Research enrichment

Turn scattered sources into records a person can inspect.

Aran can prepare structured research from approved sources and existing records, carrying source, context, and ownership into review. The output is evidence for a decision—not an automatic verdict.

Scope an enrichment pilot Inspect the record path

Approved inputs · source-linked output · named human decision owner

Input: Approved sources and records
Transformation: Structure, enrich, deduplicate
Outcome: Reviewable evidence

The record path

Enrichment should add context without hiding where it came from.

A useful record keeps the original input, the source used for each addition, the state of the evidence, and the person who owns the downstream decision.

When those fields disappear, polished output can look more certain than the underlying material allows.

Enrichment ledger · evidence stays attached
Field	System action	Review signal	Owner
Identity	Match an approved source to an existing record.	Source link and duplicate status remain visible.	Aran prepares
Context	Extract relevant facts and connect them to the working brief.	Evidence can be opened and checked in context.	Aran prepares
Freshness	Record when the source was checked and flag missing evidence.	Unknown or incomplete fields are not silently filled.	review signal
Relevance	Organise the record against the agreed requirements.	Relevance is presented as preparation, not a final decision.	human decides
Action	Prepare the approved next step for review.	Permission and consequence determine the checkpoint.	human approves

Workflow shape

One governed path from source to decision.

The source policy, record schema, review state, and action boundary are mapped before the workflow expands.

01 / define Agree the research question
Name the decision the research is preparing, the fields that matter, and the people accountable for review.
02 / permit Set an approved source policy
Document which existing records and external sources may be used, plus handling and retention constraints.
03 / prepare Research and structure the evidence
Extract relevant information, resolve duplicates, attach provenance, and make missing evidence explicit.
04 / review Route ambiguity to a person
Keep exceptions, uncertain matches, sensitive fields, and consequential recommendations visible for human review.
05 / learn Refine the working rules
Use approved corrections and observed failure modes to improve the workflow without erasing accountability.

Boundaries

A fuller record is not automatically a truer record.

Enrichment needs the same care as the decision it supports, especially when records contain personal information or could lead to contact.

The workflow can support

Source-linked research across an agreed evidence set.
Record matching, deduplication, tagging, and gap flags.
Consistent preparation for an existing review process.
Explicit routing when evidence is missing or ambiguous.

The workflow does not assume

That an open source is accurate, current, or appropriate to retain.
That found contact details imply consent or permission to contact.
That machine confidence transfers accountability away from a person.
That one record schema fits every decision or sensitive context.

Where it applies

One operating use, with other applications to test.

Casting research is the operating example. Other research-heavy workflows are possible pilot shapes only after their sources, decisions, and permissions are mapped.

Operating example