Skip to content
Demiton

Uploading Your Data

Insights tier

On Insights, you don’t connect live systems - you upload historical data. Demiton applies the open-source procurement classification framework to your uploads, structures the result against domain entities, and makes it queryable through Claude via the MCP server.

This is the fastest path to value: no IT involvement, no credentials, no implementation scoping. Upload a file and start asking questions.

What to upload

Section titled “What to upload”

Insights is designed for historical contract and estimating data. The most valuable uploads are:

Data type Format What you’ll get
Historical contract rates Excel, CSV Rate comparison across projects, estimator performance, where rates held vs actuals
Tender submissions and BOQs Excel, CSV Classification against the procurement framework, rate benchmarking
Project cost summaries Excel, CSV Budget vs actual patterns, cost code breakdown
Estimating workbooks Excel Structured rate extraction, classification, query via Claude

The platform handles messy, real-world spreadsheets - headers in row 3, merged cells, multiple worksheets. You don’t need to clean your data before uploading.

How to upload

Section titled “How to upload”

  1. Sign in and navigate to Systems

    Section titled “Sign in and navigate to Systems”

    Go to app.demiton.io and sign in with your Insights account.

    Open Systems from the sidebar and click Add System (or use ⌘K → “add system”).

  2. Select Excel / CSV and upload your file

    Section titled “Select Excel / CSV and upload your file”

    Choose Excel or CSV from the connector list. Upload your file - supported formats: .xlsx, .xls, .csv.

    Give the upload a name - for example, “Pacific Highway BOQ 2024” or “FY24 Contract Summary”.

  3. Review classification

    Section titled “Review classification”

    Demiton processes the file and applies the procurement classification framework. A classification summary shows:

    • How many line items were classified
    • Which cost categories were identified
    • Any rows that couldn’t be classified (you can review and correct these)

    Classification typically completes within a few minutes for most workbooks.

  4. Connect the MCP server and start querying

    Section titled “Connect the MCP server and start querying”

    Once your data is classified, it is available through the MCP server. Connect Claude or ChatGPT and start asking questions:

    • “Which cost codes had the largest variance between estimate and actuals across my uploaded projects?”
    • “What were the earthworks rates on the Pacific Highway contract?”
    • “Which estimator’s rates held best against actuals?”
    • “Summarise the labour split across my uploaded BOQs.”

What the classification framework does

Section titled “What the classification framework does”

Demiton applies an open-source procurement classification framework to every upload. This maps unstructured line items - cost codes, descriptions, quantities, rates - to a consistent taxonomy across Australian civil construction categories.

The result is a structured view of your historical rates that can be:

  • Queried across projects (“what did earthworks cost per m³ across all uploads?”)
  • Compared against your own benchmarks
  • Asked about directly in natural language through Claude

The framework source is published at github.com/demitonapp.

Upgrading to Connected

Section titled “Upgrading to Connected”

Insights answers the question: “what do we know from our historical records?”

Connected answers: “what is happening right now, and what happened last week on site?”

When you find yourself wanting to ask questions about live project financials, current workforce allocations, or last night’s site diaries, that’s the signal to move to Connected. Contact us - the first conversation is about scoping which 1 to 2 systems to integrate.

Next steps

Section titled “Next steps”