Ralixar

Platform

Ralixar GTM — Global Trade Management built for India, designed for the world.

From purchase order to out-of-charge to bank realisation: one deterministic platform, one shipment record, AI at every step.

Book a Demo

Humans decide. AI executes. Workflows own the process.

app.ralixar.com / shipments

OPERATIONS

Work queue
Shipments
Documents
Customs
Finance

IMP-26-0148

Mumbai import · Action queue

3 risks ranked

CLEARANCE

RMS review

FREE TIME

18 hours

LANDED COST

Ready

Bill of Entry validation

92% confidence

Review

e-Sanchit document set

8 / 8 complete

Ready

Drayage slot

JNPT · tomorrow

Planned

RALIXAR FIELD

Stuffing check

Container and seal captured

Submit evidence

Import and export lifecycle

Every milestone has an owner, module and control.

Move through Indian customs and logistics events without breaking the shipment record into email, spreadsheets and disconnected portals.

  1. Purchase order

    Establish parties, terms, items and shipment intent.

  2. Supplier documents

    Ingest invoice, packing list and certificates.

  3. Product normalisation

    Resolve item descriptions and master-data gaps.

  4. HS classification

    Present top-three HS suggestions with evidence.

  5. PGA requirement check

    Apply applicable agency and licence rules.

  6. Restricted-party screening

    Check parties with fuzzy-match review.

  7. Advance BoE preparation

    Assemble validated Bill of Entry data.

  8. Freight and booking

    Confirm rates, space and routing.

  9. Arrival and IGM / SCMTR

    Reconcile arrival and manifest milestones.

  10. RMS outcome

    Route assessment, examination or query work.

  11. Duty assessment

    Compute duty and landed-cost components.

  12. Duty payment — ECL

    Control payment readiness and evidence.

  13. Out-of-charge

    Capture customs release and accountable hand-off.

  14. Drayage

    Plan pickup, free time and final movement.

  15. GRN and settlement

    Close receipt, cost and payment evidence.

One engine. Complete controls.

Six moments every EXIM team knows — and how Ralixar handles each one.

Each capability cluster below starts with a real moment in your week, shows what the AI does and what your team still decides, and names the outcome it moves.

loop · illustrative data
01 · TRADE COMPLIANCE — THE NEW-SKU MOMENT

A new product, classified and screened before the PO is released.

A buyer adds a Wi-Fi router board to next month's purchase order. Usually someone searches the tariff, picks a code, and the guess returns six weeks later as a customs query. In Ralixar, the classification is ready before the supplier ships.

AI DOES
Proposes 8517.62.90 at 0.94 with two alternatives and the tariff text; flags agency requirements that may apply; screens supplier and consignee.
YOU DO
Confirm the code and clear any screening hit; the rationale is stored for every future shipment of that SKU.
95.5% top-3 accuracy on 129,084 customs cases
PEER-REVIEWED
50–70% fewer false-positive screening alerts
PRACTITIONER
One confirmed rationale reused on every repeat shipment
PLATFORM FACT
KPI MOVED · FIRST-TIME-RIGHT FILINGS
STAYS RULE-BASED · SCOMET, PGA AND POLICY DECISION TABLES
loop · illustrative data
02 · DOCUMENTS AND CUSTOMS — THE 6 A.M. DOCUMENT PACK

Fourteen pages in the inbox at dawn. A filing pack by breakfast.

The supplier's invoice, packing list and bill of lading land overnight, each in its own format. Usually someone keys them into the broker's system mid-morning, and a mismatch surfaces at assessment. In Ralixar, they are read, cross-checked and turned into a Bill of Entry work pack before the team logs in.

AI DOES
Classifies each document; extracts every field with a confidence score; checks invoice against packing list; builds the e-Sanchit set; pre-fills repeat-lane defaults.
YOU DO
Fix the one field it flagged; the customs broker reviews and files on ICEGATE.
82% straight-through processing, 120,000+ documents a year
VENDOR CASE
2–3 minutes to under 45 seconds per container
VENDOR CASE
Replaces a 1–4% manual keying error rate
INDUSTRY DATA
KPI MOVED · STRAIGHT-THROUGH-PROCESSING RATE
STAYS RULE-BASED · DECLARATION ASSEMBLY AND IDEMPOTENT SUBMISSION
loop · illustrative data
03 · DUTY, COST AND SCHEMES — THE FTA CLAIM NOBODY MADE

Know the landed cost — and the preference you're entitled to — before the vessel berths.

Goods from an FTA partner arrive with a certificate of origin, but nobody claims the preference for fear of a CAROTAR query. In Ralixar, eligibility is pre-screened, the origin evidence is checked, and the duty difference is visible before the claim is made.

AI DOES
Flags that the shipment looks eligible; checks the origin evidence is complete; spots an unusual declared value; forecasts when an Advance Authorisation balance will run short.
YOU DO
Decide whether to claim; approve the duty payment under maker-checker.
Every rupee of BCD, SWS and IGST traced to its notification
PLATFORM FACT
Duty payable ₹3,09,800 → ₹1,80,000 on ₹10,00,000 with a verified preference
ILLUSTRATIVE
₹[X] of preference identified per [N] shipments
DESIGN PARTNER
KPI MOVED · LANDED-COST ACCURACY
STAYS RULE-BASED · ALL DUTY AND TAX COMPUTATION
loop · illustrative data
04 · LOGISTICS EXECUTION — THE HALF-EMPTY CONTAINER

Fill the box, hit the cut-off, and keep every weight limit.

An export order is split across two 40-foot containers because the load was planned by eye. In Ralixar, the solver fits it into one, checks VGM and axle load, and Ralixar Field captures the stuffing photos that prove it went to plan.

AI DOES
Builds a 3D load plan; compares routes by cost, time and carbon; recommends a tender award; warns of rollover risk on the booking.
YOU DO
Approve the plan and award; the supervisor signs off stuffing in Ralixar Field.
62% → 89% utilisation vs manual planning
VENDOR CASE
12.3% lower routing cost vs the best baseline
PEER-REVIEWED
About 2.5% gain vs an already-strong solver
PEER-REVIEWED
KPI MOVED · FREIGHT COST PER TEU
STAYS RULE-BASED · VGM, AXLE-LOAD AND CUT-OFF LIMITS
loop · illustrative data
05 · VISIBILITY AND EXCEPTIONS — THE CONTAINER ABOUT TO COST ₹1.8 LAKH

See the money at stake. Clear the costliest container first.

Four containers are in trouble across four ports, and the team chases the one whose owner called loudest. In Ralixar, the queue re-sorts by rupee exposure: the Chennai container with six hours of free time and an open query goes first.

AI DOES
Narrows the ETA band as AIS and terminal signals arrive; forecasts dwell; scores demurrage risk; ranks every exception by ₹ exposure.
YOU DO
Choose what to expedite; every exception has an owner and an SLA.
3.30 h ETA mean absolute error
PEER-REVIEWED
USD 15.4B demurrage and detention from US importers, Apr 2020–Mar 2025
REGULATOR DATA
₹[X] exposure surfaced per 100 containers
DESIGN PARTNER
KPI MOVED · DEMURRAGE AND DETENTION AVOIDED
STAYS RULE-BASED · FREE-TIME CLOCKS, SLA TIMERS AND ESCALATIONS
loop · illustrative data
06 · FINANCE, ANALYTICS AND PLATFORM — THE INVOICE THAT LOOKED RIGHT

Every invoice line matched to its shipment and contract before you pay.

A carrier invoice arrives with two detention lines and a seal charge that look routine, and it would usually be paid as billed. In Ralixar, each line is matched to the shipment and rate card, and the three that don't fit are held for review.

AI DOES
Matches 38 of 41 invoice lines; flags three accessorials; matches remittances to EDPMS and e-BRC; writes the SQL for a plain-language question.
YOU DO
Approve short-pays and disputes; run the query after reading it.
5–15% of freight invoices contain errors; audits save 3–7% of spend
INDUSTRY DATA
About 40% of error value sits in accessorials
INDUSTRY DATA
New countries ship as content packs, not code forks
PLATFORM FACT
KPI MOVED · INVOICE LEAKAGE RECOVERED · CASH-CYCLE DAYS
STAYS RULE-BASED · GL POSTING, EDPMS AND E-BRC CLOSURE

India regulatory rails

Designed around the systems operators actually use.

ICEGATEe-SanchitRMSDGFTITC-HSSCOMETCAROTAREDPMSe-BRCGST

Coverage and integrations are released by country pack and verified against the applicable regulator specification.

Know exposure before free time expires.

Model demurrage and detention risk from milestones, contracted free time and exception status—then route the next action to an accountable owner.

Estimate your demurrage exposure

RALIXAR FIELD

Stuffing evidence

Container OCR · verified
Seal OCR · verified
3 photos · uploaded
Supervisor approval · pending

Field evidence, connected to the shipment record.

Capture stuffing photos, container and seal OCR, milestones and approvals where cargo moves—without rebuilding the context later.

Explore Ralixar Field
THE RALIXAR AI STACK

Six layers. Models in the middle. People and proof on every side.

Ralixar treats AI as a component inside an accountable system. Models sit behind typed contracts and confidence gates, deterministic workflows own every process, people hold every statutory decision, and an audit rail records all of it.

The Ralixar AI Stack
Capability maturity — an honest view of what ships when
Document extraction and classification
Layout-aware document AI
STAGE 1 · LAUNCH
HS top-3 suggestion
Hierarchical classifier plus tariff retrieval
STAGE 1 · LAUNCH
Restricted-party screening
Calibrated fuzzy and transliteration matching
STAGE 1 · LAUNCH
ETA and demurrage risk
Gradient-boosted ensembles; cost-sensitive classifier
STAGE 1 · LAUNCH
Exception ranking and entity resolution
Learning-to-rank; record linkage
STAGE 1 · LAUNCH
Load planning
3D bin-packing heuristics
STAGE 1 · LAUNCH
Freight audit and invoice matching
Probabilistic record linkage; anomaly detection
STAGE 2 · CONSOLIDATE
Route and drayage optimisation
OR-Tools CP-SAT; metaheuristics
STAGE 2 · CONSOLIDATE
Status summaries, query drafts, text-to-SQL
Single-shot language models
STAGE 2 · CONSOLIDATE
Agentic exception resolution
Bounded planner with approval gates
STAGE 3 · ROADMAP
Four things Ralixar will never automate
01
Filing a customs declaration
02
Paying duty
03
Clearing a sanctions match
04
Replying to a customs officer without a person reading it first
Why the approach compounds
Every confirmation is training data
Each human decision labels that tenant's own model, so accuracy rises with use — without pooling customer data.
Regulatory depth ships as content
Tariffs, lists, validators and decision tables live in country packs, so India v1 extends to Singapore, the UAE, the US and the EU without code forks.
Trust shortens the sales cycle
A deterministic core with full model provenance answers the security questionnaire and the customs audit up front.
Stage 3 adds a bounded agent at the exception queue only, with human approval before any irreversible action.
Talk to our architects

Questions operators ask

GTM, without the black box.

Direct answers on scope, controls, implementation and commercial fit.

Bring your hardest workflow into one accountable operating model.