# When the AI has little to work from

URL: https://engiprose.com/help/write-with-ai/novel-rfp-and-thin-prior-content
Updated: 2026-09-02
Audience: member

What the Novel RFP chip and a Limited prior content marker mean, and why a lower score is not a sign your writing is weak.

Sometimes the AI has less to draw on than usual, either because the RFP
itself is unfamiliar territory or because your content library does not yet
have much like a particular section. Two indicators tell you when that is
happening, so a lower score reads as context instead of a mystery.

:::checklist title="What you'll do"
- Recognize the Novel RFP chip and read its explanation
- Recognize a Limited prior content marker on a section
- Know what to do about it
:::

## The Novel RFP chip

If at least two in five of a proposal's requirements matched past proposals
only with low confidence, the proposal is classified Novel RFP and a chip
appears in the AI Coach panel:

:::annotated-figure src="/help-figures/write-with-ai/novel-rfp-and-thin-prior-content-1.png" srcDark="/help-figures/write-with-ai/novel-rfp-and-thin-prior-content-1-dark.png" alt="The AI Coach panel with the Novel RFP chip open, showing its explanation popover" caption="The Novel RFP chip and its popover" route="/proposals/$proposalId/write/edit"
1. The Novel RFP chip
2. Its popover, open
:::

Clicking the chip opens this explanation, quoted exactly as it appears:

> **Novel RFP**
>
> This RFP covers territory your firm hasn't proposed often. Most sections
> matched with low confidence against past proposals.
>
> **What this means:**
>
> - Rubric scores may be lower than a familiar RFP. This does not mean your
>   writing is weak.
> - The rubric is checking structural alignment with the RFP, not comparing
>   you to past wins.
> - Submitting a novel-RFP proposal at 62% is often a strong result.

Select **Got it** to dismiss it. The same idea appears again, briefly, if you
choose Yolo to generate a full draft: "Novel RFP: Yolo will draft every
section using the available matches, even when those matches are limited."

## Limited prior content on a section

Separately, an individual section can show its own **Limited prior content**
marker when your content library does not have much past work like that
section yet:

:::figure src="/help-figures/write-with-ai/novel-rfp-and-thin-prior-content-2.png" srcDark="/help-figures/write-with-ai/novel-rfp-and-thin-prior-content-2-dark.png" alt="A section showing the Limited prior content marker" caption="A section with limited prior content" route="/proposals/$proposalId/write/edit"
:::

Its explanation, also quoted exactly:

> **Limited prior content for this section**
>
> Your library doesn't have much past work like this section yet, so this
> draft leans on general writing guidance rather than your firm's prior
> winning proposals.
>
> That's normal for a new client or a type of project you haven't proposed
> before. It's not a problem with your draft.

Notice that neither explanation states a count or a percentage. The AI's
sense of "limited" is deliberately kept qualitative, so it never reads as an
arbitrary number to chase.

## What to do about it

Both indicators point at the same fix over time: "To make drafts like this
stronger over time, add a few past proposals with similar sections to your
content library." See
[Upload past proposals](/help/set-up-your-firm/upload-past-proposals) to add
more. For the proposal in front of you right now, treat the AI's draft as a
starting point and lean more on your own review and rewriting than usual.

::::faq title="Common questions"
:::q question="Does a Novel RFP or Limited prior content marker mean my proposal will score badly?"
No. Both are context, not a verdict. The rubric still scores what is
actually on the page; these markers explain why the AI had less to draw on
while writing it.
:::
:::q question="Will these markers go away once I add more content to my library?"
Limited prior content on future sections will improve as your library grows
to cover more of what you propose. Novel RFP is worked out from a
proposal's own requirement matches, so it changes only if you run matching
again for that proposal.
:::
::::