Docs

Content Optimization

Honesty Engine

Content integrity verification and trust signal management.

Why honesty matters

AI agents are trained to identify and penalize inflated claims, unsubstantiated superlatives, and inconsistent data. When an agent detects that a product description overpromises or contradicts other available information, it reduces confidence in that source. This directly lowers recommendation rates.

Honest, verifiable, and complete data converts better than marketing copy in the agent ecosystem. Agents reward accuracy.

What the Honesty Engine does

The Honesty Engine is an automated verification layer that runs before any optimized content is served to agents. It compares generated content variants and structured data against your source material to ensure factual accuracy and consistency.

Every piece of content that Inception Agents serves on your behalf passes through the Honesty Engine. Nothing reaches an agent without verification.

Checks performed

The Honesty Engine runs the following verification checks:

Specification completeness

Ensures that key product specifications present in your source content are included in the optimized output. Flags cases where important details have been omitted during variant generation.

  • Feature lists match source pages.
  • Technical specifications are preserved accurately.
  • Limitations and requirements are not omitted.

Pricing accuracy

Verifies that all pricing information in optimized content matches your current published pricing.

  • Price values match source data exactly.
  • Currency and billing period are correctly stated.
  • Free tier and trial information is accurately represented.
  • Pricing qualifiers (starting at, per-user, etc.) are preserved.

Review consistency

Cross-checks review scores and customer metrics against source data.

  • Star ratings match aggregated review data.
  • Review counts are current and accurate.
  • Testimonial quotes are attributed correctly.
  • Customer count claims are verified against source.

Claim verification

Identifies and flags unsupported or inflated claims in generated content.

  • Superlatives (“best”, “fastest”, “only”) are flagged unless substantiated.
  • Performance claims are checked against published benchmarks.
  • Comparative claims are verified against available competitor data.
  • Compliance and certification claims are validated against known records.

Structured data parity

Ensures that JSON-LD and other structured data outputs are consistent with the content they describe.

  • Schema.org fields match page content.
  • Product feed data matches structured data.
  • Agent Card capabilities match actual site functionality.
  • No contradictions exist between different structured data outputs.

Trust signals

The Honesty Engine also manages the structured presentation of your trust signals to agents:

  • SOC 2 compliance — Verification status and report availability.
  • Uptime data — Historical uptime percentage, sourced from your status page or monitoring integration.
  • Review aggregation — Consolidated ratings from G2, Capterra, Trustpilot, and other review platforms.
  • Certification badges — Industry certifications, partner program memberships, and compliance standards.

Trust signals are structured in a format that agents can parse directly, using schema.org CredentialCategory and Review types alongside custom properties in your Agent Card.

How it works

The verification pipeline operates in-line with content serving:

  1. Content generation — A content variant or structured data update is produced by the optimization engine.
  2. Honesty check — The generated content is compared against your crawled source data and known trust signals.
  3. Pass / Flag — Content that passes all checks is approved for serving. Content that fails any check is flagged.
  4. Flagged content handling — Flagged content is held from serving and queued for your review in the dashboard. The previous approved version continues serving until the flag is resolved.
  5. Resolution — You can approve the flagged content (overriding the check), edit it, or regenerate it.

This process adds no perceptible latency. Checks run asynchronously during content generation, not at request time.

Dashboard alerts

When the Honesty Engine flags content, you receive notifications through your configured alert channels:

  • Dashboard notification — A flag appears on the affected content chunk in the Optimization tab.
  • Email alert — Summary of what was flagged and why, with a direct link to review.
  • Slack webhook — Real-time notification to your configured Slack channel.

Each alert includes:

  • The specific check that failed.
  • The generated content that was flagged.
  • The source content it was compared against.
  • A suggested resolution.

The competitive advantage

Businesses with verified, honest data see measurably higher recommendation rates from AI agents. This effect compounds over time as agents build trust profiles for domains they interact with.

Key findings from agent behavior analysis:

  • Consistent data across sources increases agent confidence and recommendation likelihood.
  • Acknowledged limitations (e.g., “not suitable for X”) build trust rather than reducing conversions — agents prefer complete information.
  • Verified claims with supporting data (benchmarks, third-party reviews, certifications) are weighted significantly higher than unverified assertions.
  • Transparent pricing with clear terms reduces agent hesitancy in recommendation contexts.

The Honesty Engine is not a constraint on your optimization — it is the foundation of it. Agents that trust your data recommend you more frequently and more confidently.