SUMM

Temporal emphasizes reliability as a core value, symbolized by its mascot, the tardigrade, known for resilience
The company focuses on providing a platform for building reliable, durable, and scalable agentic AI applications
Agents are complex distributed systems that require orchestration, human interaction, and parallel processing
Orchestrating these workflows and maintaining visibility and traceability is critical due to system unreliability, especially with LLMs

LLMs and agentic workflows are inherently unreliable, making debugging and testing particularly difficult
Complex distributed system reliability challenges have existed previously, and Temporal was founded to address these issues
Temporal's mission is to abstract away the challenges of reliability and scalability so developers can focus on business logic

Temporal offers idiomatic SDKs in several languages, with Python recently becoming the most popular among users
The platform manages all essential plumbing code to ensure reliable execution, offering robust guardrails for developers
Temporal is a mature, battle-tested product, used in production for over a decade by mission-critical applications
Companies such as Dust and Gorgeous (supporting brands like Reebok, Timbuktu, and Glossier) deploy agentic AI at scale on Temporal
Payment processing and other mission-critical workloads also rely on Temporal for reliability and agility

Temporal abstracts workflow orchestration and error handling into developer-friendly workflows written in code
Developers use SDKs to define workflows, focusing on orchestrating interactions between users, language models, databases, and tools
Temporal accelerates development speed, with customers reporting over a sixfold increase in feature delivery velocity
Scalability is managed by Temporal Cloud, so developers do not handle scale logic manually
The result is more reliable applications, improved customer satisfaction, and developers able to focus on core business logic

Example walkthrough of a ticket booking agent workflow, implemented in code using Temporal’s workflow abstraction
Key features include user input via signals, the ability to query workflow state, and orchestration of interactive loops
Temporal manages all retry and failure handling automatically, removing the need for manual plumbing code
Complete workflow history is stored for visibility, debugging, compliance, and can be exported as needed
Tools are wrapped as activities, supporting integration of a variety of LLMs
Rich programming model allows for loops and frequent agent use-case patterns

Temporal Cloud manages all execution state, retries, and scaling, while workflow code runs locally in the user’s environment
Integration fits with developers’ existing CI/CD workflows, requiring no major codebase changes
Temporal is open source, with publicly accessible code exchanges for reference projects and examples
Developers can quickly launch proof-of-concept applications using available credits and examples, enabling fast experimentation

Temporal Cloud is available for sign-up, with credits offered and accessible sample projects on the code exchange
Developers are invited to visit the Temporal booth (G3) for hands-on demos and further questions

Scaling AI Agents Without Breaking Reliability — Preeti Somal, Temporal