IIBM has announced a self-hosted version of Bob, its agentic development platform, for internal infrastructure, private or sovereign clouds and environments isolated from the internet.[1] This option addresses a concrete obstacle: some applications contain code, data or business logic that a company wants to keep within its own environment. It still needs to be checked where the models actually do their processing.
The assistant moves into the company’s working environment
According to IBM, Bob helps with understanding applications, planning tasks, carrying out changes and validating them. The self-hosted offering notably retains an experience within the development environment and the BobShell tool.[2] The ambition therefore goes beyond suggesting a few lines of code.
For a team maintaining legacy software, the potential benefit would be to prepare changes while taking the existing behaviour into account. Billing rules, interfaces with other systems or specific processing routines often make up a large part of the software’s value. Their confidentiality can matter as much as that of the code itself.
Self-hosting Bob does not always mean running the model locally
IBM distinguishes between models installed and administered by the customer and hybrid configurations using a compatible external service. In a supported local configuration, the code, development context and build artefacts can remain within the customer’s environment.[2]
The decisive choice concerns inference, that is, running the model when it processes a request. Hosting the assistant on your own infrastructure is therefore not enough to establish that all the data stays there. You also need to examine the models used and the connections required by the chosen setup.
IBM cites NVIDIA Nemotron and Poolside Laguna among the self-hosted models supported at launch. Customers must provide access to a supported model; some existing licences can be reused under certain conditions.[2] Compatibility and usage rights are therefore part of the project.
Four decisions that determine real control
This grid sets out the questions to examine before a deployment. It does not constitute a security certification of Bob.
| Component | Operational effect | Control question |
|---|---|---|
| Hosting | Place the service in an environment administered by the company. | Who runs the components and applies updates? |
| Model | Determine where code and instructions are processed. | Is inference local or entrusted to an external service? |
| Tool access | Allow the agent to consult or modify resources. | Which repositories, secrets and environments are accessible? |
| Validation | Check changes before they are integrated. | Which tests and what human review are required before going into production? |
Keeping data inside does not eliminate errors
An environment isolated from the internet, often called “air-gapped”, can meet a network separation requirement. IBM has announced support for this type of deployment with compatible models.[1] This property relates to the architecture; on its own, it does not prove the quality of the code produced.
The team must still check the changes, the dependencies added and the consequences for applications. An agent allowed to act across too wide a scope can cause unwanted changes even without an external connection. Permissions, tests and rollback options remain operational decisions.
For SMEs, assessing the value on a well-defined project
An SME developing proprietary software could start with a limited task, such as documenting a module or preparing its modernisation in a dedicated branch. This scenario is a suggested pilot, not a customer result published by IBM.
The assessment should measure the time saved after review and correction, the relevance of the tests proposed and the defects detected before integration. The budget should also include licences, computing resources, administration and maintenance. The two announcements consulted provide neither a typical total cost nor an independent performance comparison across configurations.
The developer stays in control of changes
This development reinforces the importance of Web Programming and Management. Understanding the code, its dependencies and business needs makes it possible to frame a task the agent can handle, and then to judge whether its proposal truly meets the project’s requirements.
Code review remains central. The developer must spot an incorrect assumption, an unsuitable library or a test that merely confirms the behaviour the agent imagined. It is up to them to compare the changes against real-world use and maintenance constraints.
Their role also includes organising delegated work: defining the scope, selecting the relevant context, monitoring actions and deciding on integration. In-house hosting brings the tool closer to the software estate, but responsibility for evolving that estate remains human.
Sovereignty is verified in the chosen configuration
The announcement opens up an option for organisations whose constraints limit the use of external assistants. Its value will depend on the whole set-up: compatible models, data flows, authorised access and the ability to operate the service. The decision should be based on a documented pilot, with measured gains and clearly assigned responsibilities.
Further reading
Four CCCLX reads to explore the control of agents and their infrastructure in more depth.
Azure Linux 4.0: is Microsoft preparing for life after Windows Server?
Open-source security scales up: IBM, Red Hat and Palo Alto join forces
GitHub Copilot bulks up for business: coding agents come under IT control
AI voice agents: Oracle details a LiveKit architecture on OCI
Sources
- 01
IBM Newsroom (1 October 2026). “IBM Introduces Self-Hosted Deployment for IBM Bob to Help Enterprises Advance AI Sovereignty and Governance”. Announcement of Bob’s self-hosted deployment and the target environments.
- 02
Mishra, P. and Sundaresan, N., IBM (30 September 2026). “IBM Bob expands to self-hosted environments for sensitive and mission-critical enterprise software”. Product details on compatible models and local and hybrid configurations.
Checked on 6 October 2026. The date used is that of the press release; the product presentation is dated 30 September. Features and configurations are as described by IBM. The table, the SME scenario and the qualification criteria are editorial analysis. The four internal links were checked during research for this series.
PDFs, presentations, audio: Adobe wants to turn Acrobat into a content factory
Acrobat no longer just wants to open your documents For more than thirty years, Acrobat has mainly been used to read, create, edit, sign and share PDFs. Adobe now wants to give it a much broader role. On 9 September 2026, the company announced new capabilities based...
Salesforce plugs Claude into its CRM: Claudeforce prepares for life after the interface
What if the CRM soon no longer needed to be opened? For decades, using a CRM meant navigating between customer records, dashboards, menus and forms. With Claudeforce, Salesforce and Anthropic want to change this logic: rather than forcing users to go into Salesforce...
Kimi K3, Mistral, DeepSeek: Oracle turns OCI into an AI model hub
Oracle wants to make OCI a one-stop shop for AI Oracle is stepping up its efforts in the AI cloud battle by significantly expanding the catalogue of models available in OCI Enterprise AI. On 4 September 2026, the group announced in particular the integration of Kimi...
Agentic Data Cloud: Google rethinks data for the age of AI agents
AI agents are forcing Google to rethink data architecture AI agents promise to search for information, make decisions and act directly within company systems, but this autonomy depends on one essential condition: accessing the right data and, above all, understanding...
Google Cloud tackles the hidden bill of AI agents
AI agents raise a new problem: how much does a task really cost? As companies move from chatbots to AI agents able to work through multiple steps, the question of cost becomes much more complex. A chatbot generally handles a request and generates an answer, whereas an...
Workday, Navan, Avalara: ChatGPT Work moves into business software
ChatGPT Work is gradually leaving the chat window With GPT-6 Astra, OpenAI is no longer simply trying to improve the quality of ChatGPT’s answers. The company wants to bring its AI closer to the applications in which employees actually work. With the arrival of GPT-6...
From chip to cloud: Groq raises 350 million to scale up
Groq now wants to sell cloud rather than just chips Groq is pursuing a major strategic transformation. Long seen as one of Nvidia’s challengers thanks to its LPUs (Language Processing Units) specialised in artificial intelligence model inference, the group is now...
Muse Code: Meta launches its AI agent to tackle giant codebases
Meta wants to bring AI agents into complex software projects Meta is stepping up its efforts in AI-assisted software development with Muse Code, a new agent designed to work on large codebases, where traditional assistants can quickly lose the context needed to...
Qwen3.8-Max reportedly coded for 16 days without human intervention
Alibaba tests how far an autonomous AI developer can go Alibaba claims to have reached a new milestone in agentic software development with Qwen3.8-Max, its new flagship model launched in early August 2026. In an internal test, the model reportedly worked fully...
Two months, $1.1 billion: River AI is already exciting investors
A spectacular raise for a start-up barely out of the shadows Two months after emerging from stealth mode, River AI has just completed one of the most spectacular funding rounds of the year in artificial intelligence. The start-up founded by Igor Babuschkin, a former...
