---
title: "Analyse the RFP and run AI assessments"
description: "Review what the AI extracted from the tender documents, run readiness, improvement and gap assessments, apply the suggestions you accept, and search for evidence your organisation already holds."
url: https://helpcenter.complytrain.net/en/articles/235-analyse-the-rfp-and-run-ai-assessments
collection: "Requirements and analysis"
updated: 2026-09-22
language: en
---

# Analyse the RFP and run AI assessments

Review what the AI extracted from the tender documents, run readiness, improvement and gap assessments, apply the suggestions you accept, and search for evidence your organisation already holds.

Review what the AI extracted from the tender documents, run readiness, improvement and gap assessments, apply the suggestions you accept, and search for evidence your organisation already holds.

Tenders uses AI in two places before you start writing. **RFP Analysis** reads the tender documents you upload and proposes the requirements, deadlines, formalities and expected outputs it finds. The **AI assessments** tab then judges how ready the bid is, suggests improvements and looks for gaps. Neither changes the tender until you choose what to keep.

## Who can do what

Product location: `/tender/workspaces/{workspaceId}` (fallback: `/tender/workspaces`). Open the tender in Tenders and choose the AI assessments tab.

<table style="min-width: 50px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>Task</p></th><th colspan="1" rowspan="1"><p>Who</p></th></tr><tr><td colspan="1" rowspan="1"><p>See past assessment runs and their results</p></td><td colspan="1" rowspan="1"><p>Team members with <strong>View Tenders</strong> (<code>tender:view</code>)</p></td></tr><tr><td colspan="1" rowspan="1"><p>Review an RFP analysis and add its items, run assessments and apply their suggestions</p></td><td colspan="1" rowspan="1"><p>Team members with <strong>Create &amp; Work on Tenders</strong> (<code>tender:manage-workspaces</code>)</p></td></tr><tr><td colspan="1" rowspan="1"><p>Use <strong>Find evidence</strong> and <strong>Cross-reference knowledge</strong></p></td><td colspan="1" rowspan="1"><p>Team members with <strong>Create &amp; Work on Tenders</strong></p></td></tr></tbody></table>

Administrators can do all of these on every tender. Everyone else works only on tenders whose team they belong to; see [the tender team](https://help.complytrain.net/en/tender/team-and-details/). Buttons you cannot use are not shown.

## Review the RFP analysis

Product location: `/tender/workspaces/{workspaceId}/rfp-analysis/{documentId}` (fallback: `/tender/workspaces`). Open the tender and open the RFP analysis that is waiting for review.

When you create a tender with **Upload a tender document**, each uploaded document is analysed separately. The wizard then continues to the Tender Assistant. Going on to the chat applies nothing: every analysis waits on the tender until you have reviewed it.

Each analysis goes through two stages:

1.  **Preparing your RFP…**: the document is read (with text recognition for scanned pages) and indexed. Audio and video recordings are transcribed first.
    
2.  **Extracting from your RFP…**: the AI works through the whole document, extracting requirements, deadlines, formalities and expected outputs in turn.
    

When an analysis finishes, the tender shows it as waiting for review. Open it to see the **RFP Analysis** page for that document. If you open it before it finishes, the page shows the current stage and moves on to the review by itself.

The review lists what was found under **Requirements**, **Deadlines**, **Formalities** and **Expected outputs**. Each heading shows how many items are ticked out of how many were found. Each item shows its type or category and, where the AI could locate it, the page (for example **p. 12**) and section of the RFP it came from. Expected outputs that the RFP makes mandatory carry a **Required** badge.

Everything starts ticked.

1.  Read each item against the RFP. Untick anything you do not want, such as a duplicate or a general statement that is not a real obligation.
    
2.  Choose **Add N to the tender**, where N is the number of ticked items. The button reads **Adding…** while it works.
    
3.  A message confirms how many items were added, and the tender opens.
    

Choose **Skip for now** to return to the tender without adding anything. The analysis keeps waiting for review, so you can come back to it. Review each uploaded document's analysis the same way; when several documents repeat the same requirement, untick the repeats in the later reviews.

## What the analysis adds

Product location: `/tender/workspaces/{workspaceId}/rfp-analysis/{documentId}` (fallback: `/tender/workspaces`). Review the analysis on the RFP Analysis page, then open the tender's Requirements, Deadlines, Formality and Output documents tabs.

Each ticked item becomes an ordinary tender item, marked as extracted by AI. You can edit or delete it like any item you added yourself.

<table style="min-width: 75px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>Found under</p></th><th colspan="1" rowspan="1"><p>Becomes</p></th><th colspan="1" rowspan="1"><p>Details carried over</p></th></tr><tr><td colspan="1" rowspan="1"><p>Requirements</p></td><td colspan="1" rowspan="1"><p>A requirement on the <strong>Requirements</strong> tab</p></td><td colspan="1" rowspan="1"><p>Text, type (Technical, Financial, Legal, Administrative or Other) and priority; anything the AI could not classify becomes Other and Medium</p></td></tr><tr><td colspan="1" rowspan="1"><p>Deadlines</p></td><td colspan="1" rowspan="1"><p>A deadline on the <strong>Deadlines</strong> tab</p></td><td colspan="1" rowspan="1"><p>Name, date and type, such as submission, Q&amp;A close or clarification</p></td></tr><tr><td colspan="1" rowspan="1"><p>Formalities</p></td><td colspan="1" rowspan="1"><p>A formal requirement on the <strong>Formality</strong> tab</p></td><td colspan="1" rowspan="1"><p>Name, type and description</p></td></tr><tr><td colspan="1" rowspan="1"><p>Expected outputs</p></td><td colspan="1" rowspan="1"><p>An output document on the <strong>Output documents</strong> tab</p></td><td colspan="1" rowspan="1"><p>Title, code, description, category and whether it is required</p></td></tr></tbody></table>

The page reference stays with each item, so you can trace it back to the RFP. See [requirements](https://help.complytrain.net/en/tender/requirements/), [deadlines and formalities](https://help.complytrain.net/en/tender/deadlines-formalities/) and [output documents](https://help.complytrain.net/en/tender/draft-map-response/).

If nothing was found, the page says so and you can build the tender by hand. If the document cannot be read, **RFP analysis couldn't finish** explains why, with **Go to tender** to continue. A recording can be analysed only when your administrator has enabled transcription.

## Analyse a document added later

Product location: `/tender/workspaces/{workspaceId}/chat` (fallback: `/tender/workspaces`). Upload the document through Open Document Vault on the tender's Documents tab, then choose Chat with AI in the tender and ask the assistant to analyse it.

Clarification answers and addenda often arrive after the tender starts.

1.  On the tender's **Documents** tab, choose **Open Document Vault** and upload the document to the tender's vault. See [finding and uploading documents](https://help.complytrain.net/en/document-vault/find-upload-documents/).
    
2.  Open **Chat with AI** and ask the Tender Assistant to analyse the new document, naming it.
    

The assistant starts an analysis of that document. When it finishes, it waits for review on the tender like any other; review it as described above. See [the Tender Assistant](https://help.complytrain.net/en/tender/assistant/).

RFP analysis reads only this tender's documents.

## Run an assessment

Product location: `/tender/workspaces/{workspaceId}` (fallback: `/tender/workspaces`). Open the tender in Tenders and choose the AI assessments tab.

Open the tender and choose the **AI assessments** tab. Choose one of the three buttons:

<table style="min-width: 75px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>Button</p></th><th colspan="1" rowspan="1"><p>Assessment</p></th><th colspan="1" rowspan="1"><p>What you get</p></th></tr><tr><td colspan="1" rowspan="1"><p><strong>Evaluate readiness</strong></p></td><td colspan="1" rowspan="1"><p><strong>Readiness check</strong></p></td><td colspan="1" rowspan="1"><p>A score out of 100, a verdict (<strong>Not ready</strong>, <strong>Needs work</strong> or <strong>Strong</strong>), a summary, scored <strong>Dimensions</strong> with a red, amber or green marker and commentary, and <strong>Top risks</strong></p></td></tr><tr><td colspan="1" rowspan="1"><p><strong>Suggest improvements</strong></p></td><td colspan="1" rowspan="1"><p><strong>Improvement suggestions</strong></p></td><td colspan="1" rowspan="1"><p>A summary and a list of suggestions you accept or reject</p></td></tr><tr><td colspan="1" rowspan="1"><p><strong>Identify gaps</strong></p></td><td colspan="1" rowspan="1"><p><strong>Gap analysis</strong></p></td><td colspan="1" rowspan="1"><p>A summary and a list of requirements that cannot yet be evidenced</p></td></tr></tbody></table>

The readiness dimensions always include requirements coverage, evidence and proof, formality compliance, output documents and timeline.

Each assessment works from what the tender records: its details, requirements with their type, priority, status and any attached document, deadlines, formalities and output documents with their status. **Identify gaps** also searches your organisation's documents for each requirement. Assessments do not read the text of your response sections, so run **Validate** on a response document to check its content against the requirements; see [compliance and coverage](https://help.complytrain.net/en/tender/compliance-matrix/).

The run is added to **History** with the badge **Queued**, then **Running**, and its button reads **Running…**. You can keep working, and you can run the other two types at the same time. When the run finishes, a message says it is ready to review or could not be completed, and the person who started it receives a notification.

If a run takes longer than about two minutes, the tab stops waiting and says it is still running in the background. Reopen the tab later, or wait for the notification.

## Read results and history

Product location: `/tender/workspaces/{workspaceId}` (fallback: `/tender/workspaces`). Open the tender in Tenders and choose the AI assessments tab.

**History** lists every run on the tender, newest first, with its type, status and time. Select a run to show its result; until you do, the tab shows **Select a run to view its result.**

A **Failed** run shows the reason. Run the assessment again once the cause is dealt with.

Results are advice. A readiness score does not change the tender's status or the submission check, and a gap does not change any requirement. Use them to decide what to fix, then fix it in the relevant tab.

## Apply improvement suggestions

Product location: `/tender/workspaces/{workspaceId}` (fallback: `/tender/workspaces`). Open the tender in Tenders and choose the AI assessments tab.

Each improvement suggestion shows a title, a severity (**Low**, **Medium**, **High** or **Critical**), the rationale and, where useful, suggested wording.

1.  Select a completed **Improvement suggestions** run in **History**.
    
2.  For each suggestion, choose **Accept** or **Reject**. Your choice is highlighted but not yet saved.
    
3.  Choose **Apply decisions**. The button reads **Applying…**, then **Your decisions have been applied.** appears.
    

Decided suggestions show **Accepted** or **Rejected** and cannot be decided again. You can decide some suggestions now and the rest later.

A suggestion marked **Will add a requirement if accepted** identifies an obligation the tender is not yet tracking. Accepting it adds a requirement of type Other, using the suggested wording, or the title when there is none. Its priority follows the severity: Critical and High carry over, and anything lower becomes Medium. The requirement is marked as suggested by AI. Other accepted suggestions are recorded as accepted and change nothing by themselves; make the change in the response or the tender yourself.

Applying the same decisions again does not add the requirement twice.

## Read a gap analysis

Product location: `/tender/workspaces/{workspaceId}` (fallback: `/tender/workspaces`). Open the tender in Tenders and choose the AI assessments tab.

Each gap names the requirement, a severity and the kind of gap:

<table style="min-width: 75px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>Gap type</p></th><th colspan="1" rowspan="1"><p>Meaning</p></th><th colspan="1" rowspan="1"><p>Typical action</p></th></tr><tr><td colspan="1" rowspan="1"><p><strong>No supporting evidence</strong></p></td><td colspan="1" rowspan="1"><p>You can probably meet it, but nothing on file proves it</p></td><td colspan="1" rowspan="1"><p>Attach or produce the proof</p></td></tr><tr><td colspan="1" rowspan="1"><p><strong>Capability gap</strong></p></td><td colspan="1" rowspan="1"><p>Nothing suggests your organisation can meet it</p></td><td colspan="1" rowspan="1"><p>Review the requirement's disposition and the bid decision</p></td></tr><tr><td colspan="1" rowspan="1"><p><strong>Missing document</strong></p></td><td colspan="1" rowspan="1"><p>A document the tender needs has not been produced</p></td><td colspan="1" rowspan="1"><p>Add or complete the output document</p></td></tr></tbody></table>

**Suggested evidence** names the kind of document that would close the gap. **No gaps found** means every tracked requirement appears to be supported by what the tender and your documents record.

## Find reusable evidence

Product location: `/tender/workspaces/{workspaceId}` (fallback: `/tender/workspaces`). Open the tender in Tenders and choose the AI assessments tab.

The **Reuse existing evidence** card finds documents your organisation already holds that may answer a requirement.

1.  Enter the requirement, or a short description of it, in the box.
    
2.  Choose **Find evidence**.
    

Each match shows the document's name, its vault and a match percentage, with the reason, for example that the document's name matches the requirement. A match on the document's name scores higher than a match on its description or number. Passages from your organisation's shared knowledge bases can also appear. The search covers only documents you are allowed to read.

The **Find evidence** button on a requirement row on the **Requirements** tab runs the same search for that requirement. When the best document match is a name match and the requirement has no document attached yet, ComplyTrain attaches that document to the requirement and says **A confident match was automatically linked to the requirement.** The attachment then shows under **Addressed by** on the **Compliance matrix** tab. Check it there, because a name match is not proof that the document answers the requirement. See [compliance and coverage](https://help.complytrain.net/en/tender/compliance-matrix/).

## Cross-reference what you know

Product location: `/tender/workspaces/{workspaceId}` (fallback: `/tender/workspaces`). Open the tender in Tenders and choose the AI assessments tab.

The **Cross-reference knowledge** card searches several sources at once and changes nothing.

1.  Enter a question, such as "ISO 27001 certification scope", in **What do you want to look up?**.
    
2.  Choose **Search**.
    

The result is a short summary of what was found, attributed to its sources, followed by counts for **Facts** (recorded facts about your organisation), **Documents** (documents you can read), **KB passages** (this tender's knowledge base and your shared knowledge bases) and **Past submissions** (past tenders with a recorded outcome and your [submission history](https://help.complytrain.net/en/tender/submission-history/)).

## Example: from RFP to readiness

Product location: `/tender/workspaces/{workspaceId}` (fallback: `/tender/workspaces`). Open the tender in Tenders and choose the AI assessments tab.

A facilities team uploads a 60-page cleaning-services RFP and goes on to the chat. When the analysis is waiting for review, the bid manager opens it from the tender: **RFP Analysis** has found 42 requirements, 5 deadlines, 6 formalities and 4 expected outputs. The bid manager unticks three requirements that repeat the scope statement and a deadline that is really the contract start date, then chooses **Add 53 to the tender**.

After setting dispositions on the **Requirements** tab, they run **Evaluate readiness**. The result is **Needs work**, 58 out of 100, with **Evidence & proof** in red. **Identify gaps** lists "Proof of COSHH training for all operatives" as **No supporting evidence**. The bid manager types that requirement into **Reuse existing evidence**, finds the training matrix in the HR vault, and attaches it from the **Compliance matrix** tab.

**Suggest improvements** proposes adding a requirement for a business continuity plan that the RFP mentions in an annex. The bid manager accepts it, rejects two stylistic notes and chooses **Apply decisions**. The new requirement appears on the **Requirements** tab, ready to be mapped to a response section.
