← All projects

Client project / Internal platform / 2026

Invoices, people and documents.
One workspace.

A web application connecting financial records, team workflows, communication and AI. From an email attachment to a paid invoice or a follow-up task.

Executive summary

Challenge
Financial records, documents and follow-up work needed to remain connected.
Approach
Design a shared workspace around everyday operations and role-based access.
Solution
Invoice extraction, payment tracking, personnel records, approved meeting tasks and document search.
Outcome
A working platform used by 14 people, with source documents and human review kept in the workflow.

How it works

From an email attachment to payment tracking.

01 / 06

An invoice arrives in a connected mailbox.

The application reads the attachment. AI identifies the financial document and prepares it for processing.

Data transfer

EmailDocument

Processing starts in the connected mailbox.

02 / 06

No need to retype the details.

AI extracts the supplier or customer, amount, currency, invoice number, payment reference and due date.

Extracted information

Supplier / customerCounterparty identified
Amount & currencyFinancial details
Invoice & payment referenceDocument identification
Due dateDate

03 / 06

Rules determine where the document belongs.

The mailbox and processing rules determine whether it is recorded as income, an operating expense or a contractor invoice.

Document record

  • Values saved in the correct fields
  • Classified by mailbox and rules
  • Included in financial records

04 / 06

The original attachment stays available.

The source document is attached to the record so users can compare the extracted information and correct it if necessary.

Record & source document

AttachmentLinked to the record
DataAvailable for review
CorrectionsAvailable to the user

Extracted details can be checked against the original

05 / 06

An unambiguous match marks the invoice as paid.

Once connected to the bank, the application compares transactions with invoices. Ambiguous cases remain for manual review.

With a bank connection

Clear match→ Record payment
Ambiguous match→ Manual review

Automatic matching only happens when the match is unambiguous.

06 / 06

An unpaid invoice can become a follow-up task.

Unpaid documents can trigger alerts and tasks for the responsible person. Payment status remains visible in the financial overview.

Follow-up workflow

  • Payment status recorded
  • Alert the responsible person
  • Create a follow-up task

An invoice becomes part of a workflow, not just a stored file.

Simplified illustration · no client dataAutomatic walkthrough · select any step

Solution

One workspace for day-to-day operations.

I built a web application that connects finance, documents, team tasks, communication and AI. It reduces repeated data entry and links the work from receiving an invoice through recording it to tracking payment.

The platform is used by 14 people. It works on desktop and mobile, with access to features and data controlled by roles and permissions.

The invoice arrives by email. Its data moves into the records.

The application reads attachments from connected mailboxes. AI identifies financial documents and extracts the supplier or customer, amount, currency, invoice number, payment reference and due date.

Mailbox-specific rules classify each document as income, an operating expense or a contractor invoice. The extracted values populate the corresponding fields and financial records.

The original attachment stays linked to the record, allowing users to check and correct extracted data.

The financial overview includes payment status.

Income, operating costs and personnel expenses feed one financial overview. The application handles multiple currencies, conversions and the payment status of each document.

With a bank connection, an unambiguous transaction match records an invoice as paid. Uncertain cases remain for manual review.

Unpaid invoices can lead to alerts and tasks for the people responsible, turning a record into a concrete next action.

Personnel records support monthly payments.

The application manages employees and contractors, compensation, bonuses, invoices and related documents. It uses these records to prepare monthly payment summaries and close completed periods.

Personnel costs also feed into the overall financial overview.

Meeting conclusions can become assigned tasks.

The team organises work on boards, assigns owners and tracks deadlines and completion. Tasks can be created manually or follow from other records, such as invoices.

The application also uses meeting transcripts. AI prepares summaries and suggested tasks; a suggestion becomes a task only after a person approves it.

  1. 01

    Meeting transcript

    Source material for the next steps.

  2. 02

    Summary & task suggestion

    AI prepares a suggestion for review.

  3. 03

    Approval

    A person decides whether to create the task.

  4. 04

    Task in the team's workflow

    The approved suggestion becomes an actual task.

A personal overview brings tasks, email and calendar together.

The home screen shows each user's tasks, emails and upcoming meetings. People can see what needs their attention without continually switching tools.

Documents and knowledge stay connected to the work.

Documents can be organised in folders, linked to records and shared according to permissions. AI search helps users find relevant information across available documents and meetings.

An AI assistant helps summarise, search and prepare text or suggested next steps. It works with information available to the user. Suggestions can be reviewed and followed up using the application's regular tools.

Outcome

Invoice data supports financial records and payment tracking. Personnel costs feed financial reporting, meetings lead to approved tasks, and documents stay linked to the relevant records. These activities now share one workspace.

Does this sound familiar?

Let's look at your process.

Contact Martin More projects ↓

Visitor analytics

I optionally use Tawk.to to understand visits to this website. The chat stays hidden. This service only loads with your consent.

This browser will remember your choice. Details