Service · AI engineering
RAG & knowledge systems
Being designed. Not yet offered — register interest via Work with us.
At a glance
Status
We are designing this service line and are not taking engagements for it yet. The approach and deliverables on this page describe how we intend to run it and may change before it opens. If its status changes, the new status will be shown here and on the services index.
- Who it’s for
- Enterprises with internal knowledge bases · Startups building search or assistants
- Practice
- AI engineering
- Solutions
- StartupsEnterprises
Contents
The problem
A RAG system can fail at retrieval (the right passage never reaches the model) or at generation (the model misuses what it retrieved). Without separate retrieval and answer metrics, teams cannot tell which component to fix.
Approach
How we intend to run this work, step by step.
Step 1: Audit the corpus
Document formats, access control, freshness requirements and the questions users actually ask.
Step 2: Build and measure retrieval
Chunking, hybrid lexical + vector search and re-ranking, evaluated with recall on labelled queries.
Step 3: Ground the answers
Citations to retrieved passages, refusal when evidence is missing, faithfulness checks.
Step 4: Keep it current
Incremental indexing, permission-aware retrieval and monitoring.
Deliverables
What an engagement is designed to produce. The final list is fixed in the written proposal.
- Ingestion and indexing pipeline
- Retrieval and answer evaluation suite
- Application with citations and access control
- Operational runbook
What this is not
- Not yet offered. This page describes a planned service line, not a service you can book today.
- Retrieval must respect the source systems' existing permissions; we do not flatten access control.
- Not a fixed package. Scope, deliverables and acceptance criteria are agreed in writing for each engagement.
Related research & insights
Paper breakdowns and guides on the topics this service draws on: Retrieval (RAG) · AI engineering.
Guide ·
A practical checklist for shipping a RAG system
What to have in place before a retrieval-augmented generation system meets real users: a retrieval evaluation set, structure-aware chunking, hybrid search, verifiable citations, permission-aware retrieval, a freshness plan and production monitoring.
Paper breakdown ·
Retrieval-Augmented Generation: pairing a language model with a searchable memory
A technical reading of the 2020 paper that introduced RAG models — how a dense retriever and a sequence-to-sequence generator are trained together, the difference between RAG-Sequence and RAG-Token, and what the results do and do not show.
Explainer ·
What counts as an AI agent? A working definition for engineers
A definition you can check against code: a model, a set of tools, a loop, state carried between steps and an explicit stop condition — plus how to tell when a fixed workflow is the better engineering choice.
Questions
Can we engage you for this now?
Not yet. This service line is planned: we are designing it and are not taking engagements for it. You can still register interest and describe what you need — it creates no obligation on either side.
Do you publish prices or timelines?
No. Scope, deliverables, acceptance criteria and any timing are agreed in a written proposal before work starts. We do not publish delivery service levels we have not committed to.
Can we share confidential material in an enquiry?
Please don’t. Describe the problem in general terms; no confidentiality obligation arises until an NDA is signed.