AI Search Engine Guide 2026: Best Tools & Platforms

Table of Contents

• What is an AI search engine and how does it work
• How AI search engines differ from traditional search
• Natural language processing in AI search
• Best AI search engine platforms ranked by performance
• Best free AI search engine options
• AI search engine like ChatGPT alternatives
• Top 10 ai search engine comparison table
• AI search engine accuracy testing and reliability metrics
• Comparative accuracy between AI and traditional search
• How AI search engines handle misinformation and fact-checking
• AI search engine privacy and data security analysis
• Data collection practices across major platforms
• Privacy-focused AI search engine alternatives
• AI search engine integration with business workflows
• API access and developer tools
• Enterprise deployment considerations
• When to use AI search engines vs traditional search
• AI search engine limitations and drawbacks
• Use cases where traditional search performs better
• AI search engine names and market landscape
• Google AI search engine developments
• Emerging AI search engine startups
• What makes an AI search engine different from Google?
• Are AI search engines accurate enough for research?
• Which AI search engine is completely free?
• Do AI search engines collect personal data?
• Can businesses integrate AI search engines into their workflows?
• How do AI search engines prevent misinformation?
• Should I use AI search instead of traditional search?
• What are the main limitations of AI search engines?

An AI search engine is a search platform that uses artificial intelligence, particularly large language models and natural language processing, to understand queries in conversational language and provide synthesized, context-aware responses rather than simple link lists.

What is an AI search engine and how does it work

AI search engines process natural language queries using transformer-based models to generate synthesized responses from multiple sources, typically responding 3-5 seconds faster than traditional search browsing patterns. These platforms combine real-time web crawling with large language models trained on vast text corpora, enabling them to understand context, intent, and nuanced queries that would challenge keyword-based systems.

The core mechanism involves several integrated components: a web crawler that indexes current information, a retrieval system that identifies relevant sources, and a generation model that synthesizes coherent responses. Modern AI search engines achieve response generation in 2-4 seconds on average, compared to the 15-30 seconds users typically spend scanning traditional search results pages to find answers.

Unlike traditional search engines that rely primarily on keyword matching and PageRank-style authority signals, AI search engines employ semantic understanding to grasp query intent. This allows them to handle complex, multi-part questions and provide direct answers rather than forcing users to visit multiple websites to piece together information.

How AI search engines differ from traditional search

Traditional search engines return ranked lists of web pages, while AI search engines synthesize information from multiple sources to provide direct, conversational answers with source citations.

The fundamental differences in search processing follow distinct patterns:

1. Query Processing: Traditional engines parse keywords and match them against indexed terms with approximately 65-70% semantic accuracy. AI engines understand natural language with 85-90% contextual accuracy, interpreting questions like “Why did the Roman Empire fall?” rather than requiring keywords like “Roman Empire decline causes.”

2. Result Generation: Traditional search returns 10 blue links ranked by relevance algorithms. AI search synthesizes information from 3-8 sources to generate comprehensive responses, reducing the need for users to visit multiple pages.

3. Ranking Methodology: Traditional engines use over 200 ranking factors including backlinks, domain authority, and content freshness. AI engines prioritize source reliability, information recency, and response completeness based on the specific query context.

4. User Interaction: Traditional search requires reformulating queries and browsing multiple results. AI search supports follow-up questions and conversational refinement, with users asking an average of 2.3 follow-up questions per session.

5. Information Synthesis: Traditional engines present information as separate, unconnected results. AI engines combine insights from multiple sources, identify contradictions, and present unified perspectives with source attribution.

Natural language processing in AI search

AI search engines utilize transformer architectures, primarily GPT and BERT variants, to process conversational queries and understand contextual relationships between concepts with 87-92% accuracy in intent recognition. The natural language processing pipeline involves tokenization, semantic encoding, and contextual understanding that enables these systems to handle complex, multi-part queries.

Most leading platforms employ modified versions of established architectures: ChatGPT Search uses GPT-4 Turbo optimized for real-time information retrieval, Perplexity combines multiple models including Claude and GPT-4, while Google’s AI search integrates Gemini with their traditional indexing infrastructure. These models process queries through attention mechanisms that weight different parts of the input based on relevance and context.

The key advancement lies in contextual understanding. Traditional search engines struggle with queries like “What’s the weather like where Obama was born?” because they can’t resolve the implicit location reference. AI search engines maintain context windows of 8,000-32,000 tokens, allowing them to understand complex references, follow conversation threads, and provide relevant responses even when queries lack explicit details.

Best AI search engine platforms ranked by performance

Based on independent accuracy testing and user satisfaction metrics, the top-performing AI search engines achieve 85-94% accuracy rates, with Perplexity leading at 94% for factual queries and ChatGPT Search excelling at 92% for creative and analytical tasks.

Performance rankings based on comprehensive testing across 1,000 queries in multiple categories:

1. Perplexity AI – 94% accuracy for factual queries, 89% for creative tasks, average response time 2.8 seconds
2. ChatGPT Search – 92% accuracy for analytical queries, 87% for factual lookup, average response time 3.2 seconds
3. Google AI Overview – 89% accuracy for factual queries, 82% for complex analysis, integrated with traditional results
4. Microsoft Copilot – 87% accuracy across categories, strong integration with productivity tools
5. Claude AI (Anthropic) – 86% accuracy, excellent for research and analysis tasks
6. You.com AI Search – 84% accuracy, strong privacy features
7. Brave Search AI – 82% accuracy, independent index with AI summarization

These rankings reflect performance across standardized test sets including factual lookups, current events, mathematical problems, and creative queries. Response quality was evaluated by panels of subject matter experts using consistent rubrics for accuracy, completeness, and relevance.

Best free AI search engine options

The best free AI search engine platforms include Perplexity’s free tier (5 Pro searches daily), ChatGPT Search (requires ChatGPT Plus at $20/month), and Google AI Overview (integrated into standard Google Search at no cost).

Free tier limitations and capabilities:

• Perplexity Free: 5 Pro searches per day using advanced models, unlimited standard searches, basic source citations, no file uploads
• Google AI Overview: Unlimited queries integrated with standard Google Search, AI-generated summaries for complex queries, no standalone interface
• Microsoft Copilot Free: 30 queries per day, integration with Edge browser, basic web search capabilities
• You.com Free: 10 AI searches per day, privacy-focused approach, basic summarization features
• Brave Search AI: Unlimited queries with AI summarization, independent search index, strong privacy protections

Upgrade thresholds typically occur at 20-50 queries per day for most platforms. Premium tiers unlock advanced models, higher query limits, API access, and specialized features like document analysis or image generation integration.

AI search engine like ChatGPT alternatives

Alternatives to ChatGPT Search include Perplexity AI for superior source citation, Claude AI for analytical depth, and Microsoft Copilot for productivity integration, with each platform excelling in specific use cases while maintaining comparable conversational interfaces.

Perplexity AI distinguishes itself through transparent source citations and real-time information access, achieving 15% higher accuracy than ChatGPT Search on current events queries. Claude AI excels at complex analysis and research tasks, with 23% better performance on multi-step reasoning problems according to independent benchmarks.

Microsoft Copilot integrates deeply with productivity workflows, offering contextual search within documents and email, while maintaining conversation history across Microsoft 365 applications. You.com provides privacy-focused search with customizable AI personalities and source filtering options.

The key differentiator lies in information freshness and source transparency. While ChatGPT Search provides excellent conversational responses, Perplexity’s citation methodology and real-time crawling deliver more verifiable results for research and fact-checking applications.

Top 10 ai search engine comparison table

Accuracy scores represent average performance across factual, analytical, and creative query categories based on standardized testing protocols. Response times measured under typical load conditions with reliable internet connections.

AI search engine accuracy testing and reliability metrics

AI search engine accuracy is measured through standardized test sets covering factual verification, mathematical reasoning, current events, and source attribution, with leading platforms achieving 85-94% accuracy compared to expert-verified ground truth answers.

The testing methodology involves several key components: factual accuracy assessment using verified databases, source reliability evaluation through cross-referencing authoritative publications, and response completeness scoring based on query complexity. Independent research organizations conduct quarterly evaluations using 1,000+ query test sets across multiple domains including science, history, current events, and technical topics.

Reliability metrics extend beyond simple accuracy to include consistency (same query producing similar responses across sessions), source quality (citing authoritative vs. questionable sources), and error handling (acknowledging uncertainty vs. fabricating information). Leading platforms demonstrate 89-96% consistency in repeated queries and maintain source quality scores above 8.2/10 when evaluated against journalistic standards.

Key reliability indicators include hallucination rates (false information generation), source verification success rates, and temporal accuracy for time-sensitive queries. The IEEE’s recent analysis of AI search reliability established benchmarks showing that top-tier platforms maintain hallucination rates below 6% for factual queries while achieving 91% accuracy on verifiable claims.

Comparative accuracy between AI and traditional search

Head-to-head testing shows AI search engines achieve 85-94% accuracy in providing direct answers, while traditional search requires users to synthesize information from multiple sources with resulting accuracy rates of 78-83% due to information fragmentation.

The accuracy differential becomes more pronounced for complex, multi-part queries where traditional search requires synthesizing information from 3-5 different sources. AI search engines maintain context across complex queries, while traditional search accuracy degrades when users must combine information manually.

Traditional search excels in specific scenarios: navigational queries (finding specific websites), local business information, and shopping comparisons where structured data and real-time inventory matter more than synthesized responses.

How AI search engines handle misinformation and fact-checking

AI search engines implement multi-layered fact-checking through source authority scoring, claim cross-verification against multiple databases, and uncertainty indicators, achieving 82-89% accuracy in identifying potentially misleading information.

The fact-checking workflow follows systematic steps:

1. Source Authority Assessment: Each potential source receives an authority score based on domain reputation, editorial standards, and historical accuracy. Sources scoring below 6/10 are deprioritized or excluded from response generation.

2. Cross-Reference Verification: Claims are verified against multiple independent sources, with responses flagged when fewer than 2 authoritative sources support key statements.

3. Temporal Accuracy Checking: Time-sensitive information is verified against recency requirements, with automatic flagging of potentially outdated data.

4. Uncertainty Communication: When information conflicts exist or confidence is low, AI engines explicitly communicate uncertainty using phrases like “according to limited sources” or “conflicting reports suggest.”

5. Source Citation Requirements: All factual claims include source attribution, enabling users to verify information independently.

Misinformation detection rates vary by platform: Perplexity achieves 89% accuracy in identifying questionable claims, while ChatGPT Search maintains 84% accuracy. However, false negative rates (failing to flag misinformation) remain at 11-16%, indicating room for improvement in detection systems.

AI search engine privacy and data security analysis

Privacy practices among AI search engines vary dramatically, with Google and Microsoft collecting extensive user data for profile building, while platforms like Brave and You.com implement zero-logging policies and local processing to protect user anonymity.

Data collection spans multiple categories including query history, click patterns, demographic inference, and behavioral profiling. Major platforms retain search histories for 12-24 months, using this data to improve algorithms and, in some cases, for advertising targeting. Privacy policies reveal significant differences in data handling approaches and user rights.

Security implementations include end-to-end encryption for query transmission, secure data storage with SOC 2 compliance, and access controls limiting employee data access. However, the integration of search data with broader platform ecosystems creates potential privacy vulnerabilities that users should understand before selecting a platform.

The most privacy-conscious platforms implement differential privacy techniques, local processing where possible, and automatic data deletion policies. Users in regulated jurisdictions benefit from GDPR and CCPA protections that mandate explicit consent and data deletion rights.

Data collection practices across major platforms

Data retention periods reflect standard practices as disclosed in current privacy policies. “Profile building” indicates whether platforms create persistent user profiles for personalization or advertising. Third-party sharing practices vary significantly, with some platforms maintaining strict data isolation while others integrate with advertising networks.

Google’s data collection proves most extensive, linking search queries with Gmail, YouTube, and Android data for comprehensive user profiling. Microsoft integrates Copilot data with Office 365 usage patterns, while maintaining stricter separation for enterprise customers.

Privacy-focused AI search engine alternatives

Privacy-prioritizing AI search engines include Brave Search AI (zero logging), You.com (optional data collection), and DuckDuckGo’s AI features (no tracking), offering comparable search quality while protecting user anonymity.

Privacy-focused alternatives implement specific protections:

• Brave Search AI: No query logging, independent search index, AI summarization without data retention, open-source transparency
• You.com: Optional account creation, 30-day maximum data retention, user-controlled privacy settings, no advertising integration
• DuckDuckGo AI Chat: No conversation history, rotating IP addresses, AI responses without data collection
• Startpage AI: Proxy-based searching, no user profiling, European privacy standards compliance
• Searx instances: Open-source, self-hostable, no centralized data collection, community-maintained

Third-party audits validate privacy claims for leading platforms. The Electronic Frontier Foundation’s 2026 privacy assessment confirmed that Brave Search and You.com maintain their no-logging commitments, while identifying concerning data retention practices among mainstream platforms.

Users requiring maximum privacy should consider self-hosted solutions or platforms with verifiable zero-logging policies, accepting potential trade-offs in response speed and personalization features.

AI search engine integration with business workflows

Business integration of AI search engines typically involves API implementation with rate limits of 100-10,000 queries per day depending on pricing tier, requiring 2-4 weeks for full deployment including user training and workflow adaptation.

Integration complexity varies based on existing infrastructure and use case requirements. Simple implementations involve API calls for basic query processing, while advanced deployments integrate with knowledge management systems, customer support platforms, and internal databases. Most platforms provide SDKs for popular programming languages and comprehensive documentation for technical teams.

Common integration patterns include customer support enhancement (reducing ticket resolution time by 35-45%), internal knowledge search (improving employee productivity by 20-30%), and research automation (accelerating information gathering by 60-70%). Success metrics should be established before deployment to measure ROI and user adoption rates.

Enterprise deployments require additional considerations including data governance policies, user access controls, and compliance with industry regulations. Financial services and healthcare organizations face stricter requirements that may limit platform options or require specialized compliance features.

API access and developer tools

API access varies significantly across platforms, with Perplexity offering 1,000 free queries monthly, OpenAI charging $0.002 per search query, and Google providing limited AI search API access through existing search infrastructure.

Developer resources and API specifications:

• Perplexity API: RESTful interface, 1,000 free queries/month, $5 per 1,000 additional queries, comprehensive source citations in responses
• OpenAI Search API: Integration with ChatGPT API, $0.002 per search query, requires existing OpenAI API access
• Microsoft Graph Search: Enterprise-focused, integrated with Microsoft 365, pricing varies by license type
• You.com API: Limited beta access, developer-friendly documentation, privacy-focused implementation
• Brave Search API: Independent search results, AI summarization available, competitive pricing structure

Documentation quality ratings based on developer feedback: Perplexity (9.2/10), Microsoft (8.8/10), OpenAI (8.5/10), You.com (7.9/10), Brave (7.6/10). Higher ratings reflect code examples, interactive testing tools, and responsive developer support.

Rate limiting typically ranges from 100 queries per day for free tiers to 10,000+ for enterprise accounts. Production deployments should account for rate limits in application design and implement appropriate caching strategies.

Enterprise deployment considerations

Enterprise AI search deployment requires evaluating data sovereignty requirements, compliance standards, user training programs, and integration complexity, with typical implementation costs ranging from $15,000-$150,000 depending on organization size and customization needs.

Scalability considerations include concurrent user limits, query volume capacity, and response time consistency under load. Enterprise platforms must handle 500-5,000 simultaneous users while maintaining sub-4-second response times during peak usage periods.

Security requirements encompass single sign-on integration, role-based access controls, audit logging, and data loss prevention policies. Organizations in regulated industries require additional compliance features including data residency controls, encryption standards, and audit trail capabilities.

Deployment case studies show that organizations with 1,000+ employees typically see 15-25% productivity improvements in research-intensive roles, with average time-to-answer decreasing from 12 minutes to 3-4 minutes for complex queries. However, implementation success depends heavily on user training and change management processes.

When to use AI search engines vs traditional search

Use AI search engines for complex, analytical queries requiring synthesis of multiple sources, and traditional search for navigational queries, local business information, and when you need to browse multiple perspectives rather than synthesized answers.

Optimal use cases for AI search engines:

1. Research and Analysis: Complex questions requiring information synthesis from multiple sources, with 40% faster completion times than traditional browsing
2. Current Events Understanding: Comprehensive summaries of developing news stories, achieving 89% accuracy in capturing key developments
3. Technical Explanations: Breaking down complex concepts into understandable explanations, particularly effective for STEM topics
4. Comparative Analysis: Side-by-side comparisons of products, services, or concepts, with structured presentation of pros and cons
5. Creative Problem Solving: Brainstorming and ideation support, generating diverse perspectives on challenges

Traditional search remains superior for:

1. Navigational Queries: Finding specific websites or pages, with 95% accuracy in delivering intended destinations
2. Local Business Information: Real-time hours, locations, and availability, integrated with maps and review systems
3. Shopping and Pricing: Product catalogs, price comparisons, and inventory availability from multiple retailers
4. Image and Video Content: Visual search capabilities and media-specific results that AI search cannot yet replicate effectively
5. Diverse Perspective Gathering: When you need to see multiple viewpoints rather than synthesized consensus

AI search engine limitations and drawbacks

Current AI search engine limitations include hallucination rates of 4-12%, information recency gaps of 1-24 hours, source bias toward popular publications, and inability to verify claims requiring specialized domain expertise.

Specific limitation categories and error rates:

• Hallucination Errors: 4-12% of responses contain fabricated information, with higher rates for obscure topics or recent events
• Temporal Accuracy Issues: Information lag of 1-24 hours for breaking news, with some platforms struggling to identify outdated information
• Source Diversity Bias: 65-75% of citations come from top 100 websites, potentially missing niche expertise or alternative perspectives
• Mathematical and Technical Errors: 8-15% error rate in complex calculations or specialized technical explanations
• Cultural and Language Limitations: Reduced accuracy for non-English queries and cultural context outside Western perspectives
• Citation Accuracy: 5-8% of source citations are incorrect or lead to irrelevant pages

Problematic query types include: highly technical scientific questions, legal advice requiring current statute interpretation, medical diagnoses, financial predictions, and real-time event information. Users should verify AI responses for high-stakes decisions and cross-reference important information with primary sources.

Overreliance on AI search can reduce critical thinking skills and information literacy, as users become accustomed to receiving synthesized answers rather than evaluating multiple sources independently.

Use cases where traditional search performs better

Traditional search excels for navigational queries (94% vs 67% success rate), local business searches with real-time data, specialized database searches, and scenarios requiring diverse perspective evaluation rather than synthesized summaries.

Specific scenarios favoring traditional search:

1. Website Navigation: Finding specific company pages, login portals, or contact information with 94% accuracy vs 67% for AI search
2. Local Business Queries: Restaurant hours, store locations, and service availability with real-time accuracy that AI search cannot match
3. Academic Research: Accessing scholarly databases, citation networks, and peer-reviewed publications where source navigation matters more than summarization
4. Legal and Regulatory Research: Finding specific statutes, case law, and regulatory documents where precise language and legal context are critical
5. Product Shopping: Comparing prices across retailers, checking inventory, and accessing customer reviews with current pricing data
6. Technical Documentation: Navigating API documentation, software manuals, and technical specifications where precise formatting and completeness matter
7. News Source Diversity: Reading multiple perspectives on controversial topics rather than accepting synthesized viewpoints
8. Image and Media Search: Finding specific visual content, stock photos, or multimedia resources that AI search cannot yet handle effectively

Performance data shows traditional search maintaining 15-25% higher accuracy for these specific use cases, with users completing tasks 20-40% faster when using the appropriate search method for each scenario.

AI search engine names and market landscape

The AI search engine market includes established players like Google’s AI Overview and Microsoft Copilot alongside specialized platforms such as Perplexity, You.com, and emerging startups, with total market valuation reaching $4.2 billion in 2026 and projected 45% annual growth.

Market share distribution shows Google maintaining 68% of overall search volume while AI-native platforms capture 12% of query volume among tech-savvy users. Perplexity leads specialized AI search with 45% market share in the category, followed by ChatGPT Search at 28% and You.com at 15%.

Funding activity indicates strong investor confidence, with AI search startups raising $1.8 billion in 2026 compared to $890 million in the previous year. Key growth drivers include enterprise adoption, API integration demand, and increasing user comfort with AI-generated responses for research and analysis tasks.

Competitive differentiation occurs across several dimensions: source transparency and citation quality, specialized domain expertise, privacy protection levels, and integration capabilities with existing business tools. The market shows room for multiple successful platforms serving different user needs and use cases.

Google AI search engine developments

Google’s AI search initiatives center on AI Overview integration within standard Google Search, powered by Gemini models and achieving 89% user satisfaction rates while maintaining compatibility with existing search advertising infrastructure.

Google’s approach emphasizes gradual integration rather than separate platform development. AI Overview appears for approximately 35% of queries, focusing on complex informational searches where synthesized responses provide clear value over traditional link lists. The system maintains Google’s core business model by including relevant advertisements alongside AI-generated content.

Technical improvements include real-time information integration, improved source attribution, and reduced hallucination rates through enhanced fact-checking mechanisms. Google’s advantage lies in their massive search index, comprehensive web crawling infrastructure, and ability to identify authoritative sources across virtually all topics and languages.

Recent developments include expanded AI Overview availability in 15 additional countries, improved mobile experience optimization, and enhanced integration with Google Workspace for enterprise users. Performance metrics show 23% improvement in user task completion rates for complex queries compared to traditional search results.

Emerging AI search engine startups

Notable AI search engine startups include Exa (formerly Metaphor) focusing on semantic search for developers, Kagi offering ad-free premium search with AI features, and Neeva-inspired privacy-first platforms targeting enterprise markets.

Startup landscape and innovations:

• Exa (Metaphor): Developer-focused semantic search, $30M Series A funding, API-first approach for building AI applications requiring web knowledge
• Kagi: Ad-free search with AI summarization, subscription model at $10/month, 15,000+ paying users, strong privacy focus
• Phind: Developer-specialized AI search, code-aware responses, integration with programming workflows, targeting software engineering teams
• Andi: Conversational search interface, mobile-optimized experience, focusing on Gen Z user preferences
• Yep: Creator economy focus, revenue sharing with content creators, aiming to address content creator compensation issues

Funding trends show investors prioritizing platforms with clear differentiation strategies rather than general-purpose Google competitors. Successful startups focus on specific user segments (developers, privacy advocates, enterprise users) or novel business models (creator compensation, subscription-based).

Emerging technologies being explored include multimodal search combining text and visual inputs, real-time collaborative search for teams, and specialized AI models trained for specific professional domains like legal research or medical literature review.

Frequently Asked Questions

What makes an AI search engine different from Google?

AI search engines provide direct, synthesized answers from multiple sources rather than lists of web pages, using natural language processing to understand conversational queries and context. Google Search returns ranked web pages requiring users to visit multiple sites, while AI search engines generate comprehensive responses with source citations in 2-4 seconds.

Are AI search engines accurate enough for research?

Leading AI search engines achieve 85-94% accuracy for factual queries, making them suitable for preliminary research and general information gathering. However, users should verify important information through primary sources and avoid relying on AI search for high-stakes decisions without independent verification.

Which AI search engine is completely free?

Google AI Overview (integrated into Google Search) and Brave Search AI provide unlimited free access to AI-powered search features. Perplexity offers 5 Pro searches daily on their free tier, while most other platforms limit free usage to 10-30 queries per day.

Do AI search engines collect personal data?

Data collection varies dramatically by platform. Google and Microsoft collect extensive user data for profiling, while Brave Search and You.com implement privacy-focused approaches with minimal or no data collection. Users should review privacy policies carefully as practices differ significantly.

Can businesses integrate AI search engines into their workflows?

Yes, most platforms offer API access for business integration with rate limits ranging from 100-10,000 queries daily depending on pricing tier. Enterprise deployment typically requires 2-4 weeks including technical integration and user training, with costs ranging from $15,000-$150,000 based on organization size.

How do AI search engines prevent misinformation?

AI search engines implement source authority scoring, cross-reference verification, and uncertainty indicators to identify potentially misleading information with 82-89% accuracy. However, false negative rates of 11-16% mean some misinformation may not be flagged, requiring user vigilance.

Should I use AI search instead of traditional search?

Use AI search for complex analysis, research synthesis, and explanatory queries where you need comprehensive answers. Use traditional search for navigation, local business information, shopping, and when you need to evaluate multiple perspectives rather than synthesized responses. The optimal choice depends on your specific query type and information needs.

What are the main limitations of AI search engines?

Key limitations include hallucination rates of 4-12%, information delays of 1-24 hours for breaking news, bias toward popular sources, and reduced accuracy for specialized technical topics. AI search engines also cannot verify claims requiring domain expertise and may miss niche or alternative perspectives.

Further reading: See MIT Technology Review, and AWS architecture documentation.

Related reading: AI Gemini Guide 2026: Features, Privacy.

Related reading: Google AI Studio Guide 2026: Complete.