Going Beyond Data Deliverables: Working with the Point Topic Broadband Graph

Arnold Kuersteiner
3 days ago
3 min read

Updated: 2 days ago

(Article written by human)

Recently we have been investing heavily into allowing our data to be consumed in more powerful and flexible ways. To do this, we created the Point Topic Broadband Graph, a semantic layer that links all our datasets and services into one unified environment, and the Point Topic Agent, an custom AI for interacting with this environment. It's designed to be accessible to everyone, from experts to non-technical users, even those not familiar with our data or broadband data in general.

Semantically linked: A look at our new Broadband Graph viewer (link)

Available as either a plug-and-play MCP connection for your preferred LLM, a web agent chat (link), ontology web app (link), SPARQL endpoint or conventional SQL database connection, our business data and the knowledge graph can be accessed in a variety of ways and by a variety of tools.

This has many advantages:

Data Deliverables	Knowledge Graph
Separate files	One environment
Mysterious column headers	Simple, standardised vocabulary
No AI access	Built for AI access
SQL, JSON, XSLX, etc.	Real-world relationships (subject-predicate-object)
Business knowledge essential	Accessible and open
Hard to share and access	Bring your own LLM/key/environment
Black box deliverable	Integrates with your own data
Insight for experts	Insights for everyone

Impact for our clients

For our clients, this can be game-changing. Where previously data had to be loaded into systems, contextualised, understood and then navigated by people with the necessary business knowledge, the Broadband Graph removes the need for expert jargon, hidden metadata files/logic, post-processing SQL joins, or cumbersome extraction logic.

Instead, you get one interconnected environment where data is already contextualised, post-processed and labelled for human and agentic consumption - you simply extract the insight. Data can be arranged, queried and processed in many ways, and new datasets and outputs to be generated, to extract exactly the information your are looking for. Where deliverables were once tied to a fixed set of columns that are expensive and complicated to change, datasets can now be extracted directly from the Knowledge Graph, and adjusted with ease.

Working with LLMs: The Point Topic Agent

Claude Sonnet 4.6 using the Point Topic Agent in Cursor

While LLMs are ubiquitous now, we know that agents are still limited by the data, context, knowledge and tools available to them. This is especially true in business data contexts, where data is often proprietary, lacks metadata, or needs additional knowledge for meaning to be effectively extracted from it. Our Point Topic Agent (link) is able to read the Broadband Graph natively, abstracting away a lot of this complexity, and allowing it to use a sophisticated set of custom tools built specifically for broadband and telecoms data.

See what the Point Topic Agent can offer:

Other Agents	Point Topic Agent
Generalised, superficial answer	Specific, quantitative answers
Simple web searches	Tool calls specifically for telecom data
Outdated, inconsistent responses	Up-to-date, verified data
No database access	Point Topic database access

The Point Topic Agent is available as an MCP server, so you can connect it to the LLM provider of your choice — including Anthropic, OpenAI, Copilot, Mistral, and Cursor. Your queries stay private: nothing is logged by Point Topic.

Side-by-Side: ChatGPT and Point Topic Agent

Curious if it makes a difference? Take a look at these one-shot query responses.

Detail, context, tools: How the Point Topic Agent outperforms regular chatbots

Note: Your preferred AI model can use the Point Topic Agent too. Speak to us to connect your provider.

Customising the Broadband Graph

The Point Topic Broadband Graph is modular and extensible. We can host a version on your servers, or extend and modify the graph with your business data. This means you can layer in your own data, business logic, metadata, or relationships alongside ours - creating a custom Knowledge Graph that reflects your specific needs. Whether you want to enrich our broadband data with your internal customer segments, add proprietary market definitions, or integrate with your existing BI tools, the graph adapts to you.