Paige Intelligence · Multimodal AI

Review isn't just about reading text.
It's about understanding.

Your headline, your chart, your image, your layout. Together they communicate something your content may never say out loud. Paige Review sees the whole piece, and understands what it means.

Book a discovery call → View example flags ↓

Eleviora IVF medication advertisement depicting mother and baby

Paige Review

Critical · p.1 · imagery

CRITICAL Visual analysis · imagery

Authorized Use

The imagery depicts a mother holding a baby, which implies a successful pregnancy and live birth. This outcome is not guaranteed by the drug Eleviora™ (corifollitropin alfa), which is only authorized to support follicle development in IVF treatment. The depiction of a baby is a critical violation as it suggests an outcome that the drug cannot ensure.

RECOMMENDATION Ensure imagery depicts only populations, outcomes, and use contexts consistent with the authorized indication. Replace outcome imagery that exceeds the indication with imagery representing the actual authorized use. If patient imagery is retained, verify the depicted population matches the authorized patient selection criteria in the Product Monograph.

What Paige catches

The violations text-only AI
can't see.

Pharma promotional content isn't just text. It's charts, images, layout, typography. Compliance problems hide in all of them. Here are sample findings across each content layer.

01 / TEXT & COPY

Claims and benefit language

Identifies product claims, checks fair balance across body copy, and flags unsubstantiated or superlative language using NLP that understands intent, not just exact matches.

02 / CHARTS & DATA

Misleading data presentations

Flags truncated axes, selective scaling, and chart designs that make a modest efficacy difference look dramatic. Analysis that requires actually seeing the chart.

03 / IMAGES & LAYOUT

Visual and contextual claims

Recognizes what photographs actually depict, detects implied outcomes, checks font size and prominence of safety information. The layer no text-only AI can reach.

01 / Text analysis

Paige Review

Critical · p.1 · main headline

CRITICAL Claim analysis

Fair Balance of Risk and Benefit

The main headline "85% Symptom Reduction!" and subheadline "Fast. Effective. Life-Changing." are in much larger font than the safety information in the footer.

RECOMMENDATION Ensure risk information has comparable visual prominence to benefit claims. Use similar font sizes for benefits and risks, place risk information in the main body (not just footer), apply similar visual treatments, and allocate reasonable space to safety information.

WONDERDRUG™

(miraclizumab) 100mg injection

85% Symptom Reduction!

Fast. Effective. Life-Changing.

Key Benefits:
✓ 85% of patients achieved symptom reduction
✓ Works in as little as 2 weeks
✓ Once-monthly convenient dosing
✓ Proven efficacy across all patient subgroups

Important safety information: Serious infections, malignancies, and hypersensitivity reactions have been reported. See full prescribing information.

02 / Chart analysis

Paige Review

Warning · p.2 · chart

WARNING Visual analysis · chart

Misleading data presentation

Y-axis starts at 60%, visually exaggerating the difference between WONDERDRUG (85%) and placebo (62%). The actual 1.37× ratio reads as a dramatic gap. A text-only AI sees the numbers. It cannot see that the chart distorts them.

RECOMMENDATION Extend the y-axis to zero so the visual proportion reflects the true magnitude of difference. Ensure chart design does not amplify the clinical effect beyond what the data supports.

Clinical Efficacy Results

100% 90% 80% 70% 60%

85%

WONDERDRUG

62%

Placebo

DRAMATICALLY superior efficacy!

How Paige reads your content

Most AI tools scan text.
Paige reads the whole page.

Paige reads your promotional piece the way a human reviewer does: seeing layout, charts, imagery, and copy together, and understanding what they communicate as a whole.

Reads the copy

Identifies product claims, checks for fair balance, and flags unsubstantiated or superlative language using NLP that understands intent, not just exact matches.

Reads the layout

Analyzes typography hierarchy, font sizes, color contrast, and placement. Detects when safety information is cramped, low-contrast, or relegated to a footer where it loses impact.

Reads the imagery

Recognizes what photographs and visuals actually depict. Flags when imagery implies outcomes, patient populations, or use contexts beyond the authorized indication.

Reads the code

Maps findings to the relevant regulatory guidance with severity ratings (Critical, Warning, or Info) plus plain-language explanations, location pointers, and actionable recommendations.

Built for regulated industries

Your content stays yours.

Papercurve has been purpose-built for life sciences since 2019. Data handling isn't an afterthought. It's the foundation.

✓

No cross-client training

Your Paige instance learns from your team's reviews only. No client data is ever used to train models serving other customers.

✓

Regional data residency

Data hosted in your region. Canada and US availability today, EU on the roadmap.

✓

Human-in-the-loop, always

Paige surfaces findings for your reviewers to decide on. Final approval authority stays with your team, never with the AI.

Why visual analysis matters

The compliance issues that live outside the text.

Text-only AI tools extract claims from words. But in pharma promotional content, some of the most significant violations exist in the visual layer. And those tools never see them.

Truncated axes

Charts that distort data

A y-axis starting at 60% can make a modest 23-point difference look like a dramatic gap. Text extraction sees the numbers. Multimodal AI sees what the chart actually communicates.

Image context

Photos that make implicit claims

An athlete, a smiling family, a return to work. Images imply outcomes. When those outcomes exceed the approved indication, it's a compliance issue invisible to text analysis.

Typography

Safety buried in 6-point type

Risk statements cramped into footers, low-contrast on coloured backgrounds, or de-emphasized through layout. Violations detectable only through visual document analysis.

Video & broadcast

Commercials and digital video

Paige uses multimodal AI to analyze video content for compliance, a capability no text-based tool offers. Relevant for TV, digital, and social media advertising.

Review isn't just about reading text.
It's about understanding.

The violations text-only AI
can't see.

Most AI tools scan text.
Paige reads the whole page.

Your content stays yours.

Reviews that used to take weeks
now take days.

The compliance issues that live outside the text.

Let's talk about how Paige fits
into your review workflow.

Review isn't just about reading text.It's about understanding.

The violations text-only AIcan't see.

Most AI tools scan text.Paige reads the whole page.

Your content stays yours.

Reviews that used to take weeksnow take days.

The compliance issues that live outside the text.

Let's talk about how Paige fitsinto your review workflow.

Review isn't just about reading text.
It's about understanding.

The violations text-only AI
can't see.

Most AI tools scan text.
Paige reads the whole page.

Reviews that used to take weeks
now take days.

Let's talk about how Paige fits
into your review workflow.