Pinaka · for institutions & public bodies

Merdot Pinaka

The story on the surface. The one moving underneath.

In India a single post can become a national story within six hours, and by the time it reaches television the decisions that mattered have already been made. Pinaka reads news and social media continuously, groups what it finds into narratives as they form, and shows you who is moving them, where, and what happens next.

Request access How it works

What it is

Pinaka is a Narrative Intelligence OS. It reads news and public sources, clusters coverage into narratives as they form, traces each narrative to its earliest sources, and shows who carried it and where it is heading, with evidence on every claim. Pinaka is in early access.

Visit pinaka.merdot.com

Who it is for

  • Government departments
  • Law enforcement and public safety
  • Public affairs and policy teams
  • Multi-region institutions

Core capabilities

What Pinaka does.

Narrative detection
Related coverage clustered into one narrative as it forms.
Origin tracing
Each narrative walked back to its earliest observed sources.
Trajectory projection
Where a narrative is likely to go next.
Actor intelligence
Behavioural profiles of the accounts and outlets in play.
Propagation graph
How a story moved between sources, and whether the spread is organic.
Geographic spread
Narrative pressure mapped across India.
Entity graph
People, places and organisations resolved against everything connected to them.
Evidence on every claim
Every finding linked to the source material behind it.

How it works

From setup to outcome.

  1. 1

    Ingest

    News and public sources collected continuously.

  2. 2

    Read

    Indian languages handled natively.

  3. 3

    Cluster

    Coverage grouped into narratives.

  4. 4

    Enrich

    Actors, entities and geography attached.

  5. 5

    Surface

    Alerts, investigations and briefs.

Security and data handling

  • Hosted in India.
  • Encrypted in transit with TLS and at rest with AES-256.
  • Role-based access inside each workspace.
  • Administrative actions recorded in an audit trail.
  • Handled in line with our DPDP Act 2023 practices. Customer data is not sold or used for advertising.
  • Tenant isolation at the query layer: every organisation's data is scoped and stored separately.
Trust details

Deployment and support

  • Not self-serve. Each deployment is provisioned and supported by Merdot directly.
  • Early access for a limited number of institutions.
  • Briefs and exports for internal reporting.

Sign in

580,000+
articles in the corpus
61,000
narratives detected
Continuous
ingestion, not batch
Per-tenant
isolation at the query layer

Monitoring tells you what was said. It does not tell you what is forming.

Alerts arrive too late

A keyword alert fires once something is already spreading. By the time it reaches you the response window has closed and you are reacting, not deciding.

Volume is not signal

Ten thousand mentions is not an answer. The questions are which of them belong to the same story, which story is accelerating, and who is pushing it.

Nobody can show their work

A dashboard that asserts a sentiment score without the articles behind it cannot be acted on, and cannot be defended when somebody senior asks why.

How it works

Five stages, running continuously. The work is done before you open the platform, which is why the platform opens instantly.

01

Ingest

News, wires, regional outlets and social run continuously into one corpus. Sources are pulled on their own cadence rather than a single nightly batch, so a fast-moving story is not held back by a slow one.

02

Read

Every item is language-detected, then scored for sentiment, emotion and stance. Indian language content is treated as first-class, not translated to English and flattened.

03

Cluster

Related coverage is grouped into a single tracked narrative. This is the step that turns ten thousand mentions into a handful of things you can actually act on.

04

Enrich

Each narrative gets an origin trace, a propagation path, the actors moving it, its geographic footprint and a projected trajectory.

05

Surface

Results are precomputed into read views rather than aggregated on request, so opening a dashboard is a read, not a wait.

What it can tell you

An intelligence platform, not a reporting tool. Every capability exists to answer a question an analyst actually has to answer out loud.

Narrative detection

Clusters related coverage into one tracked narrative as it forms, not a keyword alert fired after something is already spreading.

Origin tracing

Walks a narrative back to its earliest detectable appearance and shows the chain from there to now.

Trajectory projection

Projects where a narrative is heading over the next day, so attention goes to what is accelerating rather than what is loudest.

Actor intelligence

Behavioural fingerprints for the accounts and outlets in play: who originates, who amplifies, who moves in step with whom.

Propagation graph

Who carried what, when, and to whom. The shape of the spread often says more than its volume.

Geographic spread

State and district level mapping, so a regional flare-up is visible long before it reaches national coverage.

Entity graph

Resolves people, organisations and places across the whole corpus and shows how they connect, with a time scrubber to replay how a network formed.

Evidence on every claim

Each assessment carries its provenance. Open any conclusion and read the underlying articles it was built from.

One morning, hour by hour

An illustration of the shape these things take. The value is not the alert. It is having the whole file already assembled when the alert would have arrived.

  1. 06:40

    A claim appears

    A post in one district alleges something about a local incident. Volume is negligible. Nothing on any dashboard would flag it.

  2. 08:15

    Pinaka groups it

    Four regional outlets and a cluster of accounts publish variations of the same claim. Pinaka groups them into one narrative and starts tracking it.

  3. 09:30

    Shape becomes visible

    The propagation graph shows the spread is not organic: a small set of accounts is seeding it in near-lockstep. Trajectory is projected upward.

  4. 11:00

    You already have the file

    Origin, the accounts involved, the districts affected and every source article are on one screen, before the first national outlet runs it.

Inside the workspace

The surfaces an analyst actually lives in.

Command Center

The daily read. What changed since you last looked, ranked by severity rather than recency, with the reason each item is ranked where it is.

Narrative Explorer

Every tracked narrative with its timeline, volume curve, sentiment split, origin chain and the actors carrying it.

Actor Intelligence

Per-actor dossiers: posting rhythm, topics, stance over time, co-movement with other accounts, and a watchlist for the ones that matter to you.

India Theater

The geographic view. Choropleth by state, drill to district, and open any region to see what is driving its numbers.

Evidence Library

The underlying corpus, searchable. Full text, source, language, images and the narratives each article belongs to.

Ask the corpus

A question box over your own data that answers from retrieved articles and narratives, and cites what it used.

Alerts and investigations

Standing alerts on the entities you care about, and case files that persist an investigation with notes and attached evidence.

Briefs and exports

A written brief of the current picture, and exports when something has to leave the platform as a document.

Built for people who answer for what happens next

Government departments

Departments answerable for a portfolio, who need to know what is being said about it across languages and regions before it reaches them.

Law enforcement and public safety

Teams whose response window is measured in hours, where a rumour spreading through a district is an operational fact and not a media story.

Public affairs and policy

Units tracking how a policy is landing, who is shaping the reception of it, and where the opposition to it is actually concentrated.

Multi-region institutions

Organisations operating across several states and languages, where the story rarely starts in English or in the capital.

Why you can act on what it says

An intelligence product is only worth having if you can defend its conclusions to someone who did not want to hear them.

Nothing without provenance

Every score, ranking and assessment can be opened down to the articles behind it. A number you cannot trace is a number you cannot defend in a meeting.

Tenant isolation at the query layer

Scoping is enforced where the database is read, not by hiding menu items. A missing sidebar entry is never treated as an access control.

Audit trail

Who looked at what, and when. Queryable, because in an institution the question eventually gets asked.

Stated limits

Where the platform is uncertain it says so, and where a source is thin the confidence reflects it. We would rather be trusted than impressive.

Pinaka or Vyuha?

They share a backend and nothing else. Pick by the question you are trying to answer.

Pinaka

For institutions. “What is forming, who is behind it, and where is it going?”

  • Narrative-level, not mention-level
  • Actor behaviour and coordination
  • State and district geography
  • Evidence trails for every claim

Vyuha

For brands and comms teams. “What is being said about us, right now?”

  • Mentions across social, news and reviews
  • Sentiment and share of voice
  • A shared inbox for replies
  • Android app for alerts on the move
See Vyuha
Current status

Live, provisioned per organisation

Pinaka is live at pinaka.merdot.com. It is not self-serve: deployments are provisioned and supported by Merdot directly, and every organisation's data is scoped and stored separately. Ingestion runs continuously against the live corpus, so the figures at the top of this page move every hour.

Access is granted case by case. Tell us what you are trying to see and we will tell you honestly whether Pinaka can see it yet · including when the answer is not yet.

Request access

Questions

Before you request access.

Is Pinaka live?
Yes, at pinaka.merdot.com. Access is not self-serve: each deployment is provisioned and supported by Merdot directly, with every organisation's data scoped and stored separately.
Who is Pinaka built for?
Public bodies, incubators, universities and large organisations with a communications or public-affairs function: anyone who has to answer for where a narrative came from, not only that it exists.
What does it do that Vyuha does not?
Vyuha answers what is being said. Pinaka answers where a narrative began, how it spread, who carried it and what it is likely to do next, with the earliest observed sources and the actors scored on reach, velocity and coordination.
Does it handle Indian languages?
Hindi, Gujarati and other Indian languages are handled natively rather than translated to English and flattened.
How do we get access?
Request access from this page. Tell us what you are trying to see and we will tell you honestly whether Pinaka can see it yet.
Read the Pinaka release page

One ecosystem

Everything else Merdot builds.

Pinaka is part of the Merdot platform for institutions.

See it before it breaks.

The narrative that damages you is already forming somewhere. The only question is whether you find it now or read about it later.

Request access See all Merdot productsSign in

MERDOT PRIVATE LIMITED · 4-D, Vardan Tower · Near Sardar Patel Stadium, Navrangpura · Ahmedabad, Gujarat, India 380009 · contact@merdot.com