AEC project information management doesn’t fail because teams don’t care, it fails because requirements and decisions fragment across emails, meeting minutes, PDFs, chat and spreadsheets. Teams spend days reconstructing what was agreed, and drift quietly becomes rework, risk and programme pressure. Most tools manage one strand well: tasks, a CDE, meeting notes, a risk log. The failure mode is between the strands, where decisions do not reliably tie back to what was agreed and teams end up maintaining parallel trackers.
vBrief answers the failure mode between the strands. It is how architects and project teams capture, structure and keep their project information rigorous over the life of a project, quickly and while staying fully in control of it: emails, documents, notes and chats become a verified project record your team can stand behind, every item attributed, one click from its original source, findable in seconds, and kept current as the project evolves. vBrief is fast and rigorous at once, and the differentiator is that it holds both together rather than either alone: fast from day one, including on simple projects, and still fast in year two, because the rigour is what stops the record decaying.
How it works. Capture costs seconds of your time: connect Outlook, which imports emails and attachments straight into a project, or upload PDF, email, DOCX, TXT or pasted text, then move on. The reading happens without you: vBrief surfaces the requirements, risks, suggestions and asks inside each source and pre-tags them, original context attached. You review each item beside that source, the exact quote highlighted, and confirm or correct it in a click or two; if you spot something vBrief missed, you add it yourself by highlighting the text in the source. Nothing enters your trusted record until you have confirmed it. After you submit, vBrief checks each item against your existing project data in the background and surfaces anything that looks like a potential clash or duplicate, for you to resolve when it suits.
From that verified record you build the views you need. vBrief ships three view types, Sheet, Document and Timeline; adjust the filters, sorts and fields of any of them and save the result as your own custom view. A live brief, a risk register, a suggestion log, open queries, change logs, task lists, stage reports and coordination views are all examples of custom views built from the same record, not separate products. They read live from that record, so confirming a change once updates every linked view, and an export is a snapshot of its view at the moment you take it. Any view can also be rendered as a notebook, readable prose, on demand. Export to Excel (XLSX), CSV or JSON from the sheet view, and PDF, MD or TXT from the notebook view.
What makes it different. Two properties, stated as vBrief’s own without ranking any other tool. First, the record is user-structured rather than AI-structured: you decide what is captured and how it is organised, by your own zones and a fixed category list, so the record is curated by professional judgement rather than assembled by a model. Second, each item is reviewed against its original source before it is trusted. That is source-to-output review, the review direction capable of surfacing what reading an output alone would miss. No tool can promise nothing is missed: the honest goal is making omissions catchable at the point of import.
The mechanism behind both is structural. The AI is separated from the record and does not commit to it; a person confirms each item against its original source before it is trusted. The review is not a verification chore: it is minimised and timed to decisions you were already making, since reading an email and deciding what matters is the review, and there is an honest manual bypass when you are faster than the tool.
The audit trail. Every human-verified action (who, what, when, against which source) is preserved in an append-only project log, exportable on demand. It gives unbroken attribution from source document to confirmed decision, supports the traceability and evidence standards PI insurers increasingly expect for AI-assisted work, supports the traceable information your golden thread obligations under the Building Safety Act rely on (without discharging them), and helps architects evidence ARB Code of Conduct Standards 3 and 4, competence and accurate records of work and decisions. RICS now expects written justification where AI materially affects the delivery of client work in surveying practice, the decision and reasoning recorded with a named professional accepting responsibility for reliance on outputs, and an attributed append-only log supports exactly that kind of record. vBrief is not a compliance tool or a compliance guarantor: it does not provide the statutory golden thread or discharge Building Safety Act duties, which stay with the dutyholder. PI outcomes depend on your own policy and insurer, so confirm the position with your broker.
Security and data. All data (application, storage, AI processing and backups) is hosted and processed in the EEA on Microsoft Azure, with encrypted geo-redundant backups staying within the EEA. AES-256 at rest; TLS in transit, TLS 1.3 preferred and TLS 1.2 minimum. SSO with MFA supported via Microsoft Entra ID. Cross-organisation isolation is enforced at the database through row-level security, alongside application-level controls, with automated cross-tenant isolation tests run on every build; information is organised into project spaces inside your tenant, and the tenant remains the access boundary. Your content is processed solely to generate outputs within vBrief on EEA-hosted enterprise models: it is never sent to consumer AI services or third-party model providers, and it is not used to train or fine-tune any AI model without your explicit separate written opt-in. vBrief is browser-based, with no desktop installs and no endpoint agents; the optional Outlook add-in is one-way and admin-enabled, with no standing mailbox access. Breach notification within 72 hours of becoming aware of a personal data breach affecting your data is a commitment in our standard terms, as is a 99.5% availability target during business hours with downtime credits.
Certifications, honest status. vBrief is built entirely on managed Azure services that hold ISO 27001 and SOC 2, Microsoft’s certifications of the infrastructure, which vBrief inherits and builds upon. vBrief’s own ISO 27001 and Cyber Essentials certifications are on the roadmap, and it holds no organisation-level certification of its own yet. A documented UK and EU GDPR programme is in place, and the full DPIA risk assessment is in progress with initial screening complete. An EU AI Act readiness assessment and internal classification assessment are in progress; vBrief claims alignment work, not a completed formal assessment. An internal alignment assessment against ISO/IEC 42001, the AI management-system standard, is also in progress. On ISO 19650, vBrief is not a CDE and makes no conformance claim, which stays with your delivery team; the standard is revising to better encompass the granular information management vBrief has been built on since inception. vBrief maintains professional indemnity, cyber liability and public liability insurance throughout the term of every customer agreement. Each item is stated with its honest status rather than implying certifications vBrief does not hold.
Where it sits. vBrief operates at the in-house information-management layer above your BIM, CDEs and project-management tools, from the start of a project: alongside and underneath your stack, not in place of it, closing the knowledge gap between when instructions come in and information reaches your external coordination spaces. The boundaries are stated as plainly as the capabilities. vBrief is not an autonomous AI agent: the AI surfaces and flags, and humans decide. It is not RAG or an AI-automated knowledge base: the record is user-structured and human-verified, not retrieved-and-generated. It is not a general project management tool, and it does not replace BIM, CDEs or design tools. Structured export is supported today; there is no public API, and integrations with leading CDE and BIM workflows are prioritised by demand.
vBrief Ltd was founded in London in 2024, conceived by an architect (James Wright-Zhang, ARB- and RIBA-registered, Co-Founder & CEO) and built with an engineer (Libo Shen, Co-Founder & CTO, six years at Google and two as Head of Engineering at P.ai). It is an early-stage company working closely with its initial practices on live projects. Those pilots have shaped the product but have not independently proven it, and vBrief publishes no efficacy statistics or independent evaluations. The Security Overview is a public download; vBrief responds to security questionnaires and can share its sub-processor list, contract pack and Client Contract Overview on request, with more detailed security information available under NDA.
