20-test subset result
SWE-bench Pro
20 of 20 tests resolved on initial attempts using the Taijitu AI Agent & Model.
Evaluation details
Intelligence,
made useful.
Dependable results at a cost that makes sense.
Taijitu AI builds autonomous intelligence systems, develops methods to make existing AI models more reliable and efficient, and engineers digital products for real-world use.
Talk to Taijitu AIFlagship Product
AOCA®
Real-time insights.
Reliable automation.
Autonomous intelligence that works with continuous data, investigates changing conditions, takes action, verifies outcomes, and adapts.
Built for high-volume time-series data and goal-directed operation: connecting signals with context and carrying work through to verified outcomes.
Explore AOCAProcessing throughput
- Frames per second
- 10,000
- Time series per frame
- 110
Workload-specific throughput. Hardware, test duration, processing scope, and latency are not yet published.
AI Ops benchmark
Results forthcoming
Evaluating fault identification, root-cause diagnosis, and time to diagnosis from operational data.
AOCA evidence and reportingModel Reliability & Efficiency
Research & developmentImproving today's
AI models.
We are developing methods to make existing AI models more dependable and economical in workflows where rules, tools, and operating conditions change.
We work on the models themselves through adaptation and retraining, including appropriately licensed open-weight models.
More dependable performance
Work toward better task completion as requirements, data formats, and tool behavior change, evaluated against meaningful baselines.
Better operating economics
Compare the quality and total cost of successfully completed tasks: compute, time to completion, failed attempts, retries, and operating effort. Better outcomes can justify a higher cost; a lower-cost model can be the better choice when it reliably meets the need.
Digital Product Engineering
Design, implementation & integration
From concept to working digital product.
Applications, operational interfaces, and integrated software systems: the engineering needed to make technology usable and maintainable.
Product design, user experience, architecture, implementation, and integration belong together. Our focus is working software that can be operated and evolved, not just an initial prototype.
Talk to Taijitu AIEvidence & evaluation
Explore our software engineering results and plans for evaluating multi-step browser tasks.
Planned evaluation
WebArena 2.0
Multi-step browser tasks, including navigation and interaction with changing web environments.
Evaluation detailsResearch & Papers
Explore our research in common sense reasoning, neuro-symbolic architectures, and AI safety.
Common Sense Is All You Need (2025)
Common sense, contextual learning, and adaptive reasoning as foundations for AI autonomy.
Read PaperThe Common Sense Litmus Test (2025)
A formal approach to evaluating whether AI can handle genuinely novel concepts.
Read PaperTowards Common Sense AI: A Phased Approach (2024)
A phased research roadmap connecting philosophical foundations with useful, grounded systems.
Read PaperTowards Ethical AI (2024)
The ethical implications of systems lacking common sense, moral agency, and clear descriptions of their limitations.
Read PaperSelected Prior Work
SSDL: Sensor-to-Skeleton Diffusion Model with Lipschitz RegularizationInternational Conference on Multimedia Modeling, 2025
Enabling Efficient and Interpretable Cybersecurity Reasoning through HDCIEEE Transactions on Artificial Intelligence, 2025
Company & mission
Taijitu AI is an operating technology company. Our mission is to make AI reliably useful—helping people and organizations achieve better outcomes at a cost that makes sense.
Our philosophyHugo Latapie
Founder & CEO
30+ years in AI, inventor of 70+ patents. Formerly with Cisco, Ormond Aerospace, Raytheon, and NDS/News Corp. Experience building and commercializing advanced AI architectures.
Jennifer M. Heywood
Chief Legal Officer
Patent attorney and electrical engineer. Formerly with Boston Scientific and Vista Outdoor. Experience in technical innovation, IP strategy, and regulatory compliance.
Working with Taijitu
Strategic Partnerships
Begin with the problem, the desired outcomes, and a scope we can evaluate together.
- AI & Automation Consulting
- Pilot Projects
- Research Collaboration
Start with the outcome.
Tell us what you want to achieve, what is getting in the way, and the constraints that matter.
Email
hugo@taijitu.ai
Phone
(909) 654-4316
Headquarters
131 Continental Dr, Suite 305Newark, DE 19713
United States
Taijitu AI Inc. is a Delaware C Corporation, incorporated on January 22, 2025.