Due diligence is a knowledge problem. You gather information, evaluate it, make a decision. The more experience, the better the judgment. That's the standard description of the process — and it's wrong.
Or at least: incomplete.
Because before an experienced transaction manager makes a single assessment, they've already spent hours downloading documents, pulling figures from PDFs, entering data into spreadsheets, chasing stakeholders, reconciling revisions. Work that requires no judgment. Work that exists not because transactions are complex — but because the system that represents them is passive.
Due diligence is not a knowledge problem. It's a coordination problem.
And coordination problems solved by software give people back their actual work.
01 The Silent Archive
At the centre of almost every DD process sits a spreadsheet. That's no accident — Excel is flexible, familiar, and powerful enough for almost any use case.
But Excel has one property that rarely gets said out loud: it's mute.
It can't notice that a new document has arrived. It can't detect that a number on page 14 of the supplementary report contradicts the number on page 3 of the teaser. It doesn't ask. It doesn't warn. It doesn't update itself.
It's an archive — a silent, passive archive that contains exactly what someone has manually entered into it. Nothing more.
SOURCES DATASET
─────── ───────
Teasers ─────────►
Documents ─────────► Excel
Emails ─────────►
Stakeholders ─────────►
Updates ─────────►
Every arrow in this diagram looks simple. It isn't.
02 The Coordination Layer
Behind every arrow stands a person. Someone who downloads the document, reads it, evaluates it, and decides what's relevant. Someone who finds the relevant information, translates it, formats it, and enters it into the right cell. Someone who contacts the stakeholder, waits for a reply, checks the reply, and updates the spreadsheet.
This isn't work that exists because transactions are complex. It's work that exists because the archive is passive.
SOURCES HUMAN DATASET
─────── ───── ───────
Teasers ──────► [capture ] ──────►
Documents ──────► [filter ] ──────► Excel
Emails ──────► [translate ] ──────►
Stakeholders──────► [enter ] ──────►
──────► [follow up ] ──────►
──────► [reconcile ] ──────►
The person isn't between the sources and the dataset because they need to be. They're there because the system doesn't work without them. They are the coordination layer — the link between a constantly changing reality and an archive that can't update itself.
03 The Real Problem
The coordination layer is expensive. But its real problem only becomes visible when reality changes — and in an active due diligence, it changes constantly.
A new report arrives and corrects the vacancy rate. A stakeholder revises their assessment of the roof renovation. An email contains an updated tenant list that changes the aggregated figures in the teaser.
SOURCES HUMAN DATASET
─────── ───── ───────
Document v1 ──────► [entered ] ──────► Excel ✓
Document v2 ──────► [??? ] Excel ✗ (outdated)
New email ──────► [??? ] Excel ✗ (missing)
Revised KPI ──────► [??? ] Excel ✗ (wrong)
Every change creates a new routing job. The dataset is always a few handoffs behind reality. And the more deals running in parallel, the larger the backlog grows.
This isn't a question of diligence or experience. It's a structural property of a passive archive.
04 The Wrong Fix
Since ChatGPT can read documents, at least part of the problem seems solved. Upload the teaser, ask your questions, get the answers. Fast, intuitive, impressive. Anyone who tries it for the first time is rightly excited.
But ChatGPT answers a question about a document — once, in a chat window. What happens when document v2 arrives next week and corrects the vacancy rate? The chat doesn't know. The team doesn't know. Last week's answer is still sitting somewhere in a conversation nobody opens anymore.
SOURCES HUMAN + CHATGPT DATASET
─────── ─────────────── ───────
Document v1 ──► [upload · ask · read]
│
▼ Excel
[manually transfer] ──────────────►
Document v2 ──► [new chat · start over]
│
▼
Old answer: still valid?
Nobody knows.
ChatGPT makes the coordination layer faster. It doesn't solve it. The dynamic problem — a reality that changes and a dataset that lags behind — remains entirely intact.
05 What AI Changes
AI replaces the coordination layer — not the transaction manager.
The distinction matters. It's not about replacing expertise. It's about eliminating the work that requires none: capturing, routing, reconciling, following up, updating.
SOURCES AI LAYER DATASET
─────── ──────── ───────
Teasers ──────► [reads ] ──────►
Documents ──────► [extracts ] ──────► AssetOS
Emails ──────► [answers ] ──────► (live)
Stakeholders──────► [flags ] ──────►
Revisions ──────► [notifies ] ──────►
│
▼
HUMAN
(Judgment · Decision · Depth)
The archive is no longer silent. It reads incoming documents automatically, answers checklist items from document content, flags contradictions, and notifies the team of relevant changes.
The transaction manager moves out of the middle of the process — and into the role they're actually needed for: judging, deciding, going deep where it matters.
06 What This Means in Practice
What AssetOS already handles in the DD workflow today:
Automatically parsing investment memoranda. Relevant metrics are structured directly in the deal view — no manual data entry.
DD checklists as blueprints. Set up once, reusable for every deal, with clear process ownership and full auditability.
Documents organised by deal. No searching through email threads, no scattered shared drives — all relevant information in one place.
What's coming next: We're building the ability for AssetOS to ingest your existing DD checklist and have AI answer the questions automatically from uploaded documents — including flagged red flags and the ability to drill into individual items via chat. Unlike a chat window, the answers land directly in the shared dataset — versioned, traceable, current.
Are you still using an Excel checklist for due diligence today? We'd love to understand how your process works — and whether this feature would make a real difference.
Conclusion
The goal isn't less due diligence. It's better due diligence.
Those who start automating the coordination layer today don't just gain time. They gain capacity for what actually matters: screening more deals, making sharper decisions, identifying risks earlier.
The silent archive has had its day.
This article was written by Matthias Falk, Co-Founder & CTO of AssetOS. AssetOS is an AI-powered transaction management platform for institutional real estate investors in the DACH region.