FreightVeyra

WORKING MVP / AI-POWERED B2B SOFTWARE

Freight invoice review.
Built around AI matching.

For finance and logistics teams comparing carrier charges with agreed fixed rates. Use the application yourself: enter data or import CSV, review AI suggestions, inspect the evidence and export your decision.

Working early-access releaseSelf-service softwareNative model inference
Use the working AI workspace

01 / AI MATCHING

Connect different descriptions.

A sentence-embedding model compares the meaning of invoice descriptions with the agreement descriptions you supply. Review each suggestion before using it.

02 / RATE CHECKS

See the exact difference.

After you confirm a match, fixed-rate arithmetic compares billed and agreed amounts in currency minor units. Missing evidence remains unresolved.

03 / REVIEW RECORDS

Keep the reasoning.

Inspect the supplied agreement text, record follow-up decisions and export a CSV, backup or printable review. Saving on this device is available without an account.

TRY THE ACTUAL PRODUCT

A short product walkthrough.

The workspace opens empty for your own inputs. Examples are optional and labelled.

  1. Enter charges and agreement rates.

    Open the AI workspace. Enter one invoice and the fixed rates that apply, or import your CSV. Select USD, EUR, GBP or NGN.

  2. Run the model.

    Select “Suggest rate matches.” The first run downloads the model. Descriptions are processed on your device; no AI API key is needed.

  3. Confirm the match and inspect the result.

    Review the proposed mapping, confirm it or leave it unresolved, then run the cost check. Inspect billed amounts, agreed rates and the supplied evidence.

  4. Record and export your review.

    Choose what needs follow-up. Export a CSV or backup, print a report, or save it in this browser. Your team makes the financial decision.

Start with your own data

AI INSIDE THE WORKFLOW

Model inference, then explainable checks.

Core AI capabilitySemantic charge-description matching using the pinned Xenova/all-MiniLM-L6-v2 sentence-embedding model.
Inference runtimeTransformers.js with ONNX Runtime. The model runs in the browser; suggestions are computed from the user’s entered descriptions.
Human controlA user confirms proposed matches. Similarity scores are not a guarantee that a rate applies.
Financial calculationDeterministic fixed-rate comparisons after matching. The language model does not invent amounts or decide payments.
Current scopeEntered data and CSV, fixed amounts per charge code, one currency per invoice, review decisions and exports.

PDF/OCR import, Azure cloud activation, shared team roles and ERP integrations are outside the current release. The live product does not require an Azure account. Payroll preparation is a separate companion workspace.

BUSINESS MODEL / B2B SOFTWARE

One product, used by each customer.

FreightVeyra is developing a subscription software business for shippers, freight operators and logistics finance teams. Customers operate the same application with their own invoice and agreement data. Matching, checks and report generation are software functions; the workflow does not require a FreightVeyra analyst to review each invoice.

Available today: a free self-service workspace and an optional $149 guided, 30-day product pilot. The pilot adds onboarding and evaluation support to the working software; it is a one-time offer with no automatic renewal.

Recurring commercial model: organisation subscriptions with invoice-volume allowances as commercial hosting expands. Subscription checkout and automated usage billing are not active in this MVP.

Founder support is available for onboarding and feedback. The product’s purpose is to let customer teams perform repeatable invoice reviews themselves.

View current pricing and exact pilot scope