How to use AI to explain a confusing letter or bill

A letter arrives from your energy provider. It mentions "deemed contract rates," references an Ofgem price cap adjustment, and quotes figures that do not match anything on your last statement. You read it twice, put it on the kitchen counter, and tell yourself you will deal with it later. Later never comes, or it comes three weeks past a deadline you did not realise was buried in the second paragraph.

This is not a reading comprehension problem. Official correspondence, from utility bills to council tax notices to medical explanations of benefits, is written in language designed to satisfy legal and regulatory requirements. The people drafting it are not trying to confuse you, but clarity for the recipient is rarely their first concern. The result is a stack of paper and email that feels vaguely important and specifically impenetrable.

Fabric is built to help with exactly this kind of friction. You can photograph a paper letter with your phone using quick capture, scan a handwritten note with the handwriting scanner, forward an email bill from your connected Gmail, or drag a PDF into your cloud drive. Once the document is in Fabric, you open the AI assistant and ask a question in ordinary language: "What does this letter mean?" or "Am I being overcharged?" or "What do I need to do by when?"

The assistant reads the full text of the document and responds in plain, conversational language. It pulls out the key facts: what is being asked of you, what the financial implications are, what deadlines exist, and what your options might be. If the letter references terms or acronyms you have never encountered, it defines them in context rather than sending you to a glossary.

Getting the document into Fabric

The first step is always the same: get the document out of the physical world or your inbox and into a place where AI can read it. Fabric offers several paths depending on where the document lives.

For paper letters, the fastest route is quick capture. Open Fabric on your phone, photograph the letter, and it lands in your cloud drive with the text extracted and searchable. If the letter includes handwritten annotations, perhaps a note scrawled by someone at a benefits office or a doctor's margin comment, the handwriting scanner will pick those up too. You do not need to flatten the paper perfectly or worry about shadows. The scanner handles the imperfections of real kitchen-table photography.

For digital documents, you can upload PDFs, Word files, or images directly to your Fabric cloud drive. If your bills arrive by email, connecting your Gmail account lets Fabric pull in messages and attachments automatically. That energy bill you received last Tuesday is already waiting.

Once the document is stored, it becomes part of your searchable, AI-readable collection. This matters more than it might seem at first, because Fabric does not just read documents in isolation.

Asking the right questions

With your document in Fabric, you open the AI assistant and start a conversation. There is no special syntax required and no need to structure your query like a search engine keyword string. You write the way you would ask a knowledgeable friend.

Some starting points that tend to work well: "Summarise this letter in two paragraphs" gives you the headline version. "What action do I need to take?" cuts to the practical core. "Explain the charges on this bill line by line" is useful for utility statements or medical bills where the numbers seem disconnected from reality. "Is this amount correct based on what I was paying before?" brings in comparison, which is where Fabric's broader capabilities start to shine.

The assistant draws on multiple AI models, including those from OpenAI, Anthropic, and Google, selecting the best approach for your question. You do not need to choose a model or understand the differences between them. The multi-model architecture means Fabric can handle both the precise numerical reasoning a bill comparison requires and the nuanced language interpretation a legal notice demands.

One important note: the AI assistant provides explanations and context, not legal or financial advice. It can tell you what a clause means, flag a charge that looks unusual, or summarise your obligations under a notice. It cannot tell you whether to pay, dispute, or ignore something. That decision remains yours, ideally informed by the clarity the assistant provides.

Where memory and context make the difference

Reading a single letter in isolation is useful. Reading it against the background of every related document you have ever received is transformative. This is where Fabric's memory engine changes the nature of the task.

When you ask "Am I being overcharged on this electricity bill?", the assistant does not just parse the current document. If your previous bills, your original contract, your correspondence with the provider, and the welcome letter outlining your tariff are all in Fabric, the memory engine connects those related documents and draws on them to give you a grounded answer. It might note that your unit rate has increased by 14% since your fixed deal ended, or that the standing charge on this bill does not match the one quoted in your renewal letter from six months ago.

This kind of cross-document reasoning is something no single-document AI tool can replicate. It requires persistent context, a record of what has come before, and the ability to surface relevant history without being told where to look. Fabric's semantic search does this across your entire collection, finding connections based on meaning rather than exact keyword matches.

The practical effect is that the assistant gets more useful the more you use it. A first bill upload gives you a plain-language explanation. A year of bills, contracts, and correspondence gives you a financial memory that spots patterns and anomalies you would never catch on your own.

Working with different types of documents

The approach works across a wide range of official correspondence. Council tax letters often include dense references to bands, discounts, and exemptions that vary by local authority. The assistant can explain which band your property falls into, what the single-person discount means for your total, and whether the instalment plan described in the letter is your only option.

Medical explanations of benefits, common in private healthcare or insurance claims, use coding systems and procedural language that can make a routine appointment look like a billing error. The assistant parses the codes, matches them to plain-language descriptions, and highlights any charges that appear outside your plan's coverage.

Legal notices, whether from a solicitor, a landlord, or a government body, carry particular anxiety because the consequences of misunderstanding them can be significant. While the assistant cannot replace legal counsel, it can translate the formal language into something you can read without a law degree, and it can identify the specific deadlines and required responses so you know what needs attention first.

Pension statements, insurance policy documents, tenancy agreements, HMRC correspondence: the list of documents that benefit from plain-language explanation is long. The common thread is language written for compliance rather than comprehension, and Fabric's role is to bridge that gap.

Extending access through MCP and agents

If you already use ChatGPT or Claude as part of your daily workflow, you do not need to switch tools to benefit from Fabric's document store. Fabric's MCP integration lets external AI assistants access your Fabric documents with your permission, so you can ask ChatGPT about a bill stored in Fabric without leaving the conversation.

For more structured workflows, agent access provides scoped, time-limited permissions. You might grant an agent read access to your utility bills folder for thirty minutes while you work through a billing dispute, then revoke that access automatically. The document stays in Fabric, protected by AES-256 encryption and CASA Tier 2 certification, with your data never used for AI model training.

Privacy with sensitive documents

Bills and official letters contain some of the most sensitive information in your life: account numbers, addresses, financial details, medical information. Fabric treats this with appropriate seriousness. All documents are encrypted with AES-256, the platform holds CASA Tier 2 certification, and your data is never used to train AI models. The AI reads your documents to answer your questions, then moves on. There is no secondary use, no aggregation, no profiling.

This matters particularly for the kinds of documents discussed here. You should not have to choose between understanding a letter and protecting your privacy. Fabric is built so that you do not have to.

Building a household document habit

The real power of this workflow emerges when it becomes habitual. If you keep all your household documents in one place, every new letter or bill arrives into a context that makes it immediately more understandable. You can scan paper documents as they arrive and ask the assistant about them before they hit the kitchen counter pile. Over time, going paperless means that your entire correspondence history is searchable, connected, and ready to answer questions at any point in the future.

A bill from three years ago that you need for a tax query or expense claim is not buried in a drawer or lost in an email thread. It is in Fabric, indexed, and the assistant can find it and explain it just as readily as the day it arrived.

Frequently asked questions

Can Fabric read a photographed letter, or does it need a PDF?

Fabric handles both. You can photograph a paper letter using quick capture on your phone, and Fabric will extract the text automatically. PDFs, Word documents, and images all work. The handwriting scanner also picks up handwritten notes and annotations on printed letters.

What if the letter is in legal language I do not understand at all?

That is precisely what the AI assistant is designed to help with. Ask it to explain the letter in simple terms, and it will translate legal and regulatory language into plain English. You can follow up with specific questions about any clause or term that remains unclear.

Does the AI give legal or financial advice?

No. The assistant explains what documents say and highlights key facts, deadlines, and obligations. It does not recommend courses of action or provide professional advice. For legal disputes or complex financial decisions, you should consult a qualified professional, but the assistant's explanation can help you prepare for that conversation.

Can I compare a new bill with previous ones?

Yes, and this is one of Fabric's strongest features. If your earlier bills are stored in Fabric, the memory engine connects them automatically. You can ask the assistant to compare charges, spot increases, or check whether a quoted rate matches your original contract.

How does Fabric handle sensitive information on bills and letters?

All documents are encrypted with AES-256, Fabric holds CASA Tier 2 certification, and your data is never used to train AI models. The AI reads your documents only to answer your questions. There is no secondary use of your personal information.

Can I use this with bills that arrive by email?

Yes. Connecting your Gmail account lets Fabric pull in email bills and attachments automatically. Once they are in your cloud drive, you can ask the assistant about them in the same way as any other document.

What types of official letters work best with this?

The assistant handles utility bills, council tax notices, medical explanations of benefits, insurance documents, pension statements, tenancy agreements, HMRC correspondence, legal notices, and most other forms of official correspondence. Any document heavy with jargon, codes, or regulatory language benefits from a plain-language explanation.

Can I ask follow-up questions about the same letter?

Yes. The AI assistant maintains the conversation context, so you can ask an initial question, then drill into specific sections, charges, or terms without re-uploading or re-explaining the document.

What if I want to use ChatGPT or Claude to ask about my documents?

Fabric's MCP integration lets external AI assistants access your stored documents with your permission. You can ask ChatGPT or Claude about a bill in Fabric without switching tools, and agent access provides scoped, time-limited permissions for more structured workflows.

Does Fabric work with handwritten letters or notes?

Yes. The handwriting scanner recognises handwritten text on both standalone notes and annotations on printed documents. If a council officer or doctor has scrawled a note on an official letter, Fabric can read it.

Related pages

• How to keep all your household documents in one place

• How to scan paper documents with your phone

• How to find a receipt when you need it for expenses

• How to go paperless at home