IaGenify Beta is where the platform becomes a product you can work in. It is not a mock interface around a chatbot: the same SDK that applications use drives its model sessions, agent loop, tools, workflows, storage and account services.
That relationship matters. A capability proven in the beta is not trapped inside one interface, and code written with the SDK is not talking to a second, smaller platform. Both surfaces meet the same account, catalogue, permissions and execution records.
A conversation that can act
The chat starts like an ordinary conversation, then reaches the live web, files, databases, terminals, media tools and IaGenify services when the task needs them. The model picker shows the models actually available to the account, and every answer carries a credit ceiling rather than an open-ended bill.
Four modes define the boundary: Ask, Auto, Plan and Full access. Plan is read-only. Database and DNS changes still ask in Auto. Sending a message, publishing to the internet and spending credits outside the model call ask every time in every mode, because none of those actions can be quietly taken back.
Work that continues beyond one answer
A useful agent run can become a workflow instead of a prompt you have to remember. The canvas connects model calls, agents, web research, storage, documents, media, HTTP calls and service actions. A run keeps its steps and costs, can test one node in isolation, pause for a human decision, retry a failure and resume without starting over.
The same workspace holds managed databases and object storage, connected systems, hosting projects, domains and mailboxes. This is what lets an agent move from researching a launch to publishing the site, attaching its domain and routing replies — with approval at the irreversible steps.
A personal system, not a blank assistant
Skills, reusable sub-agents and MCP servers belong to the account and follow it across devices. A skill supplies a method; a sub-agent takes a bounded piece of work; MCP connects external tools without a custom IaGenify adapter. Only the tools explicitly allowed at connection time are offered to the model.
Files keep their boundary too. On a computer, the agent works with local files through the SDK. On the web, it can use IaGenify Cloud or connected storage. A phone can keep selected files in its own browser. The interface does not pretend that every device has the same filesystem.
One product, two surfaces
The beta is the fastest way to use the whole platform. The SDK is the way to put the same capabilities inside your own application, automation or worker. They share the vocabulary — models, agents, tools, workflows and storage — so moving from an experiment in the beta to code does not require learning a different system.
The goal is not to make a chatbot with more buttons. It is to give one agent enough connected context, tools and guardrails to carry work from a request to a finished result.
Why it is still called Beta
The core system is already broad, but the product is still being shaped through real use. Names can settle, workflows can gain new steps, and the SDK surface can become smaller as repeated patterns turn into first-class methods. Beta names that stage honestly.
What should remain stable is the direction: one account, one permission model and one execution layer, available as a workspace for people and as an SDK for software.
Frequently asked questions
Is the beta a separate product from the SDK?
No. The beta is the hosted product interface; it uses the SDK and the same platform services your own application can call.
Can the agent change connected systems without approval?
The mode sets the ceiling. Plan is read-only; storage and DNS writes ask in Auto; sending, publishing and certain spending always ask in every mode.
Do workflows keep running after the chat answer ends?
Yes. Workflow runs have their own status and history, can wait for human approval, and can be followed, resumed, retried or cancelled.
Where do files live?
That depends on the device and storage you choose: local files on a computer, browser-held files on a phone, or managed and connected cloud storage.