About Seme
Founded 2025 · Updated May 22, 2026
Mission
Seme's mission is to democratize identity investigation. We provide AI-powered investigation capabilities — previously available only to large institutions — to security teams, journalists, HR professionals, and compliance officers. By integrating multi-round AI search, cross-validation, and structured reporting into a single workflow, Seme reduces traditional background investigations from days or weeks to 10-30 minutes.
We believe transparency and auditability are the cornerstones of identity investigation. Every investigated fact is tagged with source and evidence level (E1-E5), trust scores are calculated based on source quality and cross-validation results, and users can trace every finding back to its original source.
Platform Statistics
Core Features
Face Search
Upload a face photo. AI uses deep learning CNN models (FaceNet, ArcFace) to extract 128-512 dimensional embedding vectors, achieving 99.5%+ accuracy on LFW benchmark. Match triggers automatic deep research investigation.
Semantic Search
Describe your target in natural language (e.g., "all researchers who worked at Tencent AI Lab"). AI interprets intent, searches the internet, identifies matching individuals and ranks candidates.
Deep Research
Provide name and known background. AI runs 3-4 multi-angle investigation rounds, cross-validates across 15-20 sources, generates structured dossier with trust score.
Platform Architecture
Timeline
Seme project started, core investigation engine development
Face Search and Deep Research features launched
Semantic Search launched, multi-language support added
Agent API released, Methodology and Resources center launched
Technology Stack
Use Cases
Talent Recruitment
Cross-validate candidate education, work history, and qualifications across multiple sources.
Corporate Security
Large-scale identity verification and access control with biometric matching and continuous monitoring.
Due Diligence
In-depth background investigation for investment and compliance with source-cited evidence classification.
Network Analysis
Cross-organizational relationship and influence mapping using knowledge graphs to visualize entity connections.
Research Methodology
Seme's investigation methodology is based on OSINT best practices, combining AI automation with structured analysis. Each investigation follows a five-step process: Research Planning (multi-angle investigation design), Multi-round Search (parallel search across 15-20 sources), Gap Analysis (automatic coverage evaluation and follow-up query generation), Cross-validation (fact verification across independent sources), and Report Synthesis (structured dossier generation). The system uses the E1-E5 five-tier evidence classification to evaluate the credibility of each finding, with trust scores calculated from three core factors: source quality, cross-validation rate, and information completeness.
During data collection, the system queries multiple search agents in parallel, including the SearXNG meta-search engine (aggregating 70+ search sources), OpenAI Web Search API, and specialized academic and social media data sources. Query results are deduplicated, ranked, and cross-validated at the aggregation layer. The system uses vector embeddings for semantic similarity matching, combined with cosine similarity algorithms for precision. For identity verification, the system additionally uses biometric feature comparison and timeline consistency checks. The entire process completes in 10-30 minutes, covering an average of 15-20 independent sources.
Industries Served
Core Principles
Transparency & Auditability
Every investigated fact is tagged with source and evidence level (E1-E5). Users can trace every finding to its original source. Trust scores are calculated based on source quality and cross-validation, ensuring reproducible results.
Privacy & Compliance
Seme uses only publicly available information (OSINT) — no private databases or restricted resources. All investigations follow data protection regulations, with encrypted user data storage and data deletion request support.
AI-Augmented, Not AI-Replaced
Seme uses AI to accelerate investigation, but final judgment stays with the user. AI provides structured data and recommendations; users decide how to interpret and use results. Every AI-generated conclusion has traceable sources.
Continuous Improvement
The investigation engine continuously learns and improves. By analyzing investigation accuracy and user feedback, we constantly refine search strategies, verification algorithms, and report quality.
Frequently Asked Questions
What is Seme?▾
Seme is an AI-powered identity investigation platform offering Face Search, Semantic Search, and Deep Research. Given any lead — a face photo, name with background, or natural language criteria — Seme finds specific individuals and generates structured investigation dossiers.
What technology does Seme use?▾
Seme is built on Next.js 16, uses OpenAI GPT-4 for multi-round investigation, SearXNG for web search, deployed on Cloudflare Workers edge network.
What is Seme's mission?▾
To democratize identity investigation — providing AI-powered investigation capabilities to security teams, journalists, and compliance officers.
Is Seme's data sourcing secure?▾
Seme uses only publicly available information (OSINT principles) — no private databases. All data transfers use TLS 1.3 encryption, and the platform is GDPR compliant.
Can Seme integrate with existing systems?▾
Yes, Seme provides a complete REST API and Agent workflow interface, supporting integration with HR systems, compliance tools, CRM platforms, and custom applications. Enterprise also provides Webhook callbacks and custom integration support.