AI adoption has moved from experimentation to expectation. Applications are expected to have AI-integrated assistance or services these days.  You either adopt AI now or risk falling behind your competition. It’s not a question of whether you should adopt AI, but rather how you approach it. 

You may find yourself choosing between building an internal AI stack, requiring extensive research and deep engagement with complex technical documentation, or adopting an AI agent platform that accelerates delivery but introduces vendor lock-in and externalizes your data to third-party cloud environments.  

It’s either picking between expensive AI sovereignty or ease of use.  
 

What if these weren’t the only options? 

AI sovereignty simplified 

SELISE created Blocks AI to provide a simple way forward for AI adoption without giving up control over your data. All the functionality is built in and configurable from the UI, making AI implementation as easy as adding a button to your application.  

Model & provider agnostic: 

With Blocks AI, users are not limited to a few models provided by a few key players. While it provides a handful of models out of the box, you can add any model from any major player simply by copying and pasting your API Keys. 

For organizations running open-weight or fine-tuned models on their own infrastructure, Blocks AI supports direct integration of custom-hosted models through its configuration interface. Simply connect your own hosted models without architectural compromise.   

Deployable on your premises: 

Blocks AI is platform-agnostic by design. While the Blocks Cloud is hosted within the EU and built with multitenancy and strict data separation as core architectural principles, cloud deployment may not be suitable for every organization. 

For such cases, Blocks AI can be deployed entirely within your own infrastructure. When combined with on-premises language models, this enables end-to-end AI sovereignty, ensuring full control over data, models, and execution environments without functional limitations. 

Also available on Cloud: 

Just because on-prem is possible doesn’t mean this is the only way forward. For customers who do not have their own infrastructure or are working with data that can be hosted in the cloud, the option to use Selise Blocks Cloud is always available. Sign up for Blocks Cloud and start using our AI with just a few clicks. 

Retain Full Control of your AI, only forgo the complexity! 

Key components 

Agent specific models 

Different AI agents are built for different purposes; some require complex reasoning and planning, whereas others prioritize a more rapid response. With Blocks AI, choose the right Model for the right agent. Select from the list of Blocks provided models or the ones added by you, set the temperature, maximum response length, and set up the base prompt to reflect how you want the model to behave. 

Fig: Agent Specific Settings 

Your own knowledgebase with retrieval testing 

Empower your agent with domain-specific knowledge by adding your documents, and transform it into a RAG powerhouse. You can upload documents, add Q&As, and specify URLs to crawl (even recursively) to form your knowledge. 

Blocks AI provides fine-grained control over retrieval behavior, including query enhancement, relevance tuning, recall depth, and the balance between keyword-based and semantic search. Retrieval testing capabilities allow teams to validate and optimize knowledge of grounding prior to production use. 

Open ecosystem of tools 

Blocks AI comes with MCP support out of the box. Different authentication mechanisms are supported; simply specify your MCP URL, add your credentials, and connect to MCP servers instantly.  

However,  it’s not limited to connecting to applications with MCP servers only.  Connect easily with any application using its APIs and expose those to the AI agents as tools. These integrations are configured directly through the UI, removing the need for custom development. This approach ensures extensibility without dependency on pre-built vendor integrations. 

Personalization: make it your own 

Make the Agent your own by customizing the Chat UI look and feel. Customize the agent name, avatar, welcome message, banners, and UI colors. 

Set some preset questions for your users to get started. Enable smart suggestions for the Agents to guide your users with their queries as the conversation progresses. 

Observable by design 

Blocks AI cites the sources of the conversations from your knowledge base and provides full traces of tool calls and responses.  This ensures the agentic journey fully observable and auditable. 

Guardrails 

To ensure responsible and controlled agent behavior, Blocks AI includes configurable guardrails in addition to system prompts. Set limitations on user queries and agent responses. Prevent prompt injection, redact Personally Identifiable Information (PII), ban keywords & patterns, and set custom rules with risk thresholds to prevent any behavior that you consider risky. 

Instant roll out 

Publish your AI agent as an embedded widget or as a standalone chat interface. As simple as adding a button to your application or website. 

Who this is for 

Regulated industries  

Data sovereignty is important for everyone, but severity varies. In regulated industries, this isn’t just important; it is mission-critical. Financial Organizations, Healthcare Providers, and Governments face challenges in adopting mass-market AI solutions due to a lack of data sovereignty-friendly solutions. 

The solution? Host LLMs and Blocks AI on your own infrastructure, and set up guardrails as needed. 

Adopt AI in compliance with regulations. 

B2C businesses 

In the age of ChatGPT, customers want responses immediately, or they move on. Allow your customers to chat with an expert agent who knows your products and services inside and out.  

Streamline your L0 and L1 support by connecting your FAQs, self-service applications, and ticket management platform as tools. 

Take your customer delight to the next level. 

Enterprises with proprietary models 

Organizations that have invested in custom or fine-tuned models often face challenges keeping pace with evolving AI architectures. Is AI Agent development and upkeep taking up more effort and resources than the core business? 

With Blocks AI, leverage the existing AI investments while Blocks AI handle Agent orchestration.Easily connect Blocks AI to existing custom hosted models without increasing operational overhead. 

Why this matters 

Local regulations and internal policies are already causing many shifts in the industry. Compliance over convenience is the new way forward. The need to keep proprietary and sensitive data within the country’s borders/company’s infrastructure prevents adopting most solutions on the market. 

As governments and organizations catch up on the implications of adopting AI, the regulatory framework will only become stricter. 

We developed Blocks AI with these in mind. We don’t just provide AI Agents as a Service, we provide AI On Your Own Terms! 

Looking ahead  

For us, 2025 was the year of laying a robust foundation for integrating AI into your applications. 2026 will be the year of capability extension. We will provide a built-in library of common applications integrated as tools, enabling your custom workflows to leverage the power of our AI agents. 

What are you waiting for! 

Get in touch with us at [email protected]