For GRC leads, security engineers and CISOs

Security questionnaires, streamlined

Most of the questions in the next SIG or CAIQ are ones your team has answered before. AnswerTrail keeps those approved answers in one knowledge base, drafts the new questionnaire from them, and shows the source of every draft so an expert can approve it with confidence.

Published prices from $199 a month. Month-to-month. No sales call.

Question 47 of 150 Matched · reviewed answer

Is customer data encrypted at rest, and how are the keys managed?

Yes

All customer data is encrypted at rest with AES-256. Keys are held in a managed key service and rotated every year.

  • Source: Vendor security assessment, row 112
  • Answered 4 months ago · approved by a reviewer
  • Scope: organisation-wide (ISMS)
Illustration with sample text, not a real customer answer.

The problem

The answers exist. Finding the right one is the work.

A mid-size deal often arrives with a 100 to 300 question spreadsheet. The answers are usually somewhere already: in last quarter's questionnaire, the Statement of Applicability, or the SOC 2 report. Tracking down the latest approved wording, and checking it is still true, is slow and falls on the same few people every time.

20–40 h

for a 200-question questionnaire without an answer library

12–18 h

for the same questionnaire with an answer library

And old answers go stale

Last year's answer may describe a control that has since changed. A copied "yes" that is no longer true can become a contractual misrepresentation.

Hours are a vendor estimate (Wolfia, "How long a 200-question security questionnaire takes"), not our own measurement.

How it works

Five waypoints from inbox to sent

  1. Upload past questionnaires

    Excel questionnaires you have already sent. Each answer is copied word for word and keeps its source document and date.

  2. Upload a new one

    AnswerTrail finds the questions and the answer columns. You confirm the mapping before anything is drafted.

  3. Review cited drafts

    Each draft shows the past answers it came from, their age and who approved them. Accept, edit or flag it.

  4. Export the original file

    Approved answers are written into a copy of the customer's own spreadsheet. Sources and reviewer notes stay internal.

  5. Answers flow back

    Reviewed answers can be added to the library, so the next questionnaire starts from more of your own approved wording.

Features

Built for security answers, not general RFPs

A wrong "yes" has consequences, so the product is built around evidence, scope and review.

Knowledge base ingestion

Past questionnaires in Excel become a searchable library. Answers are stored exactly as they were sent, with the source document, version and date.

Answer aging

Newer approved answers win near-ties, and old ones are marked stale instead of being hidden. Confirm an answer is still accurate, or retire it with a reason.

Scope awareness

Answers are labelled as organisation-wide or product-specific. Set the products a questionnaire is about, and drafts warn before reusing another product's answer.

Review queue

Accept, edit or flag every draft. When the evidence is missing the draft says so and stays empty, and you can redraft one answer with your own guidance.

Export and feedback loop

Get the customer's file back with answers written into the empty answer cells only. Reviewed answers can then be promoted into the library.

Audit trail

Every recommended answer shows where it came from: which questionnaire, when it was answered, who approved it and who added it to the library.

Bring your own LLM

Connect your own Anthropic, OpenAI, Azure, Bedrock or internal endpoint for data-residency or cost reasons, or choose a plan with AI included.

Tenant isolation

Each organisation's library and questionnaires are separated in the database by row-level security. One customer's answers are never retrieved for another.

Questionnaire hours calculator

What do questionnaires cost your team each month?

Enter your own numbers. The calculator compares answering from scratch with answering from a maintained answer library, which is what AnswerTrail gives you.

Assumptions

  • Without an answer library: 20–40 hours per 200 questions (6–12 minutes a question).
  • With an answer library: 12–18 hours per 200 questions (3.6–5.4 minutes a question).
  • Both ranges are a vendor estimate (Wolfia), scaled in proportion to the number of questions.
  • "Difference" compares the midpoints of the two ranges.

1,200 questions a month

Without an answer library120 – 240 h

$9,000 – $18,000

With an answer library72 – 108 h

$5,400 – $8,100

Difference at the midpoints

90 h and $6,750 a month

This is an estimate from published vendor figures, not a measurement of AnswerTrail and not a promise. Your result depends on how much of each questionnaire your library already covers.

Pricing

Published prices. No sales call.

Pay month-to-month by card and change or cancel your plan yourself.

Bring your own LLM

For teams that must keep model traffic on their own account.

$199/month

  • Connect your own Anthropic, OpenAI, Azure, Bedrock or internal endpoint
  • 3 seats
  • Questionnaires not capped
  • Knowledge base, review queue, export and audit trail
Choose Bring your own LLM

AI included

Pro

For teams answering questionnaires every week.

$799/month

  • AI included, nothing to connect
  • 20 seats
  • 50 questionnaires a month
  • Knowledge base, review queue, export and audit trail
Choose Pro

Have a registration code? Create your account first, then redeem the code on the plan page after you sign in.

Security and data handling

Your answers are sensitive. We treat them that way.

A questionnaire library holds policies, control descriptions and customer names. These are the rules the product is built on.

AnswerTrail is a beta-stage course project. It has not been audited and holds no certifications of its own, so we describe how it works rather than show badges.

  • Isolated by tenant

    Every organisation's data is separated by Postgres row-level security, and every retrieval query runs inside that organisation's scope.

  • Retrieval stays on our servers

    Embedding and reranking models run on our own servers, so the text used to search your library is not sent to a third party.

  • Admins manage accounts, not content

    Platform administrators can provision accounts and change subscriptions. They cannot open customer questionnaires or knowledge-base entries.

  • A person approves every answer

    Drafts are always marked as AI drafts with their sources. Only reviewed answers are written into the exported file.

  • Never used for training

    Customer data is never used to train models. On the bring-your-own-LLM plan, drafting calls go to the endpoint you connect.

  • Other customers' names stay internal

    The exported file carries answers only. Sources and provenance go in a separate internal review report.

FAQ

Questions before you set out

Which file formats can I upload?

Past questionnaires for your library are Excel files (.xlsx, .xlsm). New questionnaires to answer can be Excel, Word (.docx) or text-based PDF. For Excel you get the original file back with the answers filled in. For Word and PDF you get a separate answers spreadsheet. Scanned PDFs that need OCR are not supported yet.

Does AnswerTrail send questionnaires without me?

No. It drafts, and a person decides. Every answer is accepted, edited or flagged by a reviewer, and you download the file and send it yourself. You approve and own the final answers.

What happens when there is no past answer?

The draft says the evidence is insufficient and leaves the answer empty for a reviewer, with a note on what is missing. AnswerTrail is designed not to claim a control your library does not support.

How accurate are the drafts?

We are measuring that now on a 100-question test set with known good answers. Our targets are at least 70% of drafts accepted without edits, at least 95% citation accuracy and zero false "yes" answers. Those are targets, not results. We will publish the measured numbers when the test is complete.

What does "bring your own LLM" mean?

On the $199 plan you connect your own model endpoint (Anthropic, OpenAI, Azure, Bedrock or an internal gateway) with your own key. Drafting calls go to that endpoint and are billed by your provider. Search over your library still runs on AnswerTrail's own embedding and reranking servers.

Can AnswerTrail staff read my questionnaires?

The management portal used by platform administrators covers accounts, subscriptions and registration codes only. It has no access to questionnaires or knowledge-base entries, and the database enforces that separation for every tenant.

How do I pay, and can I cancel?

Pick a plan after you sign up and pay by card through Stripe-hosted checkout, so card details never reach our systems. Plans are month-to-month. During the beta, checkout runs in Stripe test mode and no real charges are made.

I have a registration code. Where do I enter it?

Create your account, sign in, and enter the code on the plan page instead of choosing a paid tier. The code activates the plan it was issued for.

Who is behind AnswerTrail?

AnswerTrail is a project for CS 789 at the University of Nevada, Las Vegas. It is at beta stage and runs for a limited time. Please do not upload material you are not permitted to share with a beta service.

Start with the questionnaires you have already answered

Create an account, upload a few past questionnaires and try a new one against them.

Get started