Cognitive Classification Engine

Humans, it turns out,are predictable.

We classify how everyone works — and predict what they’ll do. From a YouTube link, a transcript, a CSV, or structured data via API.

How it works →Request access

Product walkthrough

The Vision

Primitive of the next decade

The protocol layer
for human cognition.

Location has GPS. Proximity has Bluetooth. Identity has OAuth. Cognition has nothing.

Every model, every agent, every CRM, every recruiter, every ad system is operating without one. Mahakram is the primitive — a single API call that returns the cognitive architecture of whoever is on the other side of the screen, derived from observation, not self-report.

Architecture

Behind the classification.
Three stages, signal to type.

STAGE 01

Fingerprint.

A multimodal pipeline ingests video, audio, text and behavioral signal, and compresses it into a single numerical fingerprint before any model reads it.

STAGE 02

Triangulate.

Vector embeddings are cross-retrieved against a corpus hand-typed by trained operators — multiple independent similarity layers, validated where they agree.

STAGE 03

Resolve.

A proprietary elimination algorithm collapses the 2,048 candidate space to one, with confidence scored per dimension.

Proprietary elimination2,048 / 2,048

Resolved—

Real classification

One interview. This is what comes back.

The dossier below was generated from a single public interview. Below it, what the person we classified said after reading their own dossier.

MahakramFULL DOSSIER

Read the full dossier↗Download↓

Slawomir Tulski

linkedin.com/in/slawomir-tulski↗

BASED ON: 1 video · 1h 12m public interview
DATED: MAY 8, 2026

The architecture

He has already figured it out before the conversation begins, and his job in the room is to transmit the model, not to discover it. The conversation is a delivery mechanism for a conclusion that was reached privately and is no longer up for revision.

Selected highlights

3 of 14

Blind spot

He cannot read the room in real time

He reasons about what people should feel instead of registering what they actually feel. Under emotional pressure he reads as cold or absent. Not by choice — it is the part of him that does not show up.

01:18:42

“I just don't see why that would be the issue here.”

What his team says when he leaves the room

“He has already decided. The meeting is theatre.”

Almost word-for-word, predictable. It happens whether the decision was right or wrong. The complaint is structural — produced by how he thinks, not by how he behaves.

01:32:11

“I'd already worked through the alternatives by the time we sat down.”

What's coming in 6–12 months

One quiet rupture he won't see coming

On the current trajectory — no real breaks, no play, no time off-task — one important relationship goes silent on him. Not a fight. A slow withdrawal. He will think it came out of nowhere.

01:51:08

“I don't think I've taken a real break in… I don't know how long.”

9 pages · Architecture · Blind spot · Tribe · Forecast · PlaybookOpen full dossier↗

The playbooks

What else you can generate from a single classification.

Negotiation

What they'll push for, where they'll fold, and the blind spots you can use.

Evaluation

Where they'll hit a ceiling, and the failure mode nobody warns you about.

Pitch

What lands, in what order, and what they'll reject on instinct.

Managing

What motivates vs depletes them, where they need cover, and what breaks them.

Ask anything

“How will this person react if their funding falls through?”— or any question you can write.

API · early access

The classification engine,
as a single API call.

Pipe in any behavioral signal — a video, a transcript, a CSV of browsing data, structured records. Get back 9 binary dimensions of cognitive architecture, per-dimension confidence, and an evidence chain.

Classify once. Then generate tactical briefs for any scenario — negotiation, hiring, pitch, management — without reclassifying. One expensive call, unlimited cheap ones.

See the API →Talk to us

POSTapi.mahakram.in/v1/classify

early access

// Request
{
  "subject_id": "sub_4601136C",
  "sources": [
    { "type": "video", "url": "https://cdn.example.com/interview.mp4" },
    { "type": "csv", "url": "https://cdn.example.com/browsing.csv" }
  ]
}

// Response
{
  "type": "Ni/Te BS/C(P)",
  "confidence": "HIGH",
  "coins": {
    "observer": { "value": "Ni", "p": 0.91 },
    "decider": { "value": "Te", "p": 0.94 },
    "energy_animal": { "value": "Blast", "p": 0.88 }
    // ... 6 more dimensions
  },
  "dossier_url": "https://app.mahakram.in/d/4601136C"
}

Built by

Solo founder

Dossier·Founder

Subject 001

Abhas Singh

Obsessed with the human mind, technology, and the pursuit of fundamental truths. Building systems to find the pattern in human randomness.

Previously

Founder's Office·Humantic AI

Personality AI · Product Marketing, Program Management & Strategy

Founder's Office·jhana.ai

Legal AI · B2G strategy, Research

Founder·ClarityCuts

Agency · Major Indian brands + international clients

LinkedIn↗𝕏↗

Request access

Closed beta.Founders only, for now.

We’re onboarding a small batch each week. Tell us what you’d use it for — co-founder bets, senior hires, investor reads, or yourself. Specific use-cases move first.

abhas@mahakram.in · Bengaluru