AI that works.
As aspected.
Vector databases retrieve text that's similar, brilliantly. But enterprise AI needs knowledge that's correct, and that's a different problem. Aspected is a new kind of database, built to solve it.
Similarity ≠ Correctness
Vector Databases retrieve similar text. Enterprise AI needs the right knowledge.
Vector databases rank on similarity, it's exactly what makes them so powerful. But similarity is blind to metadata: whether a result is recent, authoritative, or permitted to be seen. Metadata gets added afterward, as a filter, working against the ranking rather than guiding it. So the right answer can sit just out of reach, not because the database is doing anything wrong, but because it was never built to weigh correctness. That's the gap Aspected closes.
Select one of the questions on the bottom left
Pick a question and compare a conventional vector database with Aspected, side by side, on real data. The similar-sounding answer against the correct one. They're often not the same document.
Note: Select a question on the lower left to compare the results.
"After seeing the live demo “There's your ROl, isn't it?... You guys are purely unique in this space."
Tom Lane
Content Management Lead, Capgemini UK
The limitation isn't your stack. It's underneath it
Because a traditional vector database understands similarity but not context, teams compensate by stacking on layers with metadata filters, rerankers, graph DBs and rules. Each adds complexity, latency and cost, and none closes the underlying gap. The industry has been paying this Vector-Database Tax for years, the cost of working around a known limitation, because until now there was no alternative.
Traditional Vector Database
Embedding - text only
Ranking signal - semantic similarity
Metadata - filtered after retrieval
Authority - ignored
Structure - ignored
Result - similar documents
Extra Layers
Every layer is a workaround for the same limitation.
+ metadata filters
+ rerankers
+ graph DB
+ rules
Download the Whitepaper for CIO's, The Vector-Database Tax and find out how one flaw in AI search returns the wrong answer and wastes up to a quarter of your token budget and how we fixed it.
The Aspected Database — a new kind of database
Not another layer on top of the vector database but a new index structure that closes the gap at the source. Aspected fuses meaning, metadata, authority and context into a single ranked pass, so retrieval weighs what's correct, not only what's similar. We stopped asking "what text looks similar?" and started asking "what knowledge is correct in this context?" It's patented because it's a new category, not a clever query.
Aspected
Embedding - text + metadata + context
Ranking signal - multi-aspect relevance
Metadata - part of ranking at index time
Authority - a first-class dimensions
Structure - native to index
Result - correct information
Aspected for Copilot
Copilot on SharePoint runs into the vector database's limitation at enterprise scale.
Many organizations are adopting Microsoft Copilot on top of SharePoint.
In practice, retrieval quality often becomes the limiting factor:
-
Relevant documents are missed
-
Results vary depending on phrasing
-
Near-relevant information is not surfaced
-
Users retry prompts to improve answers
In many environments, improving these results requires extensive cleanup and restructuring of SharePoint before AI performs consistently.
Aspected reduces the dependency on perfectly cleaned SharePoint environments.
Aspected for Copilot places the Aspected Database between SharePoint and Copilot to improve how context is prepared and retrieved for AI systems. Instead of relying on a vector database's single similarity query, it evaluates multiple aspects at once, meaning, time, priority, intent, context, and constraints.
Multiple Aspects
Meaning
Time
Priority
Intent
Context
Constraints
This results in:
-
More reliable Copilot answers
-
Fewer silent semantic misses
-
Lower retrieval complexity
-
Improved retrieval consistency
Already deployed in production at AGFA.
Copilot answers. SharePoint stores.
Aspected for Copilot retrieves the right context.
Whitepaper
"Why Copilot projects disappoint and how to improve results without cleaning up SharePoint first"
Organisations that deploy Aspected already today
Customers
AGFA Service engineers use AI to gather, prepare, and orchestrate enterprise knowledge, delivering the right answer from the right source, directly within their ticketing system.
Partners
Aspected works with different local and global AI consultancy and system integration partners. Such as Novadoc and Capgemini.
Proof from the field
"Aspected stood up in 30 minutes and gave us something we didn't have: a way to actually see retrieval quality side by side instead of guessing, which immediately exposed a gap in our own baseline. What stands out isn't a single "it's better" number; it's that the aspect model gives you, and an agent, real levers to shape relevance per query that a plain vector store simply can't."
Andy Copland
Director, CGA Management
"With Aspected we made metadata part of the search index. It cuts search times in half and delivers maximum accuracy at lower cost with governance built in upfront."
Wim Verheij
Managing Director, Novadoc Netherlands
"Aspected has added the context Microsoft Copilot needed. The answers are now perfectly aligned with how we organize and use our information."
Rob Kasslack
IT Business Partner, AI & Innovation, Agfa
"After seeing the live demo.. there's your ROI, isn't it? You guys are purely unique in this space."
Tom Lane
Content Management Lead, Capgemini UK
Frequently Asked Questions (FAQ)
Most RAG systems retrieve content that's semantically similar but not necessarily correct for the user's context — because a vector database ranks on similarity and treats metadata as a separate filter applied afterward. Aspected closes that gap by combining meaning, metadata, structure, authority, and context into one retrieval pass, so AI retrieves the right knowledge, not just similar text.
Aspected is built for teams developing enterprise AI, RAG applications, knowledge assistants, service automation, and AI-powered search. It is especially relevant for organizations where retrieval quality, trust, context, and source correctness are business-critical.
The Aspected Database is a retrieval engine for AI and RAG systems. It creates a unified representation of content and metadata, so queries can retrieve knowledge using multiple signals in one step instead of relying on separate filters and reranking layers.
Traditional vector databases mainly rank results by semantic similarity. Aspected ranks knowledge using multiple aspects at once, including text, metadata, time, type, structure, and context. This makes retrieval more precise and reduces the need for extra filters, rerankers, or custom rules.
Similarity search can find content that looks related, but enterprise AI often needs content that is correct, current, trusted, and relevant to a specific business context. In complex knowledge environments, “similar” and “right” are not always the same thing.
Aspected can replace parts of a traditional RAG retrieval pipeline, especially where teams currently rely on vector search plus metadata filters, rerankers, graph databases, or rules. It is designed to simplify retrieval by bringing these relevance signals together in one system.
Aspected is a Spin-off from Xillio - a global enterprise data transformation company.
Over two decades, Xillio worked inside some of the world’s most complex enterprise knowledge environments — building software to transform and migrate large-scale content systems. This experience led directly to the creation of Aspected.
We invented the Aspected Database, a retrieval engine for AI systems that enables organizations to access the right knowledge from complex enterprise data. By combining semantic understanding with operational context such as metadata and system signals, Aspected helps AI assistants retrieve relevant information instead of just similar text. This makes enterprise AI more reliable and useful in real-world operations
Yes. Aspected works with AI consultancies and system integration partners that help customers implement enterprise AI retrieval solutions. Don't hesitate to contact us here.
AI that works.
As aspected.