How Outside Ideas is set to cut tender scope reading by 70%

A Claude plugin built in five weeks now gives every Outside Ideas estimator a cited first pass on each tender, freeing senior time to price more work.

At full deployment
- 70% fewer manual scope-reading hours per tender
- $252k senior capacity reclaimed each year
- 8–10 additional tenders a year
- 3–4 months to pay back from time savings alone

The problem in numbers

Outside Ideas has delivered civil construction projects across South Australia for 22 years. Like most contractors, it wins work by pricing tenders, and pricing starts with reading. For every tender, a senior estimator spent two to three days working through specifications, drawings, schedules of quantities and addenda to work out what was actually in scope.

At around 120 tenders a year, that is roughly 2,400 hours, or about $360,000 of senior estimator time. It was spent on reading rather than pricing, and it capped how many bids the business could take on.

What we built

After an eight-day discovery with one of Outside Ideas' senior estimators, Propel narrowed nine AI opportunities to the two with the highest impact. We built both over five weeks, at a fixed price, as one installable Claude plugin.

- Tool 1, the scope assistant, reads the full tender document set from SharePoint and produces a structured scope register. Every line cites the document, drawing or sheet it came from, and quantities come straight from the schedule of quantities or existing Bluebeam markups.
- Tool 2, the quote request generator, turns that register into quote requests for suppliers and subcontractors. It matches each category against Outside Ideas' own contact list, so the estimator chooses who receives each request and sends it.

The plugin connects to Outside Ideas' own Microsoft 365 and Bluebeam Max through MCP connectors, so the tool works on tender documents where they already live.

Why estimators trust it

An AI tool in estimating is only useful if the estimator can rely on it. The plugin runs on four rules:

1. The AI drafts and the estimator decides. Nothing leaves the business without review.
2. Every item cites its source. Any line can be checked in seconds.
3. Ambiguity is flagged, never guessed. Conflicting sources are named, not quietly resolved.
4. Quantities are never inferred. If no measured source exists, the field is left blank.

Built in close collaboration with that senior estimator as the subject-matter expert, the tool reflects how Outside Ideas actually estimates, not a generic idea of it. It drops straight into the team's existing workflow, and Outside Ideas owns it outright.

Where it stands

The tool is rolled out to every estimator at Outside Ideas, across its South Australian and Victorian offices. At Outside Ideas' 30% win rate and $1.2 million average project value, the extra tenders it makes room for represent more than $3 million in potential revenue a year.

If your experts are spending their days reading instead of deciding, talk to us at propelventures.ai/contact.