Safety & Security·6 min read·Updated 1 April 2026

Responsible AI Use

Guidelines for using AskBiz's AI responsibly. What the AI can and cannot do, how to interpret AI-generated advice, AI limitations, and how to disable AI learning from your sessions.

AI as a Tool, Not an Authority

AskBiz uses AI (Claude by Anthropic) to analyse your business data and generate insights, summaries, and recommendations. The AI is a powerful analytical tool — but it is not infallible, and it is not a substitute for professional judgement.

Think of AskBiz as a very capable analyst who works quickly and has read a great deal about business and finance — but who can make mistakes, miss context, and whose advice should be verified before acting on it for major decisions. This is not a limitation unique to AskBiz — it applies to all AI systems currently available.

What the AI Is Good At

AskBiz AI performs well at:

  • Summarising and synthesising large amounts of your business data quickly
  • Spotting patterns and anomalies across time series data
  • Explaining business concepts in plain English
  • Running calculations — margins, landed costs, FX impact, churn rates
  • Generating structured frameworks for decisions (export market scoring, supplier grading)
  • Drafting first versions of analysis or reports that you then review and refine
  • Answering factual business questions — tax regimes, trade regulations, common business metrics

Where AI Has Limitations

Be appropriately cautious in these situations:

High-stakes financial decisions: AI-generated analysis should inform your thinking, not replace it. For decisions involving significant capital (large stock orders, market entry, major supplier changes), use AskBiz insights as one input alongside your accountant, finance team, or business advisors.

Legal and regulatory questions: AskBiz can explain general regulatory frameworks (VAT, import duty, GDPR) but cannot give legal advice. For compliance decisions, consult a qualified solicitor or regulated professional.

Predictions and forecasts: The AI generates probability-based scenarios, not certainties. Revenue forecasts, market growth projections, and churn predictions are directional guides — not guarantees. Market conditions change in ways no AI model can fully anticipate.

Data it cannot see: The AI only knows what your connected data shows. If your Shopify data has gaps, or you have not connected your accounting software, the AI is working with an incomplete picture and will tell you so (via confidence indicators).

Very recent events: The AI's general knowledge has a training cutoff. For fast-changing regulatory or market conditions, always verify against current authoritative sources.

AI Learning and Your Data — Settings

By default, AskBiz uses anonymised, aggregated usage patterns (not your actual business data) to improve the product. You have full control over this:

To disable AI learning from your sessions:

1. Go to Account Settings → Privacy → AI & Learning

2. Toggle off Allow usage data for product improvement

3. This disables PostHog session analytics and any aggregated pattern collection from your account

4. Your business data, conversations, and financial figures are never used for training regardless of this setting — see our AI Training Policy

What this setting controls:

  • On: AskBiz can see anonymised interaction patterns (e.g. which features you use, question types — not the actual content)
  • Off: No interaction data is collected from your account beyond what is required for the service to function (authentication, billing, error logging)

What this setting does NOT affect:

  • Your business data is never used for AI training in either setting
  • Your conversation content is never used for AI training in either setting
  • This setting does not affect the quality of answers you receive

Interpreting AI Confidence Indicators

Every AI answer in AskBiz carries a confidence indicator. Use these to calibrate how much weight to give the answer:

  • 🟢 High confidence — based on complete, recent, directly connected data. Generally reliable for decision-making.
  • 🟡 Medium confidence — based on partial data or data older than 30 days. Use directionally but verify key figures.
  • 🔴 Low confidence — limited data available. Treat as a starting point for investigation, not a conclusion.
  • 〜 Estimate — no directly relevant data. The AI is using sector benchmarks or inference. Explicitly note this if sharing with others.

Never present Low confidence or Estimate outputs as definitive figures to investors, lenders, or partners without clearly labelling them as estimates.

Using AI-Generated Outputs Responsibly With Third Parties

If you share AskBiz outputs with external parties (investors, banks, suppliers, partners):

  • Label AI-generated outputs clearly — note that figures were generated by AI analysis and should be verified
  • Cross-reference key figures — before sharing revenue or margin data externally, verify it matches your accounting software
  • Do not present AI scenarios as forecasts — a scenario analysis showing what happens if FX moves 10% is not a financial forecast. Label it correctly.
  • UK/EU legal context: Under the UK Consumer Rights Act and EU AI Act (applicable to high-risk AI uses), AI-generated outputs used in commercial contracts or financial decisions should be clearly identified as AI-assisted. AskBiz is not currently classified as a high-risk AI system under the EU AI Act, but responsible disclosure is still good practice.
  • US context: The FTC expects truthful and non-deceptive claims — presenting AI-generated projections as independently verified data in investor materials could raise regulatory concerns

AI Content Boundaries

AskBiz AI is designed for business intelligence and will decline requests that fall outside this scope. Specifically, it will not:

  • Generate content designed to deceive customers, regulators, or investors
  • Provide advice on illegal tax avoidance or fraudulent accounting
  • Generate fake reviews, false testimonials, or misleading marketing claims
  • Assist with activities that violate the terms of service of connected platforms (Shopify, Amazon, etc.)
  • Generate harmful, discriminatory, or harassing content

If the AI declines a request, it will explain why. If you believe a refusal is an error, contact support@askbiz.co with the context.

Frequently Asked Questions

Was this article helpful?

Still stuck? Email our support team.