SI Agent News
DevelopingGeminiAgents

Bespin Global launches HelpNow Computer, an AI agent that operates business software

Korean cloud firm Bespin Global released an agent that clicks through browsers and terminals, with a control layer for retries, stops and approvals.

9execution tools the agent can use

The short version

Why it matters

Agents that act inside company systems are only useful if they can be limited and stopped. This launch shows one vendor treating the control layer as the headline feature, not an afterthought. Buyers comparing agent products can use the same checklist: who approves risky steps, where does the agent run, and which models can it use.

What Bespin Global launched

Bespin Global, a Korean cloud services company led by Heo Yang-ho, announced an enterprise AI agent called HelpNow Computer, as aitimes.com reported. Unlike a chatbot that only answers questions, it operates browsers, applications and terminals itself to get work done.

It is built into the company's enterprise AI workspace, HelpNow Flex, and uses 9 execution tools. These include natural-language SQL generation (turning a plain request into a database query) and browser and computer control.

The control layer

The feature Bespin stresses is an agent runtime, a layer of software that breaks a user's goal into steps. It then manages how each task runs, when it is retried and when it is stopped.

That matters because an agent that clicks through real systems can repeat a mistake or keep going when it should halt. A runtime that owns those decisions gives companies a place to set limits.

What it can do and how it is fenced in

The agent can move through systems such as ERP (enterprise resource planning) and groupware, entering data and filing requests. It can also write PDF, Word, Excel and Hangul documents. Through MCP (Model Context Protocol, a standard way to connect AI agents to tools) and APIs, it links to email and CRM (customer relationship management) systems.

On security, it supports role-based access control and blocks personally identifiable information. All work runs in an isolated sandbox, and high-risk tasks need user approval. It also connects to existing identity systems such as Keycloak and Okta.

Flexible deployment

The product works with a range of models: GPT-OSS, Llama, Qwen, HyperCLOVA X, EXAONE and Gemini. Companies can run it on major public clouds, on their own servers, or in Korea's CSAP (cloud security certification) environments.

Buyers can choose a subscription SaaS version or an installed one. The source gives no price, customer names or performance results.

What to watch

Try it

When evaluating any workplace AI agent, ask the vendor three questions: which actions need human approval, where does it run, and what happens when a step fails.

Sources and links

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

Listen to the AI Wire headlines

Listen to the AI Wire

Listen to our daily AI shows every morning: on your commute, in the shower, wherever you are

Listen to the daily shows

First reported by aitimes.com · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

More AI news