The Development Bureau (DevB) AI Technical Circular (Works) No. 3/2026 sets new expectations for how consultants and contractors manage, check, and evidence compliance across tendering and project delivery. Pyramid helps teams meet these requirements fast — without rebuilding their entire document workflow.
Starting 1 June 2026, the DevB AI Technical Circular (Works) No. 3/2026 makes AI adoption a required capability for works delivery. It raises expectations for AI-assisted document checking, stronger governance, and more auditable compliance workflows. Pyramid helps consultants and contractors meet these requirements — automating tender and requirement checks, creating audit-ready evidence trails, and accelerating review cycles — without changing how project teams work.
The Technical Circular formalises a shift. At a practical level, it pushes project teams to adopt AI-assisted workflows, improve traceability, standardise information handling, and reduce downstream risk.
Especially tender and contract documentation — rather than relying solely on manual checking.
Teams must be able to explain what was checked, against which requirements, and what evidence supports the conclusion.
Project information must be searchable, auditable, and reusable — not trapped in PDFs and email threads.
Detect missing requirements, inconsistencies, and non-compliance earlier — before submission, approvals, or downstream disputes.
If you're delivering works in Hong Kong, the bottom line is: AI adoption is becoming part of being compliant and competitive.
Any team handling tendering, contract obligations, specifications, method statements, document control, or audit evidence on Hong Kong public works projects.
Most teams are trying to meet AI adoption requirements with fragmented PDFs, manual checklists, spreadsheet trackers, and ad-hoc ChatGPT usage — no audit trail, no enterprise control. That creates real risk.
Missed requirements buried across thousands of pages of tender, contract, and specification documents.
Inconsistent interpretation across reviewers, packages, and projects.
No defensible evidence trail when auditors, clients, or disputes demand it.
Duplicated work across bids and projects — the same checks redone every time.
Long review cycles and senior bottlenecks slowing every submission.
Ad-hoc ChatGPT usage that leaks IP, breaks governance, and can't be evidenced.
Implement mandate-ready AI with real governance — not "AI experiments."
Pyramid automatically extracts requirements from tender documents, highlights gaps and inconsistencies, and produces structured check outputs your team can review quickly. This directly supports the mandate direction toward AI-enabled document checking and improved quality control.
Instead of "the AI said so," Pyramid helps teams produce reviewable outputs. Every conclusion traces back to its source. This supports governance, defensibility, and internal QA.
Pyramid speeds up first-pass checking and triage so senior reviewers focus on exceptions. Teams comply without adding headcount.
Designed for contractors and consultants — Pyramid aligns with the artefacts teams already produce every day.
Designed to map directly to the AI Technical Circular (Works) No. 3/2026 requirements and to the real tender workflows used in HKSAR public works.
Purpose-built for document intelligence, requirements, and compliance — trained on Hong Kong's codes, standards, and practice notes.
Every answer cited. Every check traceable. Outputs that can be reviewed, verified, and used operationally — not opaque black-box answers.
Self-serve onboarding and pre-built construction agents mean teams ship a custom workflow under their own brand in under four weeks.
Funding note. Pyramid is a CITF Pre-Approved AI solution (Ref# PA26-033), which can significantly reduce adoption cost for eligible customers. Talk to us to see if your team qualifies.
A live demo on your real tender documents.
Pre-Approved · Ref# PA26-033