Kohēsis Enrichment Platform

·Sessions

Enrichment sessions

Create a new session, or resume one in progress. Each session saves as you go.

·Access & Security

Access & Security

Enter the admin access code to manage people and permissions for your institution.

·Kohēsis Control

Kohēsis Control

The control plane: master-graph curation, the ontology, and the enrichment method. Staff only.

Master graph curation

Signed-off papers wait here for a steward to commit them into the shared master graph. You decide what enters each domain's graph, and you can walk any commit back.

Awaiting commit —

In the master graph · committed papers you can walk back

AI models · chosen here only; researchers never pick a model

Enrichment & analysis Graph building Model the repair step tests against

Before a scholar is asked to repair anything, each loaded term is put to this model and only the ones it gets wrong are asked about. Set it to the tier of model the graph is meant to serve, not the strongest one: a frontier model already reads most published terms correctly, so testing against it would ask almost nothing and leave the graph thin.

Applies to every request from the next one on. Opus 5 and Fable 5.1 run with automatic fallback, so a request their safety filters decline is retried on another model instead of failing.

Security · applied to every account

Require two-factor authentication for every account

On (recommended): anyone without two-factor sets it up at their next sign-in, and the platform stays closed to them until they do. Turning it off doesn’t remove anyone’s two-factor.

Graph settings · applied to every user & every new paper

Merge duplicate concepts Similarity threshold Hide unconnected concepts

Build against a draft ontology

For a module still under review. A graph built this way records its version as …-draft, so it never claims to conform to a published one.

Lower threshold merges more aggressively (0.80 recommended). Dedup applies to papers built from now on; rebuild a paper to re-apply a new threshold. Hiding unconnected concepts is applied live.

01New session

Upload your paper

Kohēsis reads the full text and prepares your enrichment. Nothing in the paper is changed.

↑

Drop a PDF, or click to browse

Kohēsis reads the whole paper and detects the language.

PDF · up to ~50 pages

Stored privately in your Kohēsis workspace.

02Repair possible misreadings

Where today’s AI may misread this paper

Kohēsis read your paper. These are places a current AI model could flatten your meaning. Put each one right in your own words. Your corrections become the context for every enrichment question that follows.

03The enrichment ahead

Set the context AI gets wrong

These seven layers are the parts of cultural context an AI model routinely misreads. Left to fill them in on its own, it guesses, and its guesses become confident claims about your community that no one in it would recognize. You are the authority that prevents that. Questions are prepared by Kohēsis as part of the enrichment process; every answer is yours. They come one at a time, each chosen from your paper and your earlier answers. Answer in your own words, or say when something isn’t known, doesn’t apply, or rests on a premise that doesn’t hold. Finish whenever you’ve said what you can.

Enrichment

Manuscript

·

Kohēsis

Your response

Anything you type above is kept with the option you choose.

Structured for the graph

04Review

Your enriched manuscript

Your paper, with every contribution set in the margin beside the passage it deepens. The manuscript stays yours. Review it, then sign to lock.

Author Sign-off

I confirm these enrichments reflect my expertise and that I retain authorship of this contribution. I authorize Kohēsis to structure this enriched record for inclusion in the Kohēsis knowledge graph, and to publish and license it under my institution’s agreement.

05Knowledge graph

Built from the paper, with your notes attached · click any concept to see its notes and passages · scroll to zoom, drag to pan

◱

Turn this contribution into a knowledge graph

Kohēsis builds two graphs — the paper on its own, and the paper with your enrichment — so you can see and interrogate exactly what your expertise added.

Connection settings

New session