Skip to content
BuildGuider

THE MECHANISM

Drawings, notes, invoices in. Reviewable outputs out.

The same project evidence layer powers every agent. The same human-review step approves every output. Here's how the work actually flows.

THE DECISION CHAIN

The eight questions every job answers.

What does the drawing show?

The Estimator Agent reads rooms, viewports, components, and the existing-vs-proposed split.

Approved by your estimator.

What has the client or builder said?

Voice notes, WhatsApp messages, and emails are captured against the right context.

Approved by your project lead.

What work is implied?

Scope items are atomised, linked to source, and surfaced with assumptions and confidence.

Approved by your estimator.

What is missing or unclear?

Open questions are flagged, not silently guessed.

Approved by your estimator.

What needs measuring or pricing?

Quantities are checked against scope; the QS Agent compares revisions and surfaces cost movement.

Approved by your QS.

What supplier or cost history can help?

The Procurement Manager Agent reads every invoice you've processed and suggests the right supplier and the right number.

Approved by your procurement lead.

What decision or change needs recording?

The Contract Manager Agent drafts variations and keeps decision logs timestamped and traceable.

Approved by your contracts lead.

What should be reviewed before it ships?

Every output above passes through human approval before it becomes part of the quote, programme, purchase, or contract record.

Approved by your team.

INPUTS · EVIDENCE · OUTPUTS

One evidence layer. Seven agents.

Inputs

drawings (PDF), voice notes (mobile), invoices (supplier PDFs), decisions (chat / email / meeting notes), past projects.

Project evidence layer

rooms, viewports, components, revisions, transcripts, line items, decisions — all timestamped and cross-referenced.

Outputs

scopes, variations, buying lists, task lists, decision logs, change logs, margin alerts. Each labelled with the agent that produces it.

HUMANS IN CONTROL

Humans in control. Always.

The evidence layer is the read side. Reading is automatic.

The output side is reviewed. Every scope item, variation, buying suggestion, task, or change log starts as a draft. Your team approves before it ships to a quote, a programme, a purchase, or a contract.

Your team approves before anything ships.

See the audit trail in action

WHAT IT DOESN'T DO

What it doesn't do.

It does not replace estimators, QSs, site managers, or contract managers.

It does not guarantee accuracy. Every output has a confidence indicator and open questions surfaced inline.

It does not "automate construction." It reduces repetitive review and admin work so senior judgement gets to where it matters.

COMMON QUESTIONS

Common questions.

We tried estimating software and no one used it.

Most tools demand a workflow change. BuildGuider works from the files you already have. Your estimator opens their normal drawing pack; we add to it. We'd want to start on one live project together, not a system migration.

AI is going to be wrong on construction-specific stuff.

Often. That's why every output has an evidence link back to source, a confidence indicator, and an open-questions section. Your estimator approves before anything ships. The point isn't replacing their judgement — it's getting them to the point of judgement faster.

My estimator will hate this.

Counter-intuitively, they usually don't, once they've seen it. The thing they hate is repetitive copying from drawings into spreadsheets. We take that off their plate. The work they love — judgement, pricing, supplier conversations — stays theirs.

What about my drawing pack — does it need a specific format?

No. Send us what your architect sends you. PDFs at any quality the eye can read are fine. If the drawing pack is a mess of half-revised sheets, we'll show you what we can extract before any commercial conversation.

What if you go out of business?

Fair. All your data is exportable. Your project records are yours. We're not a lock-in play.

How much does it cost?

The first conversation is a pilot — thirty minutes, with a real drawing pack from one of your live projects. Pricing comes after we both know it'll work for your business.

START SMALL

See it on your drawings.

Send us a real drawing pack from a real project. We'll show you what BuildGuider reads, what it flags, and what it produces — before any commercial conversation. Thirty minutes.

Start the pilot →