Business case: a research intelligence console for our corpus

The problem

We hold a corpus that matters. Documents, oral histories, records, a specialist archive built over decades. Almost nobody reads it in depth, because the access tools are 1995. A new researcher can't form a useful question because the archive's contents are opaque from the outside. Fluency takes years of working alongside people who lived through the period, and those people are leaving, retiring, or busy elsewhere. Institutional memory does not transfer through PDFs.

Proposed engagement

Eight to fourteen weeks. Syntheos ingests the corpus, deploys the console, and trains our researchers on the first wave of investigations. Our subject matter experts name the themes that matter for our domain and judge the answers on real questions. Where our questions turn on relationships that passage search can't follow, Syntheos builds a knowledge graph over the corpus as well. By week fourteen the console runs on our infrastructure, our researchers have used it for real work, and the documentation for adding new material is in our team's hands.

What we get

• Researchers asking the corpus questions and getting answers in seconds. The depth of investigation that previously required years of corpus familiarity is available to a new researcher in their first week.
• Institutional memory that survives the people who carry it. The console is the durable form of what currently lives in a few people's heads.
• Every answer cited to the page, with inference marked. A researcher always knows whether they're reading the archive or reading the model.
• Independence from the vendor. The pipeline, the console, and the source material run on our infrastructure under our license. We own the system.
• A corpus that grows. Documentation for ingesting new material is part of the engagement, so adding to the archive is a routine operation our team performs without Syntheos.

Risks and mitigations

• What if Syntheos disappears. The console, the pipeline, and the source material all run on our infrastructure under our license. The system continues to operate without Syntheos personnel.
• What if a researcher cites an inference as if the archive said it. Citations are direct quotes with a page range and a link to the source document. Inference is stated in prose without a citation number, and in figures it is drawn dashed where cited material is drawn solid. The console refuses to hide the distinction.
• What if our corpus contains sensitive material. The console can be deployed in air-gapped or access-controlled environments. Source material does not leave the boundary we set.
• What if our researchers do not adopt it. Adoption is a real risk for any new tool, and we treat it as a scoping question rather than an afterthought. The first wave of investigations is designed with our team so the console launches on use cases that matter, and a researcher cohort is trained during the engagement.
• What if we want to leave. License terms allow our team to fork and operate the console independently. We own the console, the documentation, and the data outright.

Success metrics

• A new researcher builds a productive investigation thread in their first week.
• Researchers find archive material via the console that traditional search did not surface.
• Chat answers reliably cite passages a researcher can verify against the source on the spot.
• Twelve months in, our team is adding new corpus material without Syntheos in the loop, and the archive the console covers is growing.

Investment and timeline

Eight to fourteen weeks for the first deployment. Cost varies with corpus size, whether the work needs a knowledge graph, and the metadata quality of the source material, so a single public range would be misleading. A scoping conversation produces a fixed first-year number before any contract is signed. For most engagements that first-year investment lands close to the cost of one medium consulting engagement. Annual hosting and support is a small percentage of the first-year cost. We can self-host or use Syntheos-managed infrastructure depending on the sensitivity of the source material.

Recommended next step

Send Syntheos fifty documents from our corpus. Within two weeks they'll run them through the ingestion pipeline and walk us through the console answering questions on our material. There is no fee. By the end of the two-hour walkthrough, we'll know whether the system handles our corpus the way we need it to. Syntheos asks one thing in return. Whoever owns the decision is in the room for the walkthrough.