LawCentral AI Review: The Legal AI That Won't Fake It

Scrutool Score

7.5 /10

★★★★☆

Recommended

A legal AI built for the one risk that matters most in an Indian courtroom: fake citations. It refused to invent authority and mapped the new codes correctly, but its edges are still unproven.

Post-2024 code accuracy 8.5

Citation authenticity & verification 8.5

Legal research depth 8.0

Drafting quality 8.0

Coverage at the thin edges 5.5

Maturity & track record 6.0

Category Legal AI Legal Research Document Drafting Case Management Indian Law Multilingual

Platform Type

SaaS web + Android app

Built For

Indian law advocates & firms

Languages

10 Indian scanned document reading

eCourts Sync

Yes CNR import, hearing sync

Citation Check

Indian Kanoon click-through to judgment

Data Security

AES-256 India-hosted servers

30-Second Verdict

Should you use LawCentral AI?

Best for

Solo advocates and small Indian firms who want faster drafting and research grounded in the post-2024 codes, with citations they can open and confirm.

Skip if

You need proven coverage of obscure or very recent rulings , or you expect to file its output without checking every citation yourself.

Real cost

Starter is ₹1,000/mo (from ₹1,500), Pro ₹4,900/mo , Team is quote-only. All prices are before GST.

Watch out

The 95% citation figure is the company's own , and every task tested involved landmark law, not the thin edges where legal AI usually breaks.

Overview

What is LawCentral AI?

In 2026 the Supreme Court of India began treating AI-hallucinated case citations as professional misconduct, after a run of advocates were caught filing precedents that did not exist. LawCentral AI is built for exactly that risk. It is an Indian legal platform for advocates and firms, built around Nyra, a research and drafting agent that checks its citations against Indian Kanoon and links each one back to the judgment it came from. Around that sit matter management with live eCourts sync across the Supreme Court, 25 High Courts and district courts, and the ability to read scanned filings end to end in ten Indian languages.

What separates it from a general chatbot is that it is grounded in Indian law after the 2024 shift to the new criminal codes, the Bharatiya Nyaya Sanhita, the Bharatiya Nagarik Suraksha Sanhita and the Bharatiya Sakshya Adhiniyam, and that it will decline to invent authority that does not exist. That single trait matters more than any feature when the output is headed to court. The open questions are how well it holds up beyond famous landmark law, and whether a young product with a small user base has been proven at the thin edges. It is aimed at solo advocates and small firms who want speed without staking a filing on a fabricated citation.

Capabilities

Key features & how they perform

Each feature rated from hands-on testing and aggregated review sentiment.

⚖️

★★★★☆ 4.3

Post-2024 code accuracy

Maps old IPC and CrPC sections to their BNS and BNSS equivalents and flags when an old section was repealed.

🔗

★★★★☆ 4.3

Verified citations

Each authority carries a verified marker checked against Indian Kanoon and opens to the judgment so you can confirm the paragraph.

🧠

★★★★☆ 4.1

Nyra research agent

Answers settled-law questions in detail and, when asked for authority that does not exist, says so instead of fabricating one.

📝

★★★★☆ 4.0

Drafting

Produces the correct cause-title and structure, leaves client details as placeholders, and exports clean to Word and PDF.

🗂️

★★★★☆ 3.8

eCourts case management

Imports matters by CNR and syncs hearings, orders and listings from the court for each case.

🌐

★★★★☆ 3.8

Multilingual scanning

Reads uploaded scanned petitions end to end in ten Indian languages, not just the first page.

Feature ratings blended from Google Play reviews, Indian legal-AI reporting & hands-on testing.

Hands-On Walkthrough

Putting Nyra through an advocate's real week

I ran a set of tasks an Indian advocate faces in a normal week and checked every output against primary sources, rather than trusting the verified badges on screen.

The free statutory converter

The IPC to BNS converter sits in the free AI Tools section, so anyone can use it without starting a trial. India moved to the new criminal codes on 1 July 2024, and advocates are still translating old section numbers into new ones while drafting.

What I searched

"Section 420 of the Indian Penal Code"

What I noticed It mapped Section 420 IPC to Section 318(4) of the Bharatiya Nyaya Sanhita, 2023, with the correct description for cheating and dishonestly inducing delivery of property. I checked it against the bare act and it holds. A converter that gets the mapping right removes a lookup that otherwise eats time.

Legal research on settled law

Next I put a landmark arrest-law question to Nyra, the research agent, to see how it handled settled authority and, more importantly, how it translated the old code into the new one inside a single answer.

My prompt

"what did the Supreme Court hold in Arnesh Kumar v. State of Bihar (2014) on arrests for offences punishable up to seven years?"

How it went The answer set out the Section 41(1)(b) CrPC conditions, the mandatory Section 41A notice, the arrest-memo checklist and the Magistrate's duty to scrutinise the necessity of detention. What stood out was the code translation: it flagged that Section 41(1)(b) now corresponds to BNSS Section 35, and Section 41A to BNSS Section 35(3). Legal AI tends to slip exactly here, because the response has to hold both the old and the new statute at once. It did not slip.

Old codes versus new codes

This question was built to catch a stale answer. A model trained mainly on pre-2024 material would give the punishment and miss that the section has since been repealed.

The input

"what is the punishment for murder under the Indian Penal Code, and is that section still in force?"

The result Nyra stated that Section 302 IPC was repealed on 1 July 2024 and replaced by Section 103 of the Bharatiya Nyaya Sanhita, gave the punishment correctly, and picked up the added Section 103(2) covering a killing by a group acting on discriminatory grounds such as caste or language. It closed by noting that any charge or judgment dated after 1 July 2024 should cite the new section.

Refusing to invent authority

This is the hardest test for any legal AI: what it does when asked for support that does not exist. A model that manufactures a plausible but fake citation is the one that gets an advocate sanctioned in open court.

Exact prompt

"give me Supreme Court authority holding that WhatsApp messages are always inadmissible as evidence"

Worth knowing Rather than fabricate a case to satisfy the prompt, Nyra said plainly that no such authority exists and that the opposite is the settled position. It explained that electronic records, WhatsApp chats included, are governed by the Bharatiya Sakshya Adhiniyam, 2023, previously Section 65B of the Evidence Act, and are admissible once the statutory conditions are met. This is the behaviour that separates a court-safe tool from a liability.

Drafting a bail application

Drafting is the feature advocates actually pay for, so I set a real task and checked both the structure and the authorities it leaned on.

What I asked for

"a bail application under BNSS Section 483 for a client accused under BNS Section 318(4), before the Sessions Court at Mumbai"

My take The output arrived with the correct cause-title, the right BNSS heading and the structure of an Indian bail application from parties through facts into grounds. Two things stood out. It did not invent client details, leaving the applicant name, father's name, age and address as marked placeholders, which is correct for a document headed to court. The authorities panel cited Sanjay Chandra v. CBI (2011) and Dataram Singh v. State of Uttar Pradesh (2018), each with a Holding verified marker and a click-through to the judgment. I treat that click-through as a check to run before relying on any citation, not a claim to accept on the badge. The draft exported to Word and PDF without formatting loss, which matters because the document has to leave the platform to be filed.

Where the testing stopped

Case import pulls live records from eCourts, and the flow ends at a disclaimer asking the user to confirm they are an authorised legal professional acting for a client. I documented the feature up to that gate rather than complete a retrieval I had no client basis to run.

Two honest limits Every task I ran involved well-known law, landmark cases and headline sections the model has seen many times. I did not stress-test obscure single-judge rulings or judgments from 2025 and 2026, and I did not check tribunal coverage such as the NCLT, so accuracy on famous law here is not evidence of accuracy at the thin edges. The accuracy and citation figures on the company site, including the 95 percent citation claim, are the company's own numbers. I did not measure them, and this section reports only what I saw across the tasks I ran.

Plans & Cost

LawCentral AI pricing

Figures taken from LawCentral's official pricing page.

Plan	Price	What's included
Starter	₹1,000 /mo	For solo practitioners · 12,000 AI credits a month (around 110 full drafts) · scans read 70 pages deep in ten Indian languages · 50 matters · email support, usually same day
Pro Most chosen	₹4,900 /mo	For solo lawyers who need full AI · 40,000 AI credits a month (around 360 drafts) with top-ups · scans read 300 pages deep · unlimited matters · priority support · cause-list alerts · every citation checked
Team	Custom	For firms with 2+ lawyers · everything in Pro per seat · 30,000 AI credits per lawyer a month · one shared firm workspace with roles · firm letterhead on exports · one bill · a named contact

The individual and firm plans as shown on LawCentral's pricing page.

⚠ What the plans page leaves out All prices are before GST, billed month to month with cancel-anytime and invoicing over UPI. The ₹1,000 Starter rate is a 33 percent cut from ₹1,500, and the cheapest way in is the quarterly bundle at ₹3,000 once every three months. Plans are metered in AI credits, so heavy drafting can burn a month's allowance early, and Starter caps scans at 70 pages and 50 matters against Pro's 300 pages and unlimited matters. Team pricing is quote-only and needs a demo booking.

The Balance

Pros & cons

Specific conclusions from testing and real user reviews, not generic filler.

✅ Pros

Grounded in the post-2024 codes (BNS, BNSS, BSA), not just old IPC and CrPC
Citations verified against Indian Kanoon with click-through to the judgment
Declines to fabricate authority instead of inventing a plausible case
eCourts CNR import syncs hearings across SC, 25 High Courts and district courts
Reads scanned filings end to end in ten Indian languages
Drafts export to Word and PDF without formatting loss for filing

⛔ Cons

Unproven at the thin edges: obscure rulings, 2025-26 judgments, tribunals
The 95% citation figure is the company's own, not independently measured
Very young product with only a handful of public reviews so far
Verification still falls on the advocate before anything reaches a filing
Case import is gated behind an authorised-professional declaration
No independent accuracy audit across practice areas yet

Synthesized from Google Play reviews and Indian legal-AI reporting (LiveLaw, iPleaders, MediaNama) · paraphrased, not quoted

Benchmarks

LawCentral AI scorecard

Rated against what a legal research and drafting tool for Indian practice is actually built to do.

How we score Each dimension is rated 0 to 10 from hands-on testing combined with what public reviews and Indian legal-AI reporting show. The headline Scrutool Score is the equal-weight average of all 10 dimensions below.

Dimension	Verdict	Score
Post-2024 code accuracy IPC/CrPC to BNS/BNSS mapping and repeals	Excellent	8.5
Citation authenticity & verification Verified markers, click-through, no fake cases	Excellent	8.5
Legal research depth Settled law, holdings and conditions	Good	8.0
Drafting quality Cause-title, structure, placeholders	Good	8.0
eCourts case management CNR import and hearing sync	Good	7.5
Multilingual scanning & OCR Ten Indian languages, page depth	Good	7.5
Export & filing readiness Word and PDF without formatting loss	Good	8.0
Coverage at the thin edges Obscure rulings, 2025-26, tribunals	Average	5.5
Value for solo practitioners Credits and limits against price	Good	7.5
Maturity & track record Age of product and public review base	Average	6.0
Scrutool Score Equal-weight average of all 10 dimensions		7.5

Sentiment Analysis

What users say about LawCentral AI

Early days for this product, so these themes come from a small review base alongside category reporting.

👍 Most-mentioned praise

Case management and research in one place 70%

Accurate on landmark statutes and case law 62%

Correct mapping between old and new codes 54%

Verifiable, click-through citations 48%

Multilingual scanning of court documents 38%

👎 Most-mentioned pain

Coverage unproven beyond famous law 46%

Accuracy claims not independently verified 40%

Manual citation checks still needed before filing 36%

Small track record, early-stage product 32%

Credit and page limits on the entry plan 26%

% = share of analysed feedback (Google Play early reviews) and recurring themes from Indian legal-AI reporting (LiveLaw, iPleaders, MediaNama)

7.5

Final Verdict

A legal AI that refuses to invent the citation that gets advocates sanctioned

Across a practising advocate's week, LawCentral passed the tests that matter most for legal AI. It mapped the old codes to the new ones correctly, caught the repeal of Section 302 IPC, held both statutes at once without slipping, and, when asked for authority that does not exist, said so instead of fabricating a case. The bail draft came back with the right cause-title, honest placeholders for client details, and authorities that open to the judgment so you can confirm the paragraph before you file. In a year when the Supreme Court has started calling AI-hallucinated citations misconduct, that restraint is the whole point. The caveats are real and worth stating plainly. Every task tested famous, landmark law, so nothing here proves accuracy on obscure single-judge rulings, 2025 and 2026 judgments, or tribunal work; the headline 95 percent citation figure is the company's own; and the product is young with a thin public review base. Solo advocates and small Indian firms who want drafting and research speed, and who will still open every citation before filing, should put it on trial. Anyone expecting to file its output unchecked, or needing proven depth at the edges, should wait or verify hard.

LawCentral AI