Vectruva
AI Practices

Responsible AI

Last updated 29 April 2026

AI is central to how Vectruva works — and we believe transparency about how we use it is essential. This page explains our approach to AI-augmented advisory, including how we manage its limitations.

1. How We Use AI

Vectruva uses large language models (LLMs) and other AI systems to assist with analysis, document generation, hypothesis development, quality review, and advisory responses. AI is used as a force-multiplier for our consulting methodology — not as a replacement for human judgement.

Our AI agent system is built around five categories of specialist agents: discovery (data profiling, quality audit), diagnosis (hypothesis generation, analysis, independent review), design (deliverable authoring, financial modelling), quality assurance (internal review), and team coordination (engagement management). Each agent operates within a defined scope and follows our brand and methodology guidelines.

2. Human-in-the-Loop

Every AI-generated deliverable passes through at least one human review step before it reaches a client. Our delivery process includes seven explicit client checkpoints (CP1–CP7) at which the client can review, comment on, or reject AI-generated outputs. No finding is treated as final without client acknowledgement.

Where AI outputs carry uncertainty, our system classifies confidence as High, Medium, Low, or Unknown, and flags items that require human escalation before proceeding. Outputs classified as Low or Unknown are always reviewed by a Vectruva Principal before being shared with clients.

3. Hallucination and Error Risk

Large language models can produce outputs that are plausible but incorrect — a phenomenon commonly called "hallucination." We take this risk seriously and address it through:

All Deliverables are advisory in nature. They are not legal, financial, tax, audit or engineering advice. You remain responsible for decisions made based on them.

4. Training Data

We do not use client data to train AI models — ours or our providers'. Our agreements with AI processing providers prohibit the use of client inputs for model training purposes. Client data is used solely to perform the services described in your engagement scope.

5. AI Provider Transparency

We use AI provided by third-party technology companies. We select providers based on their security posture, data-handling commitments, and suitability for enterprise use. A list of our current AI processing sub-processors is available on request — email support@vectruva.com.

6. Bias and Fairness

AI models trained on general data can reflect societal biases. Our methodology is designed to ground conclusions in client-specific data, quantitative analysis, and established business benchmarks — reducing but not eliminating the risk of biased outputs. We welcome feedback from clients if they observe outputs that appear unfair, inconsistent, or contextually inappropriate.

7. Continuous Improvement

We review our AI methodology regularly and update our skill library, prompt design, and quality controls as the technology evolves. Material changes to how AI is used in your engagement will be communicated before they take effect.

8. Contact

Questions about our AI practices should be sent to support@vectruva.com.