AI products
Analyst-built tools. Human-owned judgement.
JMN builds AI tooling to compress the mechanical parts of research — extraction, formatting, first drafts — so named analysts spend their time on judgement. Every product is designed around one rule: AI accelerates the work, it never signs off on it.
Product 01
JMN Excel Assistant
An in-workbook assistant that pulls figures straight from filings and transcripts into your live models — your fields, your layout, your formulas — cutting hours of manual data entry from every earnings cycle.
Extracts line items directly into your templates
Flags period-over-period changes automatically
Every figure traceable to its source document
Product 02
JMN DraftPilot
A drafting copilot for morning notes, earnings reviews, and initiation sections — producing structured first drafts in your house style that a named analyst then edits, validates, and owns before anything ships.
First drafts in your format and voice
Analyst edits and signs off on every output
Full audit trail of AI-drafted vs. analyst-written
Product 03
Research Automation & AI
The workflow layer that connects extraction, drafting, and delivery into one pipeline — filing ingestion, NLP pre-processing, and formatting automation that shorten turnaround without touching the analyst's judgement.
Automated filing & transcript ingestion
NLP surfaces anomalies for analyst review
Delivery to your systems, before the open
How AI fits into the process
Structured. Reviewed. Accountable.
Every deliverable follows the same five-step process, whether it touches AI or not. AI accelerates extraction and structuring — it never writes the judgement that goes out under your name.
01
Data ingestion
Filings, transcripts, and pricing data sourced from your existing systems pre-market
02
AI pre-processing
NLP flags anomalies, extracts key figures, surfaces changes vs prior period
03
Analyst judgement
Named analyst validates, interprets, and writes — AI assists, humans decide
04
Quality audit
Second analyst data check · Calculation audit · Senior sign-off
05
Client delivery
Your format · Your systems · Your inbox. Before the opening bell.
AI, used honestly
What the AI does. What it never does.
AI-enabled outsourcing has a credibility problem industry-wide — confidently wrong outputs, unclear audit trails, vendors who can't say which tool touched what. We answer the four questions a sophisticated research head will ask before trusting any AI-assisted deliverable.
01
AI never makes the call
AI extracts, flags, and structures. It never writes a recommendation, sets a target price, or finalises a thesis. Every judgement-bearing sentence is written and owned by a named analyst.
02
Full audit trail, every time
Every deliverable carries a record of what was AI-extracted versus analyst-written. If a client's compliance team asks "what did the AI touch," we have the answer in writing.
03
Your data never trains a model
Client data, filings, and proprietary methodology are never used to train or fine-tune any AI system — ours or a third party's. Stated explicitly in every engagement agreement.
04
You own every output
All IP — AI-assisted or not — belongs to you on delivery. We retain no rights to republish, reuse, or reference any deliverable produced under engagement.