Mimir
01Product

One record. Every way to ask it.

Everything below reads the same resolved corpus. Nothing here is a separate dataset you have to reconcile, and nothing is behind a plan gate.

02The eight

What people come here to answer.

Eight questions cover almost every session. The rest of the terminal exists to make these eight faster.

03Why it is one product

Eight answers, one resolution pass.

Every screen above is a different question asked of the same canonical record. That is the reason the answers agree with each other — the portfolio view and the concentration view cannot disagree, because they are reading the same resolved person.

30 → 1
screens, all reading one resolution pass
  1. 01One canonical id per building

    Block-and-lot, parcel number and document id collapse to a single record, so a building appears once and only once.

  2. 02One canonical id per person

    A dozen name spellings across a dozen entities resolve to one controller, which is what makes a portfolio a portfolio.

  3. 03Deterministic, not statistical

    The same inputs produce the same ids on every run, so an answer you cited last quarter is the answer you get today.

  4. 04Provenance carried per field

    Citations survive resolution, joins and export — they are not re-attached at the end.

04Everything else

The rest of the terminal.

Every screen, included on every plan. There is no feature gate between you and a question you need answered.

Search/
Ask in plain English

Address, entity or person over one resolved corpus.

Ask/ask
Answered from the record, or not at all

Questions compiled into a query over the corpus.

Map/map
See a market on the ground

Parcels, owners and distress, mapped.

Comps/comps
What traded nearby

Recorded sales around a building, with the deed behind each.

Assemblage/assemblage
Adjacent lots and who holds them

Neighbouring parcels grouped by resolved controller.

Concentration/concentration
Who holds what, where

Exposure by borrower, market and asset class.

Maturity/maturity
Loans coming due

Recorded debt approaching term, by market.

Lenders/lenders
Who is actually lending

Active lenders and recent recorded terms.

Sponsor profile/sponsor
What they own and what they owe

One person, every entity, every building.

Relationships/relationships
How parties connect

Agents, co-borrowers and shared addresses across deals.

Dossiers/dossiers
One cited file

A committee-ready document per building or borrower.

Watchlist/watchlist
Track what matters

Buildings, borrowers and searches you follow.

Pipeline/pipeline
Deals in flight

Your working set, alongside the record.

Lists/lists
Build and keep a shortlist

Saved sets of properties and owners.

Exports/exports
Take it with you

CSV and API delivery with provenance intact.

Investigate/investigate
Follow the chain by hand

Traverse ownership hop by hop, both directions.

Enrich/enrich
Resolve your own list

Upload addresses or entities and get them matched to the corpus.

Compare/compare
Put buildings side by side

Fields, debt and distress across a set.

Worksheet/worksheet
Work the numbers

Underwriting against cited inputs.

Feed/feed
What just recorded

New filings across your markets as they land.

Coverage/coverage
What is in the corpus

Counts by state and source, measured.

Bring a building you already researched by hand.

The fastest way to judge the system is to watch it re-derive something that took you a day, with a citation on every step.