TrioClick Intelligence AI Agent

AI that understands
your project as a
connected structure —
not just a search.

So it can reason across your entire project — not just search within it.

It knows that a drawing belongs to a package, a package belongs to a discipline, a discipline sits within a stage, and that stage connects to the tender it feeds. Ask it anything — it traces the connections to give you a complete answer, with sources.

Get Early Access See what it does →
TrioClick · AI Agent
What is the status of the Tower 1 façade package?
Agent Response
5 drawings — 3 at Rev C, 2 at Rev B, all issued.
2 open queries, oldest 8 days unanswered.
1 of 3 bidders submitted. Bid close in 6 days.
GraphRAG Neo4j Claude Sonnet Project Files
What the agent does

Three things.
All grounded in your data.

The agent doesn't generate from thin air. Everything it extracts, summarises, or drafts comes from your project — the graph, your documents, and your uploaded files.

Extract
Ask any question about your project in natural language. The agent traverses the project graph — connecting documents, packages, queries, bidders, and stages — to give you a sourced, complete answer. Not a search result.
Example
"How many open RFIs are on the Substructure package, and who raised them?"
Summarise
The agent reads across your project data and generates structured summaries — package status, bid progress, query logs, document transmittal histories. Things that currently take an hour to compile, done in seconds.
Example
"Summarise the current status of all tender packages and flag anything overdue."
Report
The agent drafts reports directly from live project data — tender evaluation summaries, package progress reports, bid comparison narratives. You review and send. Same output, a fraction of the time.
Example
"Draft a tender evaluation summary for the STR-001 Substructure package."
Architecture

How it works.

Not a generic chatbot connected to a folder. A purpose-built intelligence layer that models your project as a graph — and reasons across it.

01
Project Graph — Neo4j
Your project is modelled as a connected graph in Neo4j — documents, packages, disciplines, stages, queries, bidders, and relationships between them. The agent doesn't find files; it traverses a structured model of your project.
Neo4j · GraphRAG
02
Your Project Files
Drawings, specifications, BOQs, correspondence, and any documents uploaded into your project workspace are indexed and available to the agent — not just filenames, but content.
Project Files · Uploaded Docs
03
Retrieval-Augmented Generation
When you ask a question, the agent retrieves the relevant nodes, relationships, and document content from your project — then passes that context to the language model to generate a grounded, sourced response.
RAG · Grounded responses
04
Claude Sonnet — the LLM
The language model that generates the response is Claude Sonnet by Anthropic — one of the most capable and safety-focused models available. Your data is passed only for the duration of the query and is never used to retrain the model.
Claude Sonnet · Anthropic
Consent & Data Control
Disabled by default. AI must be explicitly enabled by a Project Administrator before any data is processed.
Per-project consent. Enabling AI on one project does not affect others. Each project is independently controlled.
In-app notice. All project users are notified when AI is activated and what data categories may be processed.
Revocable any time. Disable AI in project settings. No further data is transmitted to the language model once disabled.
Never used to retrain. Your project data is never stored by or used to improve the underlying language model.
Language Model
Claude Sonnet · Anthropic
Enterprise-grade, safety-focused, SOC 2 compliant. Your data is never stored or used for model training.
Honest about scope

What it doesn't do.

We'd rather tell you what the agent can't do than have you find out mid-project. These are deliberate boundaries, not gaps.

It doesn't make decisions for you. The agent extracts, summarises, and drafts. Every output is for your review. Professional judgement stays with your team.
It doesn't know what's outside your project. The agent reasons from your project data only — the graph, your documents, and your uploaded files. It has no access to the internet or external data.
It doesn't proactively surface risks. It answers questions you ask and generates outputs you request. It does not monitor your project in the background or alert you unprompted.
It doesn't replace professional review. Tender evaluation summaries, reports, and drafts are starting points — not final outputs. Your commercial and legal review obligations remain.
Data & security

Questions we get
from IT and procurement.

Is our project data sent to Anthropic or stored externally?
Your project data is passed to Claude Sonnet transiently — only for the duration of the query — and is never stored by Anthropic or used to train the model. TrioClick's Data Processing Agreement with Anthropic enforces this. Full details are in our Privacy Policy.
Can we control which projects have AI enabled?
Yes. AI is disabled by default on every project. A Project Administrator must explicitly enable it, and all project users receive an in-app notice before any data is processed. You can disable AI on any project at any time from project settings.
Does the agent have access to data from other projects in our organisation?
No. The agent is scoped strictly to the project it is enabled on. It has no access to other projects in your organisation, other organisations' data, or any external data source. Project-level isolation is enforced at the data layer.
What data does the agent actually process?
The agent reads from the project graph (structured metadata — packages, documents, queries, bidders, relationships), the content of project files (drawings, specifications, BOQs, correspondence), and any documents you've uploaded directly. Depending on your configuration, TrioClick can anonymise project identifiers before transmitting to the language model.
Is this compliant with our data residency requirements?
TrioClick's primary hosting operates in regional data centres in the UAE, KSA, and Australia. AI query processing involves a transient cross-border transfer to Anthropic's API. This is disclosed at the point of AI consent and covered by Standard Contractual Clauses where required. Contact us to discuss specific residency requirements for your organisation.
Early Access

Ask your project
anything.

The AI agent is available now in the Documents and Tender modules. Get early access and start getting answers that used to take an hour.

Get Early Access

AI is enabled per-project by the Project Administrator. See our Privacy Policy for full data handling details.