Skip to content
The platform

How Analect works

Analect is the platform from ETT that shows what your AI is fed, what it costs and whether what it says is supported by its source. Its workflows cover the path from document to checked output: they read and label documents, test whether sending a model less still gets the same answers, price each workload across around 200 models, and certify AI outputs claim by claim with evidence an auditor can check.

Underneath the three services are seven workflows, each answering one question. Two of them are what the services run on, the Assurance Pack being a mode of the Evidence File’s. Customers sign in to the ones their work uses, and nothing stored in one organisation’s account is visible from another’s.

The workflows

Each workflow answers one question

They sit beside your AI, not in the path of it: nothing here routes a request or changes a model. Each one measures a decision and leaves a record of it.

What your AI is fed

What is in the documents your AI reads, who may see them, and does sending a model less still get the same answers?

What it costs

Which model fits each job, what will running it cost, and what will the platform under it cost?

Whether what it says is supported

Does the source support what the output says, what did the output leave out, and can anyone check that we said so?

Certificates anyone can check

A certificate is only worth what somebody else can verify

Every Certificate of Accuracy carries a reference and the SHA-256 digests of the exact source and output it was issued against. Change a word of either and the digest no longer matches.

Anyone holding a certificate, a client, an auditor or an insurer, can check on the public verification page that we issued it, that it has not been withdrawn, and that the file in front of them is the one it covers. Nothing is uploaded to do it: the file is digested in their browser, and the registry it is checked against holds digests and dates, never the documents.

Open the verification page

The knowledge engine

Read a document once, not once per question

A language model has no index into a document. To answer a question it reads everything, so every question pays for every token, and the same document is paid for again on the next question. Analect's knowledge engine reads a document once, into its sentences and an index of what each sentence is about. A question can then find the sentences that hold its answer rather than the pages around them. What a model reads is the document's own words, and each one can be traced to its place in the source.

Measured, not assumed. Whether that works on your material is something to measure, not to assume. Analect Proof runs the comparison on your own documents and says holds, does not hold or not shown for each model. We publish no figure for it until one holds beyond our own sample, with the method beside it. The original document is always kept.

Your data

What we keep, and what we never do

Your bills stay in your browser

The AI Spend Forecast reads your billing files in your browser and never uploads them. We keep the forecast built from them, and you can leave the cost figures out.

Your documents are not kept

The Accuracy Evidence File never keeps your source documents or the outputs themselves. We keep the certificates: the claims, and the quotes each verdict rests on.

Named models, named terms

Checks run on models from Anthropic and OpenAI, through Vercel's AI Gateway, under their commercial API terms. Our sub-processor list is available on request.

How an engagement grows

Start with a service. Go further only if it earns it

  1. Step 1

    A fixed-price service

    Three weeks on your own bills or outputs, or a quarter for the Assurance Pack. You keep the report and the working file whatever you decide next.

  2. Step 2

    The platform

    Where the service shows something worth going further on, your team gets Analect itself: the workflows your work needs, set up by ETT on your own material, with the evidence trail kept as you go.

Questions

About the platform

Is Analect a RAG tool?

Its knowledge engine does a related job. Retrieval sends a model part of a document rather than all of it, and so does the engine: it indexes each sentence by what it is about and sends the sentences a question needs. Whether that beats the retrieval you already have is not something we claim. Analect Proof measures it on your documents, with plain search at the same budget as the control, and reports the questions where it did no better.

Do we have to delete our source documents?

No, and nothing about Analect suggests you should. Fragments are an index into your knowledge, not a replacement for the record. Each fragment links back to the sentence it came from, which only works while the source exists.

Which workloads does the knowledge engine not help with?

Five of them. Summarise-everything work, because the model needs the whole document anyway. Questions that need every instance of something across a long document, a complete list of obligations for example, because retrieval sends the passages most like the question, not all of them. Corpora that change constantly, because decomposition is paid again on every change. Documents you question once, because there is nothing to amortise against. And small corpora, where reading everything is already cheap.

Who can sign in to Analect?

Customers, with accounts set up by ETT. Each account sees only the workflows it has been granted, and nothing stored in one organisation's account is visible from another's.

See it on worked material, then on yours

Thirty minutes with the team that runs Analect: the AI Spend Forecast and the Accuracy Evidence File on worked examples, and what either would look like on your bills or your outputs.

Book a demo