๐ผVentureBeatโขStalecollected in 2m
Sakana AI launches 'ultra deep research' agent for enterprises
๐กFirst 'Virtual CSO' agent that trades speed for 8-hour deep reasoning to produce 100-page enterprise strategy reports.
โก 30-Second TL;DR
What Changed
Marlin functions as an autonomous 'Virtual CSO' for corporations and think tanks.
Why It Matters
This signals a shift in enterprise AI from rapid, shallow text generation to methodical, long-horizon reasoning, potentially disrupting traditional strategy consulting workflows.
What To Do Next
Evaluate whether your current research-heavy workflows can be automated by testing Marlin's long-horizon reasoning capabilities for your next strategic report.
Who should care:Enterprise & Security Teams
๐ง Deep Insight
Web-grounded analysis with 25 cited sources.
๐ Enhanced Key Takeaways
- โขSakana AI's foundational philosophy is rooted in nature-inspired intelligence, emphasizing collective intelligence and evolutionary model merging as an alternative to developing single, massive AI models.
- โขMarlin integrates Sakana AI's proprietary research, including Adaptive Branching Monte Carlo Tree Search (AB-MCTS) โ recognized with a spotlight award at NeurIPS 2025 โ and a workflow automation framework derived from their 'AI Scientist' project, which demonstrated fully autonomous scientific discovery.
- โขPrior to its official launch, Marlin underwent a closed beta phase starting in April 2026, involving approximately 300 professionals across diverse industries to refine its research quality and long-duration operational stability.
- โขSakana AI has secured substantial funding, including a $30 million seed round, a $200 million Series A in 2024, and a ยฅ20 billion (approximately $135 million USD) Series B round in November 2025, which valued the company at around $2.6 billion.
- โขBeyond Marlin, Sakana AI's product portfolio includes Sakana Chat and Sakana Fugu, and the company is actively developing an 'AI Scientist' model aimed at automating the entire scientific research process.
๐ Competitor Analysisโธ Show
| Feature / Product | Sakana AI Marlin | OpenAI's Deep Research (ChatGPT agent) | Relevance AI | Lyzr AI |
|---|---|---|---|---|
| Core Function | Autonomous 'Virtual CSO' for long-horizon strategic research (up to 8 hours, 100-page reports) | General autonomous research assistant for web search and structured reports (15-30 min) | Orchestration platform for multi-agent systems (e.g., competitive intelligence) | Enterprise-grade framework for secure, private OGI and competitive monitoring |
| Key Differentiator | 'Ultra deep research' with continuous, long-horizon reasoning; nature-inspired AI foundation (AB-MCTS, AI Scientist) | Transforms ChatGPT into a research agent; desktop integration, general task completion | No-code Agent Studio for custom multi-agent workflows; supports various LLMs | Focus on privacy and OGI; deployable within secure cloud/on-premise infrastructure |
| Output | Comprehensive strategy reports, executive slides, verified citations | Structured reports with citations, synthesized findings | Daily briefs, synthesized data (customizable) | Continuous monitoring reports, actionable insights |
| Pricing Model | Pay-per-use service (Pro, Team, Enterprise tiers) | $20/month for Plus subscribers | Not specified (platform-based) | Not specified (enterprise solution) |
| Benchmarks / Speed | Weeks of work compressed into 8 hours | 15-30 minutes for comprehensive topics | Real-time monitoring | Continuous monitoring |
๐ ๏ธ Technical Deep Dive
- Marlin's core architecture integrates multi-model coordination optimization and advanced long-horizon reasoning technologies.
- It leverages the Adaptive Branching Monte Carlo Tree Search (AB-MCTS) technique, which frames reasoning as a tree-search problem, enabling the AI to autonomously evaluate, pursue, or discard hypotheses.
- The agent incorporates a workflow automation framework developed from Sakana AI's 'AI Scientist' project, which demonstrated fully autonomous scientific discovery from ideation to peer review.
- Marlin is designed to execute hundreds to thousands of Large Language Model (LLM) queries per session, dynamically allocating computational resources to promising research avenues.
- Sakana AI's broader research emphasizes nature-inspired intelligence, including 'Evolutionary Model Merge' (e.g., Model Merging of Natural Niches - M2N2), which 'breeds' new models by combining strengths of existing ones without extensive retraining.
- The company also develops 'Continuous Thought Machines' (CTM), AI models that utilize the synchronization of neuron activity as a reasoning mechanism, drawing inspiration from biological neural networks.
- Sakana AI is actively researching Recursive Self-Improvement (RSI) technology, aiming for AI systems that can autonomously redesign and improve their own development processes.
๐ฎ Future ImplicationsAI analysis grounded in cited sources
AI agents like Marlin will significantly reduce the time and cost associated with strategic market research for enterprises.
By automating tasks that traditionally take weeks into a matter of hours, these agents enable faster decision-making cycles and more efficient allocation of human resources within corporations.
The 'nature-inspired' approach to AI development, championed by Sakana AI, will gain increasing prominence as a sustainable alternative to brute-force scaling of large models.
This methodology offers a more resource-efficient and adaptive path to developing capable AI, particularly beneficial for specialized applications and regions with computational constraints.
The widespread adoption of autonomous AI agents will necessitate the development of new regulatory frameworks and ethical guidelines for AI oversight and accountability in enterprise decision-making.
The self-contained, long-horizon operation of agents like Marlin, which produce comprehensive reports without continuous human intervention, raises critical questions about potential biases, error detection, and the ultimate responsibility for strategic outcomes.
โณ Timeline
2023-07
Sakana AI founded by David Ha, Llion Jones, and Ren Ito in Tokyo.
2024-01
Sakana AI develops a method to build new AI models by 'breeding' existing models and raises $30M in seed funding.
2024-08
Sakana AI publishes 'The AI Scientist' report, outlining a fully AI-driven system for automated scientific discovery.
2024-09
Sakana AI announces Series A funding round, raising approximately $200M.
2025-05
Sakana AI releases the Continuous Thought Machine (CTM), an AI model inspired by biological neural networks.
2025-11
Sakana AI secures ยฅ20 billion (USD 135 million) in Series B funding, valuing the company at approximately $2.6 billion.
2026-04
Sakana Marlin launches in closed beta with approximately 300 professionals.
2026-06
Sakana AI officially launches Marlin, its first commercial product.
๐ Sources (25)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
๐ฐ
Weekly AI Recap
Read this week's curated digest of top AI events โ
๐Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: VentureBeat โ