Solutions · Private equity operators
Operate the portfolio, not just the deal.
A business harness uses AI to run one company’s operations. A private equity harness runs every company in a portfolio — and the portfolio itself. Each platform company becomes a governed workspace with an AI manager, a memory, a ledger and an audit trail; the operator gets a control plane that funds, prices and reads balance and burn across the book. Nothing irreversible runs without a human yes.
01 — What changes on day one
Three things happen on day one. Each one leaves a row.
A company is created, funded and priced, and its first diligence is ordered — from one console, in one sitting.
01
Provision a claimable company
From Admin → Accounts, one action mints a real sign-in, an isolated workspace, an owner seat, a credit grant and a shareable link with an access password. The operating lead claims the workspace with their own email, and the credits stay with the company. Partner seats provision client workspaces from templates with starter credits.
02
Fund it and price it
Gift credits by workspace, and read the live price sheet every company meters against: a full-depth research report is 800 credits, a legal draft 250, a lead 8 per lead found, a daily briefing 10. Edit a price and every workspace reads the new one.
03
Order the first diligence
Order a Competitive Landscape Analysis on the target — 800 credits at full depth, 1 to 50 pages, cited, in minutes with a live ETA — and a Private Investigation on its founder for a stated business purpose: 1,200 credits standard, 2,800 deep, with a red-yellow-green risk panel. A failed run refunds in full.
02 — Operating
Each company gets its own manager, and the manager gets a mandate.
One loop per company: a mandate, objectives with an envelope, plans drawn as flow charts, a five-minute patrol, a nightly memory consolidation, and a trust ladder climbed one graded decision at a time.
Objectives and plans you approve
The mandate is versioned: mission, brand voice, a never-do list, reserved matters and the minutes per week the manager may ask of the board. Objectives carry a north-star metric, key results and a share of the envelope. State an intent and the manager drafts the plan as a flow chart, every step bound to a registered action with its risk class and credit estimate. Approve, request changes, or decline with a reason that becomes a lesson. Safe steps fire; anything that contacts the outside world or spends beyond the envelope waits at its gate. A drafted plan costs 15 credits; one of the 8 command playbooks costs nothing.
The tactical floor
Operation lanes, a diary of executed actions in plain sentences, an AI receipt in cents on every job, and the five-minute patrol — a SQL check of tripwires, error spikes and KPI drift that usually costs zero tokens. Pause everything halts the manager within one tick.
| Level | Rung | What it may do |
|---|---|---|
| L0 | Watching | Shadow mode. It reads, remembers and proposes nothing. |
| L1 | Suggesting | Plans and proposals land in the decision queue; nothing runs until you say go. |
| L2 | Acting within budget | Safe steps run inside the envelope; money and irreversible steps still wait for a click. |
| L3 | Fully trusted | Expanded autonomy, earned through graded decisions and Brier calibration — demotion is automatic. |
03 — Diligence and capital
Diligence targets and price companies on the same desk.
A thesis is saved before anything hunts; a valuation is computed before anyone narrates it.
The Investment Thesis and the Wolf Hunt board
The thesis holds the mandate in machine-readable form — sectors, geographies, size bands, deal types, must-haves, red lines, target IRR and scoring weights — and nothing on the buy side runs until it is saved. The Wolf Hunt board moves targets Scent → Stalk → Ambush → Feast, then Eaten or Passed; a database constraint refuses illegal moves, and each card carries its fit signals with sources.
The DILIGENCE ENGINE™ and Private Investigation
15 orderable report types: 12 web-research papers — Executive Business Summary, Competitive Landscape, Investment Analysis, Risk, Executive Diligence, Game Theory, Pricing and more — at a depth of 1 to 50 pages, plus data-grounded internal reports at 400 credits, each with a bibliography and a machine-readable appendix that feeds memory. Private Investigation researches a person from the public web only, for a business relationship, fraud prevention or litigation research; employment, tenancy, credit, insurance and licensing purposes are refused server-side.
Capital Command, the Valuation Workbench and Data Rooms
Capital Command scores a company’s readiness for capital deterministically — no language model — from the reports it holds, its memory floors, its financial intake and its valuation scenarios; every unearned point is a named gap with the link that closes it. The Valuation Workbench prices the company by DCF, multiples and unit economics, with round and SAFE modelling, a sensitivity grid, a football field and an assumptions ledger that names the source of every driver. A model may narrate the result; it may not produce a figure.
A Data Room reuses the reports the company already holds, opens by a token link with an optional password and expiry, stamps a per-visitor watermark, and shows which sections were read, for how long, and who came back.
04 — Oversight
Read the book from one plane.
The Admin plane is the cross-company read: a tenants ledger with balance, reserved and 30-day burn per company, a cross-company audit, a global broadcast and Mission Control.
| Operating task | Surface | What is minuted |
|---|---|---|
| Stand up a platform company | Admin → Accounts | The provisioning row, the credit grant, the claim by the operating lead’s own email. |
| Fund, top up, or reprice | Admin → Credits & Pricing | A gift row in the append-only ledger with actor and reference; a new price row, latest wins. |
| Mandate, plans and gates | Strategic Objectives | Each ratified mandate version, the approved plan graph, every gate release and every decline with its reason. |
| Order diligence on a target or a person | DILIGENCE ENGINE™ · Private Investigation | The report row and bibliography, reserve → settle, the purpose attestation with who signed it and when. |
| Hold the acquisition mandate | Investment Thesis · Wolf Hunt board | The saved thesis; every stage move, checked by the database. |
| Share with investors | Data Rooms | Gate passes, section views, dwell, downloads and questions, per visitor. |
| Read balance and burn across companies | Admin → Tenants | Balance, reserved and 30-day burn per workspace; CSV export. |
| Audit across companies | Admin → Audit | Every action by a person, the manager or an integration, by workspace, actor, action and date. |
Roadmap Roadmap
What the control plane does not read yet.
The four surfaces below are not built. They are built on request, on the same substrate — see Consulting.
Cross-company cockpit
RoadmapObjectives, manager status, spend and risk for every company on one board.
In-app company switcher
RoadmapStep between the companies you are seated in from the header.
Bulk actions
RoadmapThe same report, gift or flag across many companies in one audited action.
Team invitations
RoadmapInvite colleagues into a company from inside the product; today the owner seat is created at provisioning.
05 — The departments you get
Inside every company, the same departments.
An MCP server with 8 tools lets the operator’s own agent drive any company through the same audited registry; holdcos that resell seats get the partner console, with a wholesale statement at 50% / 45% / 40% of retail by tier.
Daemon Operator
Strategic Objectives, Tactical Operations, the Monday pack, the kill switch and the trust ladder.
Department of Intelligence
DILIGENCE ENGINE™ with 15 report types, purpose-gated Private Investigation, the Legal Department with 26 document types, the Executive Daily Briefing and Event Finder.
Fundraising & Investments
Capital Command, the Valuation Workbench, Investor Finder from live web research, token-gated Data Rooms, the Investment Thesis and the Wolf Hunt board.
B2B DealFlow
Lead Generator at 8 credits per lead found, the Lead Database, outreach from the company’s own mailbox, RFP Radar, and the CRM.
Social & Online Presence
Social Media Poster drafting for 12 platforms, Brand Radar across 14 source types, a customer Chatbot trained on Company Memory, classifieds drafts.
Company Memory
14 floors with a 3D Memory Stack, Theory of Mind on rivals and counterparties, Brain Dump for the company’s own records, and a portable Mind File that leaves with the company when it is sold.
06 — Governance
Governance a board can sign off.
Enforced in the database and the action registry, not in a prompt, and re-checked in CI.
Money law
The price is shown before every metered run; the run reserves it, settles at the actual cost and releases the rest. A failed run refunds in full; an overrun is absorbed, never charged. No language model sits in the money path.
Audit across companies, append-only
Update and delete on the audit trail are revoked at the database grant layer — even for the service role — and blocked again by trigger. Root reads every company’s trail; a company reads its own.
Isolation and the human gate
Row-level security on all 167 tables, 183 policies; a CI tripwire refuses any policy that re-grants a cross-tenant read. Every action carries a risk class the server decides; an agent cannot skip the ceremony by omission.
Sealed secrets, in-house AI
Keys and tokens are sealed before they reach the database. AI runs on in-house keys across 8 providers — Anthropic, OpenAI, Google, xAI, DeepSeek, Qwen, Moonshot, Z.ai — with per-prompt model pins. No portfolio company manages a provider account.
Nothing irreversible runs without a human yes.
07 — What it costs
One platform company’s first month, priced from the sheet.
Credits are granted and metered per company. 1 credit is 10¢ CAD; partners pay the wholesale tier.
| Item | Where it runs | Credits each | Quantity | Credits |
|---|---|---|---|---|
| Research reports on the target, full depth | DILIGENCE ENGINE™ · cited PDF | 800 | 4 | 3,200 |
| Private Investigation on a founder, standard | Purpose-gated · risk panel | 1,200 | 1 | 1,200 |
| Background check on a counterparty | Business-relationship purpose | 600 | 1 | 600 |
| Legal drafts: NDA, SOW, advisor agreement | Legal Department | 250 | 3 | 750 |
| Executive Daily Briefing, one per day | Grounded, every item linked | 10 | 30 | 300 |
| Plans drafted by the manager | Strategic Objectives | 15 | 4 | 60 |
| Month, total | 6,110 | |||
| Scale plan | $749 a month · 1 credit = 10¢ CAD | 12,000 | ||
| Scale covers it with 5,890 credits left. | 5,890 |
Plans: Core $149 a month for 1,500 credits · Growth $349 a month for 5,000 credits · Scale $749 a month for 12,000 credits. Packs: 500 for $45 · 2,000 for $159 · 8,000 for $549. The full live price table is published inside every workspace; detail on the pricing page.
Where next
Other doors into the same harness.
Related
Existing businesses adopting AI
What a platform company with a working stack keeps, and how its agents drive it.
Related
Business owners
The research department and ops floor inside each company you own.
Related
Partner programme · PE firms
Run the harness as your firm’s operating layer, on wholesale credits.
Next step
See the harness on your own portfolio.
A private demo is a real workspace, provisioned with credits and an access password — the same action you will use to provision your first company.