Artificial intelligence (AI) for reporting and analytics: ask your data, explain it, and draft the pack

Natural-language querying, narrative summaries and report drafting all run in production today, provided the AI is given a curated map of what the data means. Accuracy is a context and verification problem, not a model-power problem, so the groundwork is data definitions rather than a cleverer model.

• The best-documented finding in this field is that accuracy comes from context and verification, not from model power. In Anthropic's own deployment, agents with raw data access scored no better than around 21 per cent on internal evaluations; adding structured skills that route the model to trusted definitions lifted accuracy consistently above 95 per cent in aggregate, and to around 99 per cent in some domains. That is the difference between a toy and a tool, and it is entirely buildable now.[1]
• The honest counterweight is the gap between demos and enterprise reality. On the older, tidy Spider 1.0 benchmark, text-to-SQL looks near-solved, above 90 per cent accuracy on small databases. On Spider 2.0, which uses massive real schemas averaging hundreds of columns, multiple SQL dialects and external business logic, the leading models succeed only about 21 per cent of the time. Production systems close that gap with heavy context engineering: Uber's internal QueryGPT needed a prompt enhancer and curated example sets before it was trustworthy, after which query time fell from about 10 minutes to about 3. Pilot on your own schema before you trust anything.[1][2]
• On adoption, a Gartner survey of 403 analytics and AI leaders found over 50 per cent already using AI for automated insights and natural-language queries, and Gartner predicts 75 per cent of new analytics content will be contextualised through generative AI by 2027. Read those figures as direction of travel, not a guarantee for your firm: analyst projections and headline deployment numbers come from other organisations and from parties with something to sell, so measure the gain on your own reports before you bank it.

Keep a person accountable for any number that reaches a client, the board or HMRC. In the UK the governing rules are the UK GDPR, the Data Protection Act 2018 and ICO guidance rather than an AI statute, and connecting AI to financial and customer data still requires access controls and an audit trail.

• Confident wrong answers are the core risk: a language model will happily return a plausible figure from the wrong column, or with a silent join error, and you cannot tell a right answer from a wrong one just by looking. Definitions, tests and spot-checks matter more than the model, and rigour has a price: in Anthropic's deployment, adding adversarial self-review improved accuracy by about 6 per cent but cost roughly 32 per cent more tokens and 72 per cent more latency. The practical rule is to let the AI draft, explain and accelerate, but have a named person own and approve the numbers that go to clients, the board or the tax authority.[1]
• Connecting AI to customer and financial data is governed in the UK by the UK GDPR and the Data Protection Act 2018, enforced by the ICO, whose guidance on AI and data protection covers fairness, transparency, lawfulness, accountability and when a data protection impact assessment is required. You need a lawful basis, access controls and an audit trail before an AI reads personal data, and the higher maximum penalty is 17.5 million pounds or 4 per cent of total worldwide annual turnover.[1]
• If analytics output feeds automated decisions about individuals, staff scoring, customer credit terms or pricing tied to a person, a specific regime applies. Since 5 February 2026, new UK GDPR Articles 22A to 22D, inserted by the Data (Use and Access) Act 2025, generally permit solely automated significant decisions, but only with safeguards: the person must be informed, able to make representations, able to obtain meaningful human intervention and able to contest the decision. Meaningful involvement has to be genuine, not a rubber stamp. Numbers that flow into tax reporting carry their own obligations too, so what is filed with HMRC should always pass through human review and approval.[1]