Hoshang Sheth logoHoshang Sheth
All Services
Service Deep Dive

RAG & AI Agents.

Your knowledge, instantly queryable.

Book a Discovery Call Ask a question
At a glance
  • Timeline · 8-12 weeks
  • Investment · ₹55,000 - ₹90,000
  • Model · Fixed scope, fixed price
  • Ownership · Full source, no lock-in
<5s
answers with a source link in under 5 seconds
Source-cited
every answer shows exactly where it came from
Fewer pings
your senior people stop getting pinged for things the system knows

Figures from founder-built products and published benchmarks. Client results vary by scope, data quality, and use case.

The Problem

Why this matters right now.

Your company’s knowledge is buried in PDFs, SharePoint, Google Drive, and Notion. New hires can’t find answers. Senior staff get pinged for the same questions forever. Every decision made on incomplete information costs you.

The Approach

How I build this.

  1. 1We go through every place your knowledge lives - documents, folders, wikis and decide what should be searchable, what's out of date, and what's the authoritative version.
  2. 2Everything that makes the cut gets indexed so the system can find the right answer quickly, even when the question is phrased differently each time.
  3. 3Access controls are built in from the start so employees only get answers from documents they're allowed to see.
  4. 4Every answer comes with a link to the source so your team can verify it, not just trust it.
  5. 5It works where your team already works - Slack, Teams, or a web console - so there's no new tool to learn, and it can act on your behalf: creating tickets, booking calendar slots, or looking up an order status.
What you get

Deliverables

  • Internal knowledge base and support assistant that answers from your documents, SOPs, and product data with source citations
  • Compliance and policy assistant for regulated teams
  • Multi-step research agent that pulls from multiple internal sources and can take follow-up actions
  • Slack or Teams bot with tool use - ticket creation, calendar booking, order lookups - without leaving your workflow
Under the hood

Technology stack

LangChainLangGraphPinecone / QdrantFastAPISlack / Teams API

Vector store chosen based on where your data legally has to live, managed cloud or self-hosted. LangGraph handles multi-step agent logic that a simple chain cannot.

Case Study

DocuLens - query structured and unstructured data in plain English

Founder-built product · In Development
Problem

Most business data lives in two places that don't talk to each other, a database full of structured records and a folder full of documents, PDFs, and reports. Asking a question that spans both means running a SQL query in one tab and ctrl+F in another. Neither gives you the full picture.

Solution

DocuLens routes every natural language query to the right engine automatically - SQL for structured data, vector search for documents, or both in parallel when the question needs it. Results are merged, ranked, and returned with source attribution. One interface for everything the business knows, regardless of where it lives.

FastAPIPostgreSQLChromaDBLangChainNext.js
· In Development
Investment

₹55,000 - ₹90,000

Fixed scope, fixed price

Scales with the number of knowledge sources, access-control complexity, and agent capability - this is the most technically demanding of the five services and is priced accordingly.

Every project is quoted as a single fixed price once scoped. The range above reflects how much project size varies - not variable billing within a project.

Typical Delivery
8-12 weeks
Payment Terms
50% to begin · 20% at prototype · 30% before handover
Frequently Asked

Questions specific to rag & ai agents.

It depends on where your data legally has to live and how much of it there is. Some options are fully managed in the cloud, others run entirely on your own servers. I'll recommend the right one based on your volume, compliance requirements, and budget and explain the tradeoffs before anything is chosen.
Access controls are built in from day one. Every document is tagged with who's allowed to see it, and those rules are enforced before any answer is generated, not after. An employee asking about a restricted policy document simply gets told the system doesn't have that information for them.
When a document changes, the system picks it up automatically, either triggered immediately on update or on a scheduled refresh. Old versions are kept, not deleted, so the assistant can always tell you which version of a document an answer came from.
Yes, that's where agents come in. Ticket creation, calendar booking, order lookups. Every tool call is logged and reversible where possible.
PDFs, Word docs, Excel sheets, plain text, Notion pages, Google Drive, Confluence, SharePoint, and most tools with an export or API. If your knowledge lives somewhere specific, mention it on the discovery call and I'll confirm whether it can be connected.
Next Step

Ready to scope rag & ai agents for your business?

Tell me where your knowledge lives whether in documents, folders, wikis, Notion and I'll show you how a RAG system would work across it.

Book a Discovery Call See all services