AI agents for 14 tasks
The 14 tasks for which T-NEX builds AI agents, with the shared components, a pilot with a measurable goal and the rules for running the agents.
Learn moreAI agents for companies
Run agents in a cloud account, with approved data and an access log
T-NEX runs AI applications and agents in its own cloud account or in the customer's cloud account. The agents access only approved data, access is logged, and language models can be swapped or avoided altogether, depending on the data.
- 01Use cases for operation, data access and model choice
- 02Rules that stand before anything is connected
- 03Components for running agents
Aligned with your processes
In daily work
Use cases for operation, data access and model choice
This page covers the layer beneath an agent: where it runs, which data it may see, what is recorded about its access and which language model sits behind it. Which case we start with is agreed in the project.
- An agent in the customer's cloud account: The application, database and file storage run in an account your company holds. T-NEX receives the access needed for setup and support.
- An agent in T-NEX's cloud account: T-NEX organises the agreed infrastructure. Who owns the account and who handles each technical task is settled beforehand.
- Access to approved data: Before anything is connected, we define which sources the agent may use and which staff may see what. It does not reach anything else.
- Logged access: The agent's access is recorded. What the log contains, who may read it and how long it is kept are defined in the project.
- A language model chosen for the data: Whether an external service, local processing or additional computing capacity suits the task depends on what content the model may receive. The model can be swapped.
- An agent without a language model: For maintained answers with a source reference, such as questions about internal work instructions, a chatbot can return the approved answer unchanged. No language model rephrases it.
Overview
Rules that stand before anything is connected
Before an agent sees any data, we fix which sources it may use and for which users. An answer must not reveal anything that is not meant for the person asking.
For the language model, we clarify which content it receives, whether the task needs personal data for that, and which settings the chosen service offers for this use. If the model interface is unavailable, the workflow does what was agreed for that case beforehand.
For the log, we settle before launch what it contains, who can read it and how long it is retained. The project documents record which access development and support receive, the countries it is actually planned from and the subprocessors involved. Whether your existing data protection documents match the processing chosen is reviewed against the concrete workflow.
Overview
Components for running agents
T-NEX sets up the operating environment as part of hosting and operations. The agreed scope can include updates, monitoring, backups and incident handling, each with a named owner. Owning a cloud account does not establish who checks backups or fixes a fault, so we record responsibilities independently of the account model.
On the application side are the AI chatbot with source references and AI knowledge management. The chatbot returns maintained answers with their source. A supplementary RAG function that composes answers from document passages is developed and tested as a separate scope. Knowledge management keeps sources, ownership and access rights together.
What an AI use case can improve is shown by an AI pilot with defined evaluation criteria, run on your task and your sample data. A successful pilot is not yet a finished production solution; what operation requires is identified as next steps.
Overview
More companies for this task in Malaysia
The register of AI companies in Malaysia lists further companies for this task, among them GPU cloud and AI data centres, model providers, AI services of the large cloud platforms, and AI-based cybersecurity and protection of AI systems. This page describes only what T-NEX itself builds and runs: AI applications and agents in a cloud account.
All 14 tasks for which T-NEX develops AI agents are listed in the overview.
The workflow
How we introduce the solution.
- 01
Choose a use case: We start with one defined agent or application and settle which cloud account it runs in, yours or T-NEX's, and who in your company is responsible for decisions and approvals.
- 02
Set data access, logging and model choice: Approved sources, user groups, the content a model may receive, log contents and retention are fixed before data is connected. We also settle beforehand which data may be used in testing.
- 03
Pilot with a measurable goal: Together we define what a correct result is and which errors are not acceptable. We check quality, source reference, correction effort and behaviour when the model interface is unavailable.
- 04
Operation: Tasks for updates, monitoring, backups and incidents are agreed with named owners, and changes go through a test environment and approval. A later change of model or operator is prepared too: which data can be exported and which access is handed over is settled in advance.
FAQ
Questions about the solution
Does the agent run in T-NEX's cloud account or in ours?
Either is possible. If you hold the account, T-NEX receives the access needed for setup and support; in T-NEX's account, T-NEX organises the agreed infrastructure. In both cases, account ownership and each technical task are explicitly defined. Your company names a contact for business decisions and approvals.
Does every agent have to use a language model?
Not in every case. A chatbot with maintained answers returns the approved answer and its source unchanged, with no language model rephrasing it. Where a model makes sense, its choice and replacement depend on the data: which content does it receive, does the task need personal data for that, and does an external service, local processing or additional computing capacity fit? A supplementary RAG function is a separate scope with its own testing.
Is this a security service or an AI data centre?
No. This page describes running AI applications and agents with limited data access and an access log. It makes no statement about a particular level of protection and does not cover GPU cloud, data centres, model providers or security services such as a SOC or penetration testing. The register of AI companies in Malaysia lists companies for those topics.
Related options
You may also be interested in these.
Data protection and hosting
Establish contracting parties, data flows, model access and operating tasks for your application, with additional information on this website’s privacy.
Learn moreHosting and operations
T-NEX plans hosting and support for web applications and AI solutions, with defined accounts, operating tasks, ongoing costs and a practical handover plan.
Explore service
AI chatbot for companies with source references
Internal knowledge search and chatbots built around your content: approved answers or RAG with a language model. T-NEX scopes access, data and implementation.
Learn more
AI knowledge management for companies
T-NEX designs and builds knowledge systems with source references, evaluating document search, RAG, knowledge graphs and maintained wikis against your questions.
Explore serviceWhich task would you like to solve next?
Bring a concrete task. Together, we will define what the application needs to do.
Discuss how your agent would run