Logistics & Supply Chain
The evidence of your next loss is already in your systems.
An entry held for a missing paper while the port charge runs. A refrigerated load drifting warm overnight. A claim still inside its filing window. All recorded somewhere already, and nobody with the hours to assemble it. This page is Runink FACE, the product this company is built around.
Where the decision happens
- Records you already keepSpread across the systems you already run.
- One finding, with its evidenceDrawn together into a single proposed action.
- A person approves itThe drafted action waits; approving is what sends it, under their name.
- The work goes outAnd the reason stays on the record.
Where it goes wrong
-
The hold nobody joined to the cost
An entry sits at the port for a missing paper. The notice arrives with everything else from overnight, and the daily charge runs until somebody joins it to the invoice.
-
Claims that expire quietly
One claim means the carrier's receipt, the weighbridge reading, the rate on the date, the deadline and a drafted letter. Only the largest are worth the morning.
-
Temperature you read after the write-off
A container drifts out of range overnight and the reading is already in the sensor data. The window between saving the load and writing it off is hours.
-
Demand that grew on the way up the chain
Every stage rounds up to a full case and adds a safety margin, so the plant builds for demand that never existed. The amplification lives in the sequence, across four systems.
Who owns this
The people whose week it changes.
-
Operations director
You find out before your customer does, and the fix is drafted when you do.
-
Chief financial officer
Recoveries stop being a heroic effort and become a pipeline you can read.
-
Trade compliance
A held entry arrives with the reason, the papers that release it and the charge accruing.
-
Head of planning
Safety margins are argued from your own numbers, not by seniority.
-
Chief information officer
A shorter security review, because no outside AI service reads the data.
What Runink FACE changes
- A new instance shows you an empty queue. FACE arrives holding no findings about your operation and does not manufacture any — what appears is what it read once you connected it to something.
- What does appear is a specific proposed action with the rule it invoked and the records it cited attached, in one queue, rather than a dashboard for somebody to interpret.
- A named person approves, edits or rejects each one, and the decision is written down as an event carrying who decided and what they changed.
- Approving is what sends it, and the reply names every step: what ran, and any step that could not run, with its reason. That holds on every reply, including one where half the work went out.
- Where a check could not run, the answer is that it could not run — written out as not a finding that the thing is compliant. A quantity nobody measured is kept as unmeasured with a reason rather than rounded to zero, and a connection nobody has contacted is never reported as verified.
- Later, why a claim was filed or an entry held is answered from the record.
The jobs this covers
Each one has a page of its own: what it reads, what it hands to a person, and how you would know it worked.
-
Reverse logistics — returns, and what to do with them
Where a return goes — shelf, refurbishment, recycling or disposal — drafted at the scan, from a policy written down so that the same grade gets the same answer every time.
-
Cold chain, and what the yard camera already sees
A frame from a yard or infrared camera, checked to be an image and read by a vision model where FACE runs, not at an outside service, against the handover record and the paperwork the load travels with.
-
Route planning that keeps up with the day
The plan that was right at six in the morning is not right by ten, and nobody re-runs it because re-running it means planning a day by hand. This is about making the second run cost nothing.
-
Stock cover and supplier planning
Seeing a stockout while there is still time to order normally, rather than after the safety stock has gone and the difference is being paid in air freight.
-
Demand forecasting
A line starts moving weeks before the reorder point does. Reading that turn out of your own order history while it is still a forecasting problem.
-
Freight claims and port charges
The entry, the hold, the reason, the documents missing from the file, the days held and the per-day charge, gathered and joined, so a reviewer is left judging whether the case holds.
-
Seeing the whole chain, not your end of it
Each supplier, carrier and warehouse holds one piece of the picture, in its own format. The picture itself is not held anywhere, and assembling it is the morning nobody has.
-
Hands-free dispatch, for drivers without a screen
The person who knows a load is late is driving. Asking and answering out loud puts the fact on the record while the truck keeps moving, and the desk still makes the change.
-
Testing a plan before you commit to it
A port shuts and you have a day to pick a new route. State the change and the rules it touches — reorder points, lead times, service commitments — and read back which rules it collides with, in what order.
-
Customer data privacy and audit records
Personal details stripped out of log and diagnostic output before it is written, and every check keeping a record of its own working.
How you will know it worked
Every figure below is yours, not ours. Write down where you stand today — the baseline stops being recoverable the moment things improve.
| The measure | Your figure today | What moves it |
|---|---|---|
| Demurrage and detention days | Terminal and carrier invoices and the per-diem lines your forwarder passes through. A full quarter, by port, separating document-caused days from congestion. | Down on the document-caused share. A hold arrives as a running clock: the entry, the port, the reason, which papers would release it, how long it has sat. |
| Claim filing rate, and days from event to filing | From carrier exception records and freight audit: claim-eligible events in a quarter, how many were filed, and where each filing fell inside the carrier's window. | Rate up, days-to-file down. The receipt, the reading, the rate on the date and the deadline arrive assembled from the records FACE could reach — and the pieces it could not reach are named rather than left as a gap the reviewer has to notice. The reviewer judges whether the case holds, which is the part that needs a person. |
| Days from event to detection | Sample last quarter's holds, excursions and late vessels: when each was recorded, against when a person first acted. Few operations have counted this gap. | Down, and the rest depend on it. Checks run against every record you have connected rather than a sample somebody had time for, on a schedule you set rather than at period end. |
| Spoilage and write-off on temperature-controlled stock | The write-off account in your ledger and the quality rejection log for the same period, with temperature excursions separated from other causes. | Down, by moving loads back inside the window where an excursion is still a save. FACE reads the readings where your own systems land them — the database, the warehouse, the object store — and drafts the corrective dispatch from there. |
| OTIF, or DIFOT if that is your term | Your transport or warehouse system: confirmed delivery against the date and quantity promised on the order line, monthly and by customer. | Up, through the avoidable failures. A disruption still running arrives with the orders and customers it touches identified. A lane can then be put to the routing service you configure, and comes back with the distance and duration it returned — or with the word unavailable, which is the honest answer and the one a dispatcher can act on, rather than a blank card that reads like a measured route of zero. |
| Forecast error and days of stock cover | Your planning system's forecast against actuals by line, days of cover per line, and expedited freight coded as premium or air in accounts payable. | Error stated instead of assumed. Your order history goes through the sequence a statistician would run by hand — describe it, test whether it is stationary, read the autocorrelation, fit both a decomposition and an ARIMA, choose between them by backtest — and what comes back names the model that won and the test it was chosen against. Under five observations it declines to forecast rather than drawing a line through them. |
Two things that make the above possible
-
No outside AI service reads your data
The order files, the customs papers, the readings and the reasoning about them stay on the machines your plan names (your own servers or cloud account, or Runink's shared machines), and the model FACE reasons with runs there too: no third-party model dependency, and exactly one inference endpoint — the one set up for your plan. Two paths do deliberately reach outside, because they must: a route request goes to the routing service you configure, and open-web research puts a query to a public search endpoint. Neither carries your records.
- What you would notice
- The security review a shipper runs before it will let you hold its order data, and the customer-data clauses that today need an exception. With no model provider in the path, both have less to argue with. Ask us to walk your security lead through the boundary.
-
Open-web research with no account attached to it
When an answer needs the open web — a carrier's standing, a customs ruling, a published rate, a party you are unsure about shipping to — the search runs from the machines your plan names through a public search endpoint, and a browser on those same machines then fetches and reads the pages behind the results.
- What you would notice
- The search engine sees the query, as it would from any browser. What you get is the absence of an account: no API key, no vendor contract, no per-question bill, so no supplier is accumulating a searchable history of the consignees your company has been asking about, filed under your company's name. It also means how much checking a shipment gets is decided by the shipment rather than by a monthly query cap.
Who you would be working with
Runink is founder-led. The person in the first meeting is the person who designed the thing being discussed — his track record, what he does here, and the means to check any of it are on the company page.
See whether it fits
Bring one lane, one carrier, or one month of returns. A short conversation is usually enough to tell whether the losses you carry are the shape this addresses. It is built around freight forwarders and third-party logistics providers, manufacturers with international inbound flows, and food, pharmaceutical and chemical distributors.
Runink FACE is the product behind this page. The paper is the long version — what it reads, what it produces, who approves it and where it runs. It carries no case studies, no customer names and no return-on-investment figures.