# Classifysite

*/Startups/Classifysite*

## Startup Overview

This classification engine evaluates live web properties against custom brand-safety taxonomies. Programmatic buyers and ad networks submit URLs, and the system returns contextual classifications based on user-defined parameters. It analyzes page text, metadata, and structural signals to map content directly to proprietary risk models.

Digital advertisers and ad exchanges require strict control over where their placements appear, but standard brand-safety tools enforce rigid, universal blocklists. When media buyers encounter nuanced content, they either over-block safe inventory or accidentally bid on misaligned placements. Relying on pre-defined industry categories fails to capture distinct, brand-specific safety thresholds.

Unlike Oracle Grapeshot or DoubleVerify, which rely on static segments, or slow in-house NLP scrapers, this platform operates with sub-millisecond latency. It executes safety checks fast enough for the real-time bidding environment while allowing users to enforce fully proprietary content taxonomies. This architecture replaces generalized risk scores with precise, instant context mapping tailored to the exact requirements of individual media buyers.

## Startup Founding Hypothesis

**Approach**: that categorizes live web properties against custom brand-safety taxonomies
**Competitors**:
- [Oracle Grapeshot](/Competitors/Oracle_Grapeshot)
- [DoubleVerify](/Competitors/DoubleVerify)
- [in-house NLP scrapers](/Competitors/in-house_NLP_scrapers)
**Differentiator2x2**: sub-millisecond latency and fully customizable to proprietary content taxonomies

## Startup Solution Coordinate

**Solution**: [Contextual Classification Engine](/Software/Contextual_Classification_Engine)

## Startup Position2x2

```mermaid
quadrantChart
    title Brand Safety Classification Positioning
    x-axis Rigid Taxonomies --> Custom Taxonomies
    y-axis High Latency --> Sub-millisecond Latency
    quadrant-1 Custom & Real-time
    quadrant-2 Standard & Real-time
    quadrant-3 Standard & Batch
    quadrant-4 Custom & Batch
    Oracle Grapeshot: [0.3, 0.65]
    DoubleVerify: [0.15, 0.85]
    In-house NLP Scrapers: [0.85, 0.2]
    Classifysite: [0.9, 0.9]
```

## Startup Offer

**Proof**:
- Targeting ad networks aiming to classify 100M daily impressions without slowing bid responses.
- Aiming for DSPs seeking to replace static blocklists with dynamic, brand-specific taxonomies.
- Designed for performance marketers targeting a 30% reduction in false-positive content blocks.
**Tiers**:
- Name: Standard Taxonomy · Price: ~$0.05–$0.10 per 1,000 lookups · Inclusions: API access for up to 50M domain classifications per month against standard IAB brand-safety categories.
- Name: Custom Taxonomy · Price: ~$0.15–$0.25 per 1,000 lookups · Inclusions: Mapping to proprietary brand-safety rubrics with custom weighting, up to 250M lookups per month.
- Name: Programmatic Edge · Price: Enterprise: ~$4k–$10k/mo flat rate · Inclusions: Unlimited taxonomy lookups deployed to edge networks for sub-millisecond real-time bidding availability.
**Guarantee**: If API response times for cached domain lookups exceed 5 milliseconds during a billing period, the customer receives a 50% usage credit for that month's classification volume.
**Business Function**: ProvideService
**Objection Handlers**:
- We already use DoubleVerify; why add another layer? DoubleVerify enforces rigid, global categories; Classifysite enforces your proprietary risk taxonomy without latency penalties.
- Deep classification adds too much latency for RTB. Our architecture is designed to cache category mappings at the edge, guaranteeing sub-millisecond lookup times for live bid streams.
- How do you handle newly registered domains? The system is designed to route unknown URLs through an asynchronous NLP pipeline, categorizing and caching them within seconds of the first bid request.
**Pricing Architecture**: UsageMeter
**Agent Checkout Support**:
- agentic-commerce-protocol

## Startup Brand

**Voice**: Authoritative technical register with an emphasis on low-latency precision.
**Tagline**: Enforce custom brand-safety taxonomies with sub-millisecond latency.
**Icon Concept**: prism
**Palette Intent**: electric-signal
**Visual Identity**: A high-contrast aesthetic relying on neon green and stark black to evoke high-speed data exchanges and programmatic bid streams.
**Archetype Reference**: the-sage

## Startup Buyer Chain

**Chain**: B2B → Demand-Side Platforms (DSPs) → Brand Advertisers
**Gtm Motion**: Direct technical sales to engineering leaders at Demand-Side Platforms (DSPs) by demonstrating sub-millisecond latency reductions in the bid stream compared to legacy systems. Lands with a pilot executing a single brand's custom taxonomy, then expands via API volume tiering as the DSP routes more programmatic traffic through the classification engine.
**Agent Channel**: Intended for listing in programmatic AI capability registries and autonomous agent toolkits (like LangChain) as a brand safety URL classifier for media-buying agents to query during live bid evaluation.
**Primary Channel**: Technical searches for "low latency contextual classification API" and direct outbound targeting ad-tech infrastructure engineers optimizing bid stream latency within OpenRTB ecosystems.

## Startup Customer Journey

```mermaid
flowchart LR A[Infrastructure Engineer] --> B[API Documentation] B --> C[Latency Benchmark] C --> D[Brand Taxonomy] D --> E[Bid Stream] E --> F[Edge Network] F --> G[Case Study]
```

## Startup Proof Points

_Illustrative — target and order-of-magnitude estimate figures, not an achieved track record (this Thing is concept-stage)._

**Pilot Goals**:
- 14-day shadow bidding pilot evaluating 50M daily impressions: Target result is confirming sub-millisecond edge cache hits for domain lookups without disrupting the existing bid stream.
- 30-day side-by-side inventory analysis against a legacy brand-safety provider: Target result is quantifying the exact volume of brand-safe ad slots previously discarded as false positives.
**Target Metrics**:
- Target: <5 milliseconds average lookup latency for cached domain categories during real-time bidding.
- Aim: 30% reduction in false-positive brand safety content blocks compared to standard global taxonomies.
- Target: <10 seconds average asynchronous NLP categorization time for newly registered, unknown domains.
**Target Case Studies**:
- Mid-sized Demand-Side Platform (DSP) Product Manager: Replacing legacy static blocklists with dynamic, brand-specific taxonomies to reclaim falsely flagged inventory without compromising brand safety.
- Global Ad Network Head of Infrastructure: Deploying edge-based domain classification to evaluate 100M daily impressions while strictly maintaining sub-millisecond bid response times.
**Testimonial Targets**:
- AdTech Chief Technology Officer: Seeking validation that the edge-deployed taxonomy architecture completely removes the RTB latency penalty associated with deep domain classification.
- VP of Programmatic Advertising: Looking for confirmation that custom risk rubrics can be applied per-campaign without requiring manual blocklist updates or slowing down bid velocity.
- Performance Marketing Director: Aiming to hear that accurate custom categorizations opened up high-converting long-tail inventory previously blocked by rigid third-party vendors.

## Startup Top Risks

**Risks**:
- Severity: existential · Description: Ad-tech incumbents DoubleVerify or Oracle bundle custom brand-safety taxonomies into their existing enterprise suites, eliminating the need for a standalone vendor. · Mitigation Status: unmitigated
- Severity: high · Description: Processing volume spikes degrade the sub-millisecond latency guarantee, destroying the primary technical differentiator against incumbent solutions. · Mitigation Status: in-progress
- Severity: high · Description: Aggressive bot-protection measures by major publishers and CDNs block the classification crawlers from reading live web properties. · Mitigation Status: in-progress
- Severity: moderate · Description: Calibrating custom proprietary taxonomies for new enterprise clients requires extensive manual tuning, making sales cycles unprofitable. · Mitigation Status: unmitigated

## Startup Competitors

- [Oracle Grapeshot](/Competitors/Oracle_Grapeshot) — Incumbent
- [DoubleVerify](/Competitors/DoubleVerify) — Incumbent
- [In-House NLP Scrapers](/Competitors/In-House_NLP_Scrapers) — Status Quo
- [Integral Ad Science](/Competitors/Integral_Ad_Science) — Incumbent
- [Peer39 Contextual](/Competitors/Peer39_Contextual) — AdTech Platform

## Startup Solution Stack

- [Brand Safety Evaluation Service](/Services/Brand_Safety_Evaluation_Service) — Service-as-Software
- [Context Analysis Agent](/Agents/Context_Analysis_Agent) — Agent
- [Taxonomy Mapping Agent](/Agents/Taxonomy_Mapping_Agent) — Agent
- [Live Classification API](/Software/Live_Classification_API) — Software
- [Sub-Millisecond Scoring Engine](/Software/Sub-Millisecond_Scoring_Engine) — Software

## Startup Story Brand

**Hero**:
- **Need**: to be the performance leader delivering brand-safe reach without sacrificing bid-stream speed
- **Want**: to enforce proprietary brand-safety rubrics during real-time bidding
- **Identity**: the head of platform at a programmatic demand-side platform
**Plan**:
- Step: Map Taxonomy · Detail: Define your proprietary brand-safety rubrics or use standard IAB categories to set your protection boundaries.
- Step: Verify Performance · Detail: Run a test stream through our API to confirm sub-millisecond lookups for your active domain list.
- Step: Secure Stream · Detail: Deploy the edge integration to filter bid requests in real-time, reducing false-positive content blocks by 30%.
**Guide**:
- **Empathy**: Does your bidding process still drop high-quality impressions because of outdated static blocklists?
**Problem**:
- **Villain**: rigid global taxonomies
- **External**: Oracle Grapeshot and DoubleVerify force rigid content categories that cause high false-positives and missed impressions in the RTB stream
- **Internal**: you feel forced to choose between brand safety and winning the auction
- **Philosophical**: Content classification was built for brand protection, not as a tax on auction latency.
**Success**: Your DSP scales to 100M daily impressions with dynamic, brand-specific taxonomies enforced at the edge with zero latency penalty.
**One Liner**: Every auction, DSPs lose reach to rigid blocklists. Classifysite enforces custom taxonomies at the edge so performance marketers recapture safe inventory without adding latency.
**Positioning**:
- **So That**: enforce proprietary risk rubrics without sacrificing sub-millisecond bid response times
- **Unlike**: Oracle Grapeshot and DoubleVerify
- **For Whom**: heads of platform at DSPs and ad networks
- **Category**: Edge-based content classification for programmatic advertising
**Call To Action**:
- **Direct**: Provision API Key
- **Transitional**: Download Taxonomy Schema
**Failure Stakes**:
- Losing 30% of reachable inventory to false-positives
- Auction timeouts due to slow classification lookups
- Brand-safety breaches on newly registered domains
**Transformation**:
- **To**: the platform leader who enables precision-scale bidding
- **From**: a DSP engineer managing static CSV blocklists
**Controlling Idea**: Proprietary brand-safety taxonomies must execute at the speed of the bid stream.

## Startup Token Hero

**Genre**: founding-hypothesis
**Rendered**: Every auction, DSPs lose reach to rigid blocklists. Classifysite enforces custom taxonomies at the edge so performance marketers recapture safe inventory without adding latency.
**Mechanism**: spine-derived-v1
**Template Id**: spine-founding-hypothesis
**Vocab Fingerprint**: 7c3e12c2d2c3c195

## Startup Token Positioning

**Genre**: moore-positioning
**Rendered**: Edge-based content classification for programmatic advertising for heads of platform at DSPs and ad networks. Unlike Oracle Grapeshot and DoubleVerify — enforce proprietary risk rubrics without sacrificing sub-millisecond bid response times.
**Mechanism**: spine-derived-v1
**Template Id**: spine-moore-positioning
**Vocab Fingerprint**: 7692ae7cb14e7b53

## Startup Token Pitch Deck

**Genre**: pitch-deck
**Rendered**: Problem: Oracle Grapeshot and DoubleVerify force rigid content categories that cause high false-positives and missed impressions in the RTB stream
Solution: Every auction, DSPs lose reach to rigid blocklists. Classifysite enforces custom taxonomies at the edge so performance marketers recapture safe inventory without adding latency.
Customer: heads of platform at DSPs and ad networks
Unlike: Oracle Grapeshot and DoubleVerify
**Mechanism**: spine-derived-v1
**Template Id**: spine-pitch-deck
**Vocab Fingerprint**: 90a3b4a8014ffe1c

## Startup Token M E D D P I C C

**Pain**: Oracle Grapeshot and DoubleVerify force rigid content categories that cause high false-positives and missed impressions in the RTB stream
**Metrics**: Target: Your DSP scales to 100M daily impressions with dynamic, brand-specific taxonomies enforced at the edge with zero latency penalty.
**Rendered**: Pain: Oracle Grapeshot and DoubleVerify force rigid content categories that cause high false-positives and missed impressions in the RTB stream
Economic buyer: Demand-Side Platforms
Metrics: Target: Your DSP scales to 100M daily impressions with dynamic, brand-specific taxonomies enforced at the edge with zero latency penalty.
Competition: Oracle Grapeshot and DoubleVerify
**Mechanism**: spine-derived-v1
**Competition**: Oracle Grapeshot and DoubleVerify
**Economic Buyer**: Demand-Side Platforms
**Vocab Fingerprint**: d9e69817327bdd01

## Startup Token Cold Email

**Genre**: cold-email
**Rendered**: Subject: Edge-based content classification for programmatic advertising for heads of platform at DSPs and ad networks

heads of platform at DSPs and ad networks — Oracle Grapeshot and DoubleVerify force rigid content categories that cause high false-positives and missed impressions in the RTB stream Every auction, DSPs lose reach to rigid blocklists. Classifysite enforces custom taxonomies at the edge so performance marketers recapture safe inventory without adding latency.
**Mechanism**: spine-derived-v1
**Template Id**: spine-cold-email
**Vocab Fingerprint**: faf64388ad01f0cc

## Startup Token Agent Spec

**Genre**: ai-agent-spec
**Rendered**: Edge-based content classification for programmatic advertising. Every auction, DSPs lose reach to rigid blocklists. Classifysite enforces custom taxonomies at the edge so performance marketers recapture safe inventory without adding latency. Serves heads of platform at DSPs and ad networks.
**Mechanism**: spine-derived-v1
**Template Id**: spine-ai-agent-spec
**Vocab Fingerprint**: 524a104ad5f320bd

## Neighborhood

### Candidate solutions

- [Manifest Document Parsing](/Problems/Manifest_Document_Parsing) — candidate solution for · Problems

### What it offers

- [Contextual Classification Engine](/Software/Contextual_Classification_Engine) — offers · Software

### Composed of

- [Taxonomy Mapping Agent](/Agents/Taxonomy_Mapping_Agent) — composes · Agents
- [Live Classification API](/Software/Live_Classification_API) — composes · Software
- [Context Analysis Agent](/Agents/Context_Analysis_Agent) — composes · Agents
- [Brand Safety Evaluation Service](/Services/Brand_Safety_Evaluation_Service) — composes · Services
- [Sub-Millisecond Scoring Engine](/Software/Sub-Millisecond_Scoring_Engine) — composes · Software

### Embodies

- [Software](/Theses/Software) — embodies · Theses

### Competitors

- [Peer39 Contextual](/Competitors/Peer39_Contextual) — competes with · Competitors
- [DoubleVerify](/Competitors/DoubleVerify) — competes with · Competitors
- [In-House NLP Scrapers](/Competitors/In-House_NLP_Scrapers) — competes with · Competitors
- [Integral Ad Science](/Competitors/Integral_Ad_Science) — competes with · Competitors
- [Oracle Grapeshot](/Competitors/Oracle_Grapeshot) — competes with · Competitors

### Similar Startups

- [Lattodestone](/Startups/Lattodestone) — similar · Startups
- [Abhorrent](/Startups/Abhorrent) — similar · Startups
- [Maren](/Startups/Maren) — similar · Startups
- [Aururn](/Startups/Aururn) — similar · Startups
- [Ambyield](/Startups/Ambyield) — similar · Startups
- [Unoct](/Startups/Unoct) — similar · Startups
- [Hinder](/Startups/Hinder) — similar · Startups
- [Lumica](/Startups/Lumica) — similar · Startups
- [Dealortage](/Startups/Dealortage) — similar · Startups
- [Advetection](/Startups/Advetection) — similar · Startups
- [Pragging](/Startups/Pragging) — similar · Startups
- [Gressil](/Startups/Gressil) — similar · Startups
- [Gatortage](/Startups/Gatortage) — similar · Startups
- [Abhorrible](/Startups/Abhorrible) — similar · Startups
- [Curationdynamic](/Startups/Curationdynamic) — similar · Startups
- [Sociphan](/Startups/Sociphan) — similar · Startups
- [Adjadge](/Startups/Adjadge) — similar · Startups
- [Claimeadline](/Startups/Claimeadline) — similar · Startups
- [Adjortage](/Startups/Adjortage) — similar · Startups
- [Clearancestamp](/Startups/Clearancestamp) — similar · Startups