Kuching
26.1°C
Light rainLight rain
Samarahan
31.2°C
Partly CloudyPartly Cloudy
Serian
26.1°C
Light rainLight rain
Betong
34°C
Partly cloudyPartly cloudy
Sri Aman
34°C
Partly cloudyPartly cloudy
Sibu
31.1°C
Partly cloudyPartly cloudy
Mukah
31.1°C
Partly cloudyPartly cloudy
Sarikei
31.1°C
Partly cloudyPartly cloudy
Bintulu
31.2°C
Patchy light rainPatchy light rain
Kapit
28.5°C
Patchy rain nearbyPatchy rain nearby
Miri
30.1°C
Light rainLight rain
Limbang
29.1°C
Light rainLight rain
|
12 Nov 2025
OpenAI AgentKit Simplifies AI Agent Development

OpenAI AgentKit Simplifies AI Agent Development

SAN FRANCISCO — OpenAI has introduced a major new suite of tools for developers called AgentKit, an integrated platform designed to significantly simplify the creation, deployment, and optimisation of sophisticated AI agents.

The launch, announced by CEO Sam Altman at the company’s annual DevDay conference, targets the fragmented workflow developers previously faced when building complex, multi-step AI systems.

​AI agents, unlike traditional chatbots, are designed to execute proactive, goal-directed tasks, such as planning a sequence of steps and interacting with external applications.

Prior to AgentKit, building these ‘agentic workflows’ required juggling numerous custom connectors, orchestration scripts, and manual evaluation pipelines.

​The new toolkit streamlines this process by centralising core functions into three main components. The Agent Builder provides a visual, drag-and-drop canvas where developers can design and manage multi-agent workflows, complete with versioning and custom guardrails.

This visual approach is touted to slash iteration cycles, with early customers reporting that tasks which once took months of coding can now be prototyped in a matter of hours.

​For deployment, the ChatKit module offers a set of tools to embed customisable, chat-based agent interfaces directly into a company’s own websites or applications, making it easier to integrate AI assistants that match existing brand themes.

​Crucially, the platform also features expanded Evals capabilities for measuring performance and reliability. These new evaluation tools include trace grading, a process to analyse an agent’s step-by-step decision-making, and automated prompt optimisation. This systematic approach to testing and iteration is vital for deploying reliable agents, especially in regulated industries.

Furthermore, the kit includes a Connector Registry, a centralised admin panel that helps enterprises securely govern how their agents connect to and utilise proprietary data sources and third-party tools like Google Drive and Microsoft Teams.

​The introduction of AgentKit, which builds on the existing Responses API, signals OpenAI’s clear strategic move to become the end-to-end operating platform for the next generation of business automation.

Sources: OpenAI, Tech Crunch

By Shomirul Islam Bonny

Hi, I'm Shomirul Islam Bonny, a Front-End Web Developer currently serving as the CTO at Sarawak Daily. I'm passionate about technology and enjoy sharing my tech knowledge with others.