery and TMW
FleetLocate, OrbComm, Platform Science
Calls, texts and emails
Reads and reconciles every source into one load record
Reasons across plan, position and people
Learns your own history
Loads ranked by risk and time to act
A proposed fix for each load
Operator confirms; people or agents execute
Sets on-time pickup and delivery goals
Learns facility waits and lane delays
Encodes each customer’s appointment needs
Connects to your systems and agents
// REVENUE
Do you know exactly where risk is building in your network — before it leaks revenue?
Idaho read as balanced through day two — nothing looked wrong from the top down. But as new loads kept tendering against thinning capacity, the region started draining, a trend no single planner watching their own loads would have caught in time. By day four, Precursory had already flagged it and recommended the fix: redirect a load to a neighboring lane, and reposition trailers in before the gap became a missed pickup.
Nucleus connects truck and trailer location, driver availability, equipment health, and incoming load signals into one continuously updated view. It reasons for the whole network's balance, not any one load or planner's immediate fill, so today's assignment doesn't strand tomorrow's equipment. And it re-solves the assignment as a rolling set of scenarios in real time, adjusting as loads, drivers, and equipment shift, instead of drifting out of balance between planning cycles.
Loads, orders and dispatch plans
Truck and trailer location, driver hours, equipment health
Calls, texts and emails
Reads and reconciles every source into one network view
Reasons across the whole network, not one load
Re-solves assignments as rolling scenarios
Loads and equipment ranked by imbalance risk
A proposed reposition or reassignment
Planner confirms; people or agents execute
Learns where capacity drains and builds
Tracks availability, hours and health
Encodes each account’s service needs
Connects to your systems and agents
// OUR PROMISE Under development
We warrant our performance.
Most freight tech promises efficiency. We warrant performance: when Precursory misses expectations, we pay.
// TEAM
Built by people who ran the game.
Precursory is built by operators, not just engineers — people who've sat inside the systems this model now reasons about.
Uber and Amazon alum with a PhD in optimization. Built the driver-incentive systems that used predictive and real-time signals to keep supply where it needed to be — and saw firsthand how both systems and people learn to game an incentive. That's the lens he brings to reasoning through freight's coordination problem.
Previously ran the technology team for a freight brokerage. Spent his career building adversarial models and red-teaming systems — the same instinct for finding where a system breaks that now shapes how Precursory's Nucleus is built and tested.
// GET IN TOUCH
See the whole board.
Tired of finding out about problems after they've already cost you? Let's fix that.