Skip to content
Docs

Get a precise position

Choose an Apex solution, clear its preflight, and read the result with its evidence.

Two states to read first

Product support (claimStatus): what PosFlow states the Apex product supports. Declared technical availability (manifest runtimeStatus + accepted combinations): what a reviewed release can run; the app remains authoritative for live effective availability after operational overrides. A supported solution is therefore not automatically a runnable one. Where a supported solution has its runtime disabled, the workspace shows it as gated instead of accepting the job, and this guide gives you no procedure for forcing it through.

Before you start

Apex opens on the Pro plan or above, and a run additionally consumes credits from your included allowance, a purchased pack, or a grant. The published rate for a solution is the price of one billable station, and a run is charged that rate for every station it carries except a base, which rides free: a network solution over several stations therefore costs several times the rate. Have your observation file ready, plus a base file for the PPK solutions and the other station files for a network solution. Know your antenna type, its radome, and the height you measured: they matter more to your result than any processing option. Accuracy depends on session length, antenna calibration, and data quality, so read the formal uncertainty a solution reports rather than any headline figure.

Steps

  1. Choose a solution. The Apex workspace lists the solutions PosFlow supports and the live availability of each one; a gated solution cannot be submitted, whatever the marketing pages say it supports.
  2. Upload by role. Each solution declares the file roles it accepts, so attach every file to the role it plays: rover, base, or station. A file left in the wrong role is refused before any processing starts.
  3. Verify antenna, radome, and height. Confirm the antenna and radome the workspace resolved for your receiver, then enter the measured height together with the method you used to measure it.
  4. Repair the preflight. Every Apex run is created from a preflight check, and that check reports what it cannot accept: an unresolved antenna, a missing role, a combination the release does not declare. Fix each item and rerun the check until it passes.
  5. Confirm the credit cost and the cancellation terms the preflight quotes you, then submit with the token it returns. There is no way to start a run without a passing preflight.
  6. Follow the validation the run reports. Progress streams live, and a run that fails validation stops and names the reason instead of publishing a result.
  7. Download the result and its evidence: coordinates with their formal uncertainty, the reference frame, and the evidence artifacts the solution declares, so a reviewer can retrace the job.

Kinematic PPP

When the kinematic PPP solution is accepted for a release, it runs the installed official mode: per-epoch pseudo-kinematic PPP, with .KIN and .SUM files as its evidence. It is not a filtered dynamic trajectory, and is not sold as one.

Externally processed PPP

Submitting observations to a managed third-party positioning service is an administrator operated route rather than a self-serve option: an administrator configures it, and results come back under a generic engine label. Nothing here is a procedure for driving one of those services yourself.

Refund boundaries

What happenedCredit outcome
Credit-hold positioning runs: reported platform, engine or output-validation failure.The held credits are released when the failure is settled, including reported engine failures after compute starts.
Legacy charged runs: recognized infrastructure failure such as dispatch, missing stored file, ephemeris download, engine startup, internal error, or unusable analysis output.Automatically credited back in full.
Legacy charged runs: the positioning engine explicitly rejects the input before its first compute binary starts.Automatically credited back in full.
Legacy charged runs: compute started but the data could not be solved, or the failure is not on the refundable-code list.Charged. Running the engine consumes compute even without a usable result.
Credit-hold positioning runs: user cancellation before or after the recorded engine-start marker.Before the marker: release the hold. After the marker: charge it. The marker is set before executor entry, so waiting for an engine slot can already count as started.

Legacy automatic credit reconciliation runs every 10 minutes for tasks submitted within the current UTC calendar month. A legacy task that fails after its submission month ends is outside that refund window. Credit-hold positioning runs use their own settlement path instead.

This table describes quality-check and precise-positioning task credits, not subscription or cash refunds. A failure count is not a refund receipt; consult your credit ledger for the settled amount. Other or unclassified failures are not proof that your data was at fault.

Ready to try it with your own data? The Free plan covers every guide on this page.