Capital Refinery
The category

A private markets operating system, organized around the decision.

Private capital is the only multi-trillion-dollar asset class without a system of record for the investment decision. Every other layer of the stack — CRM, VDR, monitoring, accounting, AI assistants — is organized around a different primary object. Capital Refinery makes the IC decision the first-class object, and keeps it alive across the life of the position.

See it on a dealRead the research→
Why the existing stack is not the operating system

Each tool is excellent at one slice. None of them holds the position.

The category is not a CRM. It is not a dashboard. It is not a chat window. It is the layer underneath all of them — the structural record of the decision, alive across time.

What teams think they have
What actually happens
01DealCloud and Affinity hold the deal pipeline.
The CRM holds the relationship and the workflow. It does not hold the structured thesis after the deal closes. The IC decision is downstream of the CRM, and the CRM has no schema for it.
02iLEVEL and Allvue hold the portfolio data.
Monitoring platforms hold KPIs and dashboards. They do not hold the assumption the KPI is supposed to test. The connection from operator data to entry case lives in an analyst's memory, not in the data model.
03Datasite and Intralinks hold the diligence record.
The VDR holds the documents during the transaction. The data room expires at close. The most analytically dense work the firm did on the asset becomes inaccessible the moment it matters most.
04FundCount and Standard Metrics hold the financials.
The accounting platform holds the books. It does not hold the underwriting case the books are testing against. Quarterly reconstruction is the dominant pattern — every review starts from scratch.
05Hebbia and BlueFlame layer AI on top of all of it.
Generative AI on a fragmented data layer is a faster way to read incomplete information. It is not a system of record. It cannot test the operator data against the entry case, because the entry case is not in a form the model can test against.
The full stack
6 existing layers + 1 missing
  1. 01
    Market data & screening
    PitchBook · S&P Capital IQ · Preqin · CoStar
  2. 02
    Deal CRM & relationship
    DealCloud · Affinity · Altvia · Dealpath
  3. 03
    Diligence / VDR
    Datasite · Intralinks · Hebbia · BlueFlame
  4. 04
    Generative AI for PE
    Hebbia · BlueFlame · AlphaSense · Endex · Eilla
  5. 05
    Portfolio monitoring
    iLEVEL · Allvue · Chronograph · Cobalt LP
  6. 06
    Accounting & operations
    FundCount · eFront · Standard Metrics · QBO
  7. 07
    Decision integrity
    Missing layer
    Capital Refinery
Every existing layer was designed around a different primary object — document, relationship, dashboard, account. The decision integrity layer is the one organized around the IC decision itself.
STACK · the missing 7th
What the operating system has to do

Three properties no incumbent ships.

01 / 03

The investment decision as the first-class object

Not the document. Not the dashboard. The decision.

The platform schema has a node for the IC decision — the thesis as a structured field, the assumptions as separate testable rows, the conditions as logical clauses with quantitative thresholds. From IC approval forward, that node is the reference point for every operator update. This is the architectural choice the incumbents did not make, because their schemas were built around different primary objects.

02 / 03

Deterministic-first extraction with provenance

The brain is deterministic. The LLM is the adjudicator.

Generative-first stacks cannot guarantee provenance. Capital Refinery uses a regex-and-constructor sweep to produce candidate figures from CIMs, financials, credit agreements, and operator updates. A small local LLM adjudicates conflicts and produces summaries. Every figure has a candidate ID, a method, and a source-page provenance trail. The numbers are reproducible — the LLM is not asked to do the part of the work it cannot do reliably.

03 / 03

A continuous re-test loop across the life of the position

The decision stays alive after IC.

Each accounting feed, KPI report, and operator narrative is mapped to the entry assumption it tests. The connection is structural, not generated. When conditions change enough that the original IC decision is no longer defensible, the platform surfaces it — with the assumption that broke and the time-to-consequence ranking against the rest of the portfolio. Decision validity stops being a memory exercise and starts being a property the platform can compute.

Side by side

Where the lines fall.

CapabilityThe existing stackCapital Refinery
Primary object in the data modelDocument, dashboard, or relationshipThe IC decision
Structured investment record at ICNoYes — bound at approval
Deterministic extraction with provenanceNo — LLM-first or manualYes — every figure has a candidate ID
Operator data → entry assumption mappingNoYes — the data model
Decision validity across the life of the positionNoYes
Covenant + breach probability for credit positionsNoYes
Time-to-consequence rankingNoYes
Survives team turnover as institutional memoryNo — relies on analyst memoryYes — the structured record is the memory
Showcase · 2 min
The war room: every position, ranked by time to act.

The view that shows the entire book sorted by time-to-consequence. The positions running out of room are at the top. The IC review reorders around runway, not colour code.

Recorded on a seeded portfolio of 24 positions across credit and equity. Every score is computed live from the structured investment record.
REC · 02:08
The research behind this

The research behind the category.

The category did not exist a year ago.

Bring us a deal from your portfolio. We will parse the entry pack into a structured investment record, bind the live operator data to the assumptions, and show you the operating system the existing stack is not designed to produce.

Request accessSee how we compare