Tier 3 · scoped build
Retrieval and document intelligence build
Search and question-answering across your own contracts, policies, claims files or case law, with every answer citing the passage it came from.
Who this is for
An organization whose institutional knowledge sits in documents nobody can search, in a setting where a wrong answer has consequences.
From
$18,000
- Typical
- $28,000 to $70,000
- Other currencies
- €16,500 · £13,900 · S$24,000
Thirty minutes with the engineer who would run the work, before any paperwork.
The work, stage by stage
In order, with what happens at each phase and what it produces. Phases rather than dates, because a build is scoped per engagement and the catalog deliberately publishes a price band instead of a duration.
Corpus and question discovery
Phase 1
What the documents actually are, in what formats and what condition, and the questions the system has to answer well. That question list becomes the acceptance test, so it is agreed before anything is built rather than after.
Ingestion and chunking
Phase 2
Parsing, structure recovery and chunking tuned to your document shapes. A contract and a claims file chunk differently, and one strategy applied across both is where most retrieval quality is lost.
Retrieval and grounding
Phase 3
Hybrid retrieval with reranking, and generation constrained to cite the passage it used, so an unsupported answer is visible rather than merely plausible.
Evaluation harness, before launch
Phase 4
The golden set from phase 1 wired into continuous integration behind a groundedness threshold, so the system cannot regress silently after handover. It is a release gate rather than a report.
Deployment and handover
Phase 5
Running in your environment, with the runbook, the harness and enough documentation for your team to change it without us.
Everything that arrives, listed
What is handed over at the end, and what you own of it. If something here would not be useful to you, say so before we start and the scope changes.
- The running system, deployed in your environment rather than hosted by us
- Source code for the system written for you, which you own on payment in full under our terms
- The golden set, and the evaluation harness wired into CI behind a groundedness threshold that fails a release
- The ingestion pipeline, documented, so your team can add a new document type without us
- An architecture document giving the retrieval configuration and the reasoning behind each choice
- A runbook: what to do when quality drops, when the corpus changes, and when the model is updated underneath you
What this covers, and who takes the rest
Stated up front, because scope is what derails an engagement in week two rather than at the proposal.
Inside this build
- Ingestion, retrieval, reranking, grounding and citation across the agreed corpus
- The evaluation harness and the release gate, built before launch rather than after
- Deployment into your environment, with handover documentation
Handled separately
Named so you can plan and budget for it, whether we do it or somebody else does.
- Licensing for third-party corpora you do not already hold rights to
- Ongoing operation after handover, which the Operate retainer covers
Start with a call, not a proposal.
Thirty minutes, straight to the engineer who would run the work. We will tell you if a smaller engagement answers your question, or if nothing here does.
Typical reply within one business day · Engagements start at $2,500
Scoped per engagement