Next cohort starts 15 September 2026 · 4 weeks · ₹3,999

From PDF to decision.
In four weeks.

A hands-on course where you ship one working document agent — and finish knowing exactly how accurate it is.

One track for everyone · Bring your own documents · Nothing costs extra

4
weeks, 4 modules
1
system you ship & demo
12
seats, capped
Why this course is different

Not another prompt-engineering playlist

You write code that reads documents, checks facts, and takes action — from the very first lesson.

Build first, framework second

You build an agent from scratch before touching LangGraph, so the frameworks never confuse you again. It's all the same loop underneath.

Real documents, not sample PDFs

Scans with no text layer, stamps, rotated pages, tables that break across pages. The documents that break every tutorial you've followed.

Agents that don't hallucinate

Answer from the document, cite the exact page, and correctly refuse to guess. This is the hard, valuable part — and our day job.

You finish with a number

Not "it seems to work." A field-by-field accuracy report you can put in front of a boss or an interviewer.

The curriculum

Four modules. One mental model.

Every agent is a loop: reason → act → observe → repeat. Each module adds one capability on top — concept, live build, lab, checkpoint.

1
Week 1 · the loop

Your first agent, from scratch

Make your first model call, get reliable structured output, then build a tool-using agent in about sixty lines of Python with no framework at all. Rebuild the same thing in LangGraph so you see exactly what the framework hides.

You build: an agent that answers questions about one clean document.
structured outputReActtool callingLangGraph
2
Week 2 · the core

Real documents Grokking core

When the PDF is really a photograph. OCR, layout, reading order, tables, stamps and skew. Then structured extraction: define a schema, extract into it, validate the result, attach a confidence score, and build the path where the agent says "not in this document" instead of inventing an answer.

You build: extraction across a real document corpus, plus a list of what it got wrong.
OCR & layoutschemascitationsrefusal
3
Week 3 · the workflow

From reading to doing

The module that earns the course its name. Your agent reads a document, looks something up in a second source, compares the two, writes a row to a sheet and drafts an email — with a human approval gate on anything it isn't sure about.

You build: an end-to-end workflow that produces a decision, not just an answer.
multi-step toolshuman-in-the-loopMCP
4
Week 4 · prove it & ship it

Measure it, then show it

Build an evaluation set and measure accuracy field by field. Do error analysis on the failures that matter. Add guardrails, prompt-injection defence and on-prem privacy. Then everyone demos what they built, with their numbers on the slide.

You build: a measured before/after on your own agent, and your demo.
eval harnessguardrailson-premdemo day
Straight answers

What this course is not

Worth saying plainly, so nobody pays ₹3,999 for the wrong thing.

Not covered
  • Multi-agent orchestration. Ten minutes at the end on when it helps and when it's just extra latency. Most "teams of agents" are one agent with more failure modes.
  • Fine-tuning. Expensive, and almost never the right answer for these problems.
  • Model comparisons. Stale within weeks. The code you write is model-agnostic.
  • No-code tools. You will write Python.
What AI genuinely can't do here
  • Read a handwritten annotation reliably. Plan for a human.
  • Resolve two documents that contradict each other — it should flag the conflict, not quietly pick a side.
  • Be trusted on a legal or financial clause without review.
  • Work at all without measurement, which is why a whole module goes to it.
Who it's for

Everyone builds the same thing

No beginner lane, no advanced lane. You just point your agent at your own documents.

Engineers

See what the frameworks hide, and what breaks on a real scan.

Students & switchers

Finish with a system and an accuracy number you can walk an interviewer through.

Ops, finance & bid teams

Sitting on a pile of PDFs. Every lab ships with a working template.

WHAT YOU NEED

  • Basic Python — functions, loops, dictionaries. A primer is included.
  • Comfort running a notebook
  • About 4 hours a week between sessions
  • A document problem of your own, ideally

A setup notebook goes out a week before, so session one isn't lost to installs.

Ways to join

Pick how you want to learn

Join the next open cohort, or bring the same four modules in-house for your team.

Private team workshop

Custom quoted per team
Your dates · 2 days · on-site or private
  • Run on your team's own documents
  • Private and on-prem friendly — nothing leaves your network
  • Use-case shortlist with the risks named
  • All code and course materials, yours to keep
Talk to us →
Who's teaching

Built by people who ship this for enterprises

VM

Viswarup Misra

Co-founder, Grokking

IIT engineer, Master's in the US, ex-Amazon data scientist. At Grokking he builds the document-extraction pipelines behind agents in use at NHPC, Indian Oil and Maruti Suzuki. This course teaches that exact craft.

Questions

Good to know

Do I need to know how to code? +

Basic Python — functions, loops, dictionaries — is enough, and a short primer is included if you're rusty. Every lab also ships with a working template, so you just need to be comfortable running a notebook.

Which frameworks and models does it use? +

Patterns first, framework second. You build by hand before touching LangGraph. Tools and data are exposed over MCP. We use a hosted model plus one open model you can run locally, so you can see the code is model-agnostic.

Will this cost me anything beyond the fee? +

Model API usage across four weeks typically runs a few hundred rupees. If you'd rather not use a card at all, there's a local-model path for every lab.

Can I use my company's documents? +

Yes, and it's the better way to do the course. If they're confidential, use the local-model path so nothing leaves your machine. We also provide a public tender corpus if you'd rather not bring your own.

What makes the document module special? +

It's the hardest and most valuable part of enterprise AI: answering from real documents without hallucinating, every answer traceable to the exact page. It's also what Grokking builds commercially, so you learn it from production experience.

What if I miss a session? +

Tell us in advance and we'll help you catch up. Every lab ships with starter and solution notebooks and is designed to be completed asynchronously, and questions get answered in the group between sessions.

Can you run this for my company's team? +

Yes — a two-day private version, on-site or private virtual, run on your own documents inside your own environment. Teams that go through it often move straight into a production build with us.

Do you help with getting a job? +

We help you show the work well — a write-up of what you built, a cleaned-up repository, and how to talk through it in an interview. We don't place people and we don't promise offers.

Save your seat

Starts 15 September 2026, runs four weeks, ₹3,999 with everything included. Tell us a little about you and we’ll send joining details.

This form is processed by Formspree. See our privacy policy.

Connect on WhatsApp →
Prefer email or a private workshop? viswarup@grokking.in · Delhi NCR, India
Chat on WhatsApp