Data and Inference
Model Feasibility Assessment
A measured verdict on whether the data you hold can carry the model you have in mind, reached before any build is funded.
₹20,999GST included
Data and Inference
CR-INF-104
Document Retrieval Assistant
A retrieval-augmented assistant that answers out of your own documents, cites the passage behind each answer and admits when it has none.
Timeline22 working days
Deliverables7 itemised
PracticeData and Inference
Price₹48,999 incl. GST
01
The problem this solves
A general-purpose assistant guesses at your business, having never read a line of your documentation. We build a retrieval-augmented service across the content you provide, such as policies, manuals, support articles and contracts, embedded in a vector database with retrieval tuned using a question set you have approved. Each response carries the passage it came from, and where the corpus is silent it says as much. Both accuracy and refusal behaviour are measured, never assumed.
02
What you receive
- Ingestion covering PDF files, word-processed documents, web pages and wiki content
- Vector database with chunking and embeddings tuned to how your documents read
- Retrieval and answer service that attaches a source to every response
- A web chat widget, or a bot living inside your team chat
- An evaluation set of 50 questions, scored before and after tuning
- Guardrails for questions outside scope and for attempted prompt injection
- A cost model projecting monthly spend at your query volume
03
What changes afterwards
Every answer is anchored in your documents and cites something checkable
The questions that repeat most often stop becoming support tickets
An accuracy figure you can point to, rather than a demo impression
04
Is it the right fit?
Ideal for
- Support teams repeating the same policy answers day after day
- Organisations holding internal documentation that nobody can search
- Products where getting started means reading a lengthy manual
What we need from you
- The corpus itself, with the current version of each document identified
- 50 questions your users really ask, paired with correct answers
- A decision on the LLM provider you are willing to use
05
Before you buy
Yes. It is in Indian rupees with 18% GST included, and it is exactly what you pay at checkout. Nothing is added afterwards for the scope listed on this page.
The working days are counted from the day we hold your payment, the completed brief and everything listed under “What we need from you”. If an input arrives late, the handover date moves by the same number of days.
Ask before you buy. We will confirm in writing whether it sits inside this scope, point you to a better-fitting service, or tell you it needs separate work.
On full payment, the code, configuration, designs and documents produced for this service belong to your organisation, as set out in our Terms.
Each unit repeats the listed scope once. It does not stretch a single unit to cover more.
Often considered alongside
Data and Inference
Decision Reporting Build
Dashboards resting on governed queries and written definitions, shaped around the handful of decisions your leadership takes each month.
₹15,999GST included
Data and Inference
Scheduled Pipeline Programme
Orchestrated data pipelines with dependable retries, alerts and backfills, replacing the unwatched scheduled scripts your reporting quietly rests on.
₹37,999GST included