Making RTI smooth as butter

One request. Every branch. Every reply. A citizen-first view that turns registrations, transfers and replies into one path you can actually follow.

Not a “coming soon” slide deck. Pick a case, click through the mess or bring redacted JSON. You decide whether the answer worked; the site does not decide for you.

Five RTI plot twists

Pick your paperwork adventure.

Choose a caseUse a redacted case · Import JSON

Pick a citizen to see why they filed, their exact RTI questions, and what happened to every branch.

Maya's fictional case · Three-way parallel split

Fellowship selection records

Filed 2 Jun 2026 with Central Fellowship Selection Board (fictional)

Questions: 3
Registrations: 3
Replies: 3

Why Maya filed

Find the records behind her fellowship result.

What happened next: One application was forwarded to three CPIO sections, producing three registrations and three separate replies.

Where the trail gets messy

Maya must open every registration and manually remember which reply addresses marks, the approved cut-off, and the seat calculation.

The dependency tree below puts this history back in order.

What Maya asked in the RTI applicationThese are the original questions. The Reply Map checks each one below.

Q1
Marks record
Provide the written-test marks, interview marks, and total marks recorded for application DEMO-024.
Q2
Approved cut-off
Provide the approved category-wise cut-off document used for the 2026 selection.
Q3
Seat calculation
Provide the calculation sheet or approval record showing how the total number of seats was determined.

1 · Where did my request go?

The RTI family tree.

Three-way parallel split

Every registration, transfer and reply—reunited at last.

1. Forwarded in parallel
  1. Reply received
2. Forwarded in parallel
  1. Reply received
3. Forwarded in parallel
  1. Reply received

2 · The part that matters

Cool. But did they answer?

3 questions · 0 checked

Open a question, read the source and say yes or no. Revolutionary, apparently.

Compare the experienceProposed: answer and next step together

You're the judge here.The JSON links records to questions. Only you can say whether the information is satisfactory.

Q1Marks recordDEMO/CFSB/R/E/26/00421Check this question

Citizen askedProvide the written-test marks, interview marks, and total marks recorded for application DEMO-024.

Reply passage“Application DEMO-024 — Written test: 62/80; Interview: 16/20; Total recorded score: 78/100.”

Record checked: maya-results-reply.pdf
Page / location: Page 2 · Candidate marks table
Registration: DEMO/CFSB/R/E/26/00421
Record date: 2026-06-18

What this record showsThe requested written-test, interview, and total marks appear in the table.

Open watermarked sample PDF (new tab)

Did this give you the information you needed?

Your answer, not a status chosen by the website.

YesThis answers my questionNoI still need information

Q2Approved cut-offDEMO/CFSB/R/E/26/00421/1Check this question

Citizen askedProvide the approved category-wise cut-off document used for the 2026 selection.

Reply passage“Approved 2026 selection cut-off — Demonstration Category B: 74 marks. Approved by the Fellowship Selection Committee on 14 May 2026.”

Record checked: maya-cutoff-reply.pdf
Page / location: Attachment 1 · Page 1
Registration: DEMO/CFSB/R/E/26/00421/1
Record date: 2026-06-21

What this record showsThe approved category cut-off and approval date appear in the attachment.

Open watermarked sample PDF (new tab)

Did this give you the information you needed?

Your answer, not a status chosen by the website.

YesThis answers my questionNoI still need information

Q3Seat calculationDEMO/CFSB/R/E/26/00421/2Check this question

Citizen askedProvide the calculation sheet or approval record showing how the total number of seats was determined.

Reply passage“The final fellowship notification records a total of 120 seats. The notification is enclosed as Annexure A.”

Record checked: maya-vacancy-reply.pdf
Page / location: Page 1 and Annexure A
Registration: DEMO/CFSB/R/E/26/00421/2
Record date: 2026-06-26

What this record showsThe reply states the final total, but it does not provide the requested calculation sheet or approval record.

Still not found in the recordNo calculation sheet, formula, file noting, or approval record explaining the total was located.

Open watermarked sample PDF (new tab)

Did this give you the information you needed?

Your answer, not a status chosen by the website.

YesThis answers my questionNoI still need information

Proof, not vibes

Because “open another tab” is not a solution.

Official guidance and published files show several registrations, transfers, replies and branch-specific appeals.

The missing viewWhich branch answers each original question—and what can the citizen do next?

Official guidance

FAQ and citizen manual

Multiple registrations, separate status views, replies and branch-specific appeals.

Source 01Current Central RTI FAQ
One request can create several registration numbers.
The official FAQ says one application may be sent to several CPIOs, creating several registrations.
States the rule; it does not measure frequency.
Source 02Citizen manual · page 24
Branches sit behind another status view.
The citizen manual puts branch registrations and statuses in a separate table.
An instructional example, not a usability study.
Source 03Citizen manual · page 25
Each branch can have its own reply and appeal.
The manual expects four replies and ties an appeal to the relevant branch number.
Shows the workflow, not how often it occurs.

Published case records

AAI and TRAI examples

Real records show transfers, a chain and one reply citing two registrations.

Source 04Published TRAI record · 2025
A transferred branch gets another registration.
The TRAI record points back to the numbered DoT branch from which it was transferred.
One redacted example; transfer patterns can differ.
Source 05Published AAI record · 2024
The case becomes a chain of transfers.
AAI logs receipt, forwarding, two CPIO transfers and disposal as separate dated rows.
One published case, not a frequency measure.
Source 06Published TRAI reply · 2025
One reply can cite two registrations.
A redacted TRAI reply cites two transferred registrations in one response.
One consolidated reply, not two replies.

What these records establishBranching can occur. They do not measure prevalence, prove unlawful action or endorse this prototype.

What works. What's next.

Working now

Five cases, dependency tree, question checks, inline next-step notes and local JSON.

Not connected

Government filing, payment, live status, email or SMS alerts, document extraction and legal decisions.

Next week, if shortlisted

Round two: finish the whole journey.

01
First and second appeals
Prepare complete files, preserve branch links and hand off to the correct official route.
02
Email and SMS alerts
Remind citizens about replies, appeal actions and status changes after verified dates.
03
Officer/admin view · if time
Explore an optional work queue only after the citizen flow is complete.

For real use: consent, redaction, encryption, deletion controls, security testing and legal review would be required.

Bring your own plot twist

Your RTI chaos, now with a map.

ChatGPT prepares the JSON separately. This static site only checks and renders it in your tab.

1
Copy promptAdd redacted questions and records in ChatGPT.
2
Copy the JSONChatGPT returns one JSON object.
3
Paste or uploadCheck it, then load the tree and Reply Map.

Step 1 · Prepare JSON

Give ChatGPT the boring formatting job.

Redact first

Remove names, contact details, identity numbers, signatures, bank details and other personal information.

Separate service: ChatGPT is not called by this website. Review its privacy terms before sharing anything.

View the full prompt · optional

You are preparing one redacted RTI case for a local visualisation called RTI Reply Navigator.

PRIVACY FIRST
- Before doing anything, tell me if the supplied material still contains names, addresses, phone numbers, email addresses, Aadhaar or other identity numbers, signatures, bank details, or any personal identifier. Ask me to redact it. Do not repeat those details.
- Use "Citizen" and neutral labels when an identity is not needed.

OUTPUT RULES
- After the records are safely redacted, output exactly one JSON object and nothing else. Do not use Markdown fences.
- Use schemaVersion "1.0" and source "custom".
- Set fictional to false for a real redacted case. Use true only when every record and person is synthetic.
- Never invent a date, registration number, office, relationship, passage, page, or missing fact. Use a cautious neutral label when a display field is unknown.
- Keep the structure a dependency tree: one root application; every other node has exactly one parent; no cycles; every ID is unique; every reference points to an existing item.
- Model each registration, transfer, reply, no-reply observation, fee notice, payment, appeal, appeal order, and supplemental reply as its own dated node when the record supports it.
- A transfer notice, fee notice, or appeal order is procedural. Never treat it as evidence that an original information request was answered.
- Create exactly one mapping per original question.
- For answer_located or partially_addressed, include documentId, an exact short passage, and a precise page or attachment location. Do not paraphrase inside passage.
- Use no_matching_passage only when a substantive reply was actually inspected and contains no relevant passage.
- Use needs_human_review when no substantive reply is available or the evidence is insufficient.
- An explicit dated statement that no record exists may be answer_located; preserve the date/time qualifier and explain that this is an answer passage, not proof of legal compliance.
- Do not make a legal-compliance verdict and do not advise whether an appeal must be filed.
- Do not include assetPath or embed files/base64 data.
- Before returning JSON, check every exact passage, page, date, registration number, and relationship against the supplied redacted records. The RTI Reply Navigator website validates structure and references; it cannot inspect the source PDFs or verify factual accuracy.

ALLOWED VALUES
{
  "schemaVersion": "1.0",
  "source": "custom",
  "nodeKinds": [
    "application",
    "registration",
    "transfer",
    "reply",
    "no_reply",
    "fee_notice",
    "payment",
    "appeal",
    "appeal_order",
    "supplemental_reply"
  ],
  "edgeKinds": [
    "registered_as",
    "transferred_to",
    "split_to",
    "replied_with",
    "no_reply_observed",
    "fee_requested",
    "fee_paid",
    "appealed_as",
    "ordered",
    "supplemented_by"
  ],
  "documentKinds": [
    "substantive_reply",
    "supplemental_reply",
    "transfer_notice",
    "appeal_order",
    "fee_notice",
    "attachment"
  ],
  "coverage": [
    "answer_located",
    "partially_addressed",
    "no_matching_passage",
    "needs_human_review"
  ],
  "confidence": [
    "high",
    "medium",
    "low"
  ]
}

SHAPE EXAMPLE
Replace every example value using only the redacted records. Keep the same field names and omit optional fields only when they do not apply:
{
  "schemaVersion": "1.0",
  "caseId": "redacted-example-case",
  "source": "custom",
  "fictional": true,
  "title": "Redacted RTI case title",
  "citizenName": "Citizen",
  "citizenGoal": "Understand which supplied record answers the original question.",
  "scenario": "One application, one registration, and one reply.",
  "painPoint": "The requested record is difficult to locate inside the reply package.",
  "filedOn": "2026-01-10",
  "authority": "Redacted public authority",
  "rootNodeId": "application-1",
  "structureLabel": "Single registration and reply",
  "tags": [
    "single reply",
    "redacted"
  ],
  "questions": [
    {
      "id": "q1",
      "number": 1,
      "title": "Requested record",
      "text": "Provide a copy of the requested record."
    }
  ],
  "nodes": [
    {
      "id": "application-1",
      "kind": "application",
      "title": "Original RTI application",
      "summary": "The redacted request as filed.",
      "date": "2026-01-10",
      "status": "Filed",
      "questionIds": [
        "q1"
      ]
    },
    {
      "id": "registration-1",
      "kind": "registration",
      "title": "CPIO registration",
      "summary": "The application was registered by the responsible office.",
      "date": "2026-01-12",
      "status": "Registered",
      "office": "Redacted office",
      "registrationNumber": "REDACTED/RTI/001",
      "questionIds": [
        "q1"
      ]
    },
    {
      "id": "reply-1",
      "kind": "reply",
      "title": "Substantive reply",
      "summary": "A reply document was received.",
      "date": "2026-02-05",
      "status": "Reply received",
      "registrationNumber": "REDACTED/RTI/001",
      "questionIds": [
        "q1"
      ],
      "documentIds": [
        "document-1"
      ]
    }
  ],
  "edges": [
    {
      "id": "edge-1",
      "from": "application-1",
      "to": "registration-1",
      "kind": "registered_as",
      "label": "Registered as"
    },
    {
      "id": "edge-2",
      "from": "registration-1",
      "to": "reply-1",
      "kind": "replied_with",
      "label": "Reply received"
    }
  ],
  "documents": [
    {
      "id": "document-1",
      "title": "Redacted reply",
      "kind": "substantive_reply",
      "fileName": "redacted-reply.pdf",
      "registrationNumber": "REDACTED/RTI/001",
      "issuedOn": "2026-02-05"
    }
  ],
  "mappings": [
    {
      "id": "mapping-q1",
      "questionId": "q1",
      "nodeId": "reply-1",
      "documentId": "document-1",
      "registrationNumber": "REDACTED/RTI/001",
      "coverage": "answer_located",
      "passage": "Insert an exact short passage from the redacted reply here.",
      "location": "Page 1, paragraph 2",
      "confidence": "high",
      "explanation": "Explain factually how this passage addresses the question."
    }
  ]
}

Now wait for the redacted RTI application, registration/transfer notices, replies, fee notices, appeals, appeal orders, and attachments. Then return the single JSON object.

Step 2

Drop the JSON. Watch the case organise itself.

Checks structure and links, not source documents. Verify every passage, page, date and registration yourself.

Choose a JSON fileStays in this tab · maximum 512 KBCase JSON

Making RTI smooth as butter

Pick your paperwork adventure.

Fellowship selection records

Find the records behind her fellowship result.

Because “open another tab” is not a solution.

One request can create several registration numbers.

Branches sit behind another status view.

Each branch can have its own reply and appeal.

A transferred branch gets another registration.

The case becomes a chain of transfers.

One reply can cite two registrations.

What works. What's next.

Your RTI chaos, now with a map.

Give ChatGPT the boring formatting job.

Drop the JSON. Watch the case organise itself.