Ask two experienced investigators to research the same person and you will get two different files. One will go deep on company records because that is where they found something last time. The other will spend the afternoon on court cases. Neither file is wrong. But neither tells you what was not looked at.
That gap is the real risk in desk research. A missed directorship or an unread court order is rarely missed because it was hard to find. It is missed because nobody went looking in that place.
Anvexer is built around a simple answer to this: research by aspect.
What an aspect is
An aspect is one area of a subject's life that an investigator would want covered. Anvexer works with twelve of them, plus a general area for anything that fits none:
- Personal information
- Relationships
- Businesses
- Financials
- Criminal history
- Legal matters
- Lifestyle and assets
- Social media
- Travel and movement
- Political connections
- Timeline
- Contact details
These headings are not new. Any good due-diligence report uses something close to them. What matters is what the system does with them.
How aspect-based research works
When an investigator names a subject, Anvexer does not run one big search and summarise the results. It plans the research one aspect at a time.
Each aspect gets its own questions. Before any searching starts, Anvexer checks the case files the team has already uploaded against the open questions for every aspect. What the case files already answer does not need to be searched for again.
Aspects are researched in parallel. A coordinating agent hands each aspect to a separate research worker, with that aspect's open questions attached. The workers search the open web, read what they find and record findings as they go. A page one worker has read can be found by the others straight away, so the same source is not fetched twice.
Each aspect is closed or left open, with a reason. At the end of a round, every aspect gets a decision: closed, because the questions were answered or searching stopped producing anything new; or left open, with the reason written down. The record also shows how many questions were put to the case files and how many to the web.
Related people and companies can become subjects of their own. When research on a subject finishes, Anvexer proposes a short list of the most relevant connected people and organisations. Up to a set limit, those are researched in the same way. The investigator can remove any of them.
"Nothing found" has to be earned
The most dangerous sentence in a background report is "no record found". It can mean that there is no record. It can also mean that nobody searched properly.
Anvexer treats these differently. A research worker that stops finding anything new is allowed to stop. But the system will not record "we searched and found nothing" for an aspect unless enough searching was actually done to support that statement. Until then, the absence is held back and is not shown as a finding. If later searching does enough work, it then appears.
This rule is enforced by the software, not left to the judgement of the AI model.
What the investigator gets
At the end of a research run, the investigator does not just have a dossier. They can see which areas of the subject's life were covered, which were left open and why, and how much searching sits behind each one.
That changes the conversation with a supervising officer. "Here is what we found" becomes "here is what we found, here is where we looked, and here is what is still open."
Frequently asked questions
Can we change the list of aspects?
A research run can be limited to some of the aspects, or use a custom profile of aspects for a particular kind of subject. The full set is the default.
Does aspect-based research replace the investigator's own research?
No. It gives the investigator a structured first pass and shows where the gaps are. Deciding what to pursue, what to rely on and what to act on stays with the investigator.
What happens to findings if a research run is stopped part way?
Findings are saved as they are made. Stopping a run keeps what has been found so far, and the run can be resumed later.
Analysis published by the Anvexer Research Cell at Sobitai Solutions LLP. Exploring AI decision-support, NATO/Admiralty evidentiary frameworks, and graph intelligence for modern law enforcement.