01 · Evidence
Know what supports the deal.
Connect source documents, checked calculations, and the assumptions behind the analysis. Inspect discrepancies and see where evidence is missing.
Explore the evidence recordJudgment you can trust.
Source-linked evidence, checked calculations, and accountable human review — with uncertainty kept visible.
OloLand is the AI execution and institutional intelligence platform for private equity — connect financial work, professional judgment, and a durable system of record.
Saguaro Gardens Apartments · deal chat
GIF · loop
Source-linked evidenceExplicit assumptionsHuman approvals
Evidence. Professional judgment. Action.
Know what supports the deal. Apply your firm’s judgment. Move the work forward. Bring evidence, decisions, and delegated work into one connected acquisition process.
02 · Professional judgment
Apply the way your firm invests.
Bring your firm’s criteria and decision frameworks to the deal. Make assumptions, rationale, and unresolved questions visible to the people responsible for the decision.
Explore firm intelligence03 · Action
Move the work forward.
Delegate supported research, analysis, and drafting to agents. Keep task progress, outputs, blockers, and required approvals visible to the team.
Explore agent workflowsFollow the connection from a customer contract to an assumption, a review question, and an assigned task. Our illustrative acquisition walkthrough makes each step visible.
Follow one acquisitionProfessional skepticism, built in.
Frontier models do the reading and the reasoning. OloLand keeps the record underneath — sources, computed numbers, assumptions, contradictions, and approvals — so a judgment can be checked, not just believed.
Source drawer · cited page
The model is the analyst. OloLand is the control system.
Ask the deal. Get a cited answer.
Answers cite their sources. Where a retrieval tool returns a citation map, each claim carries a [N] marker that opens the source document — at the cited page when one is recorded; otherwise the source is named inline.
Ask what the appraisal assumed for rent, and the answer names the appraisal summary and opens it where the figure is stated.
Numbers come from engines, not prose.
DCF, LBO, Monte Carlo, comparables, and real options run as deterministic services. The model reads the output; it never does the arithmetic. Change an assumption and the valuation recomputes with lineage.
Change the exit cap rate and the valuation recomputes. The model reports the new number; it never estimates it.
Risks scored, sourced, and owned.
300+ leaf-level checks across a structured risk taxonomy. Each finding carries its severity and the source it came from — with the page when one is recorded — and, where the evidence supports them, impact, mitigation, and a confidence you can challenge or correct.
A rent finding names each document its figures came from and carries a confidence an analyst can correct on the record.
When two documents disagree, it’s flagged — never averaged.
Seller materials are reconciled against financial statements and source documents. A revenue figure that differs across two uploads becomes a discrepancy for a human to resolve, with both values preserved.
One operating figure stated three ways across the appraisal, the offering memorandum, and the sponsor pro forma is one discrepancy for review, not an average.
Data built in.
Analysis is only as good as its inputs. The public record ships with the platform; your data room is read document by document; every figure keeps its source.
Public record · SEC filings
Precedent transactions, built in.
About 61,000 US M&A transactions announced 2006–2026, with the source filing where one is recorded. Ask who has bought companies like this one, when, and for roughly how much, and cite the filing.
Every search returns a coverage note. Enterprise value is present on roughly half of rows, premium and multiples only where a filing or XBRL discloses them, and a missing value means not disclosed, never zero.
Public targets · EDGAR and XBRL
Filings read at the source.
For a public target, filings are pulled from EDGAR and XBRL facts become financial observations with the filing as their source. What the seller’s materials claim is checked against what the filings state.
A revenue figure from XBRL and the same figure from the CIM sit side by side. If they disagree, both are kept and the disagreement is flagged.
Your data room · per document
Your documents, read one at a time.
Upload folders and ZIPs or import from Google Drive. Each document is extracted on its own, so every figure keeps the document it came from, and two uploads that disagree become a discrepancy for review rather than an average.
Where a figure is stated three ways across an appraisal, an offering memorandum, and a sponsor pro forma, the record shows one discrepancy carrying all three values, not one number.
Coverage is stated, not implied. Public figures are limited to what SEC filings disclose. Private-company financial statements come only from the documents you provide; externally sourced firmographic estimates, such as headcount or a revenue band, carry their source on every field.
One record across the acquisition lifecycle.
What could make us walk away? Surface valuation gaps, document-grounded red flags, and the evidence to demand next.
Artifact · Pre-LOI Screen
See all six stagesDeal cockpit · Screen
Measured, not asserted.
Two published benchmarks with methodology, cohort, and per-case or per-category results anyone can rerun. Both are small, and both say so.
Restatement Recall v1.1
10 of 10
eligible SEC restatements caught from the pre-fraud-disclosure 10-K alone, using the Beneish M-Score, the EBITDA bridge, and the structured risk taxonomy.
n=10 · 95% CI 72%–100% · single seed
Per-case results and methodologyVals AI Finance Agent v1.1 · third-party benchmark
2.7× / 11.8×
less run-to-run variance with OloLand’s deterministic verifier tools on Claude Opus 4.7 / Gemini 3.1 Pro. Mean accuracy +3.4pt / +1.0pt. Same harness, same nine tools, same judge; the model and a per-model prompt nudge change.
Public split, 50 of 537 questions · 3 seeds per condition
Per-category results and methodologyBuilt for lower middle market private equity.
Agents do the work. OloLand preserves the record.
Use whichever frontier model or interaction surface fits the work. OloLand remains the governed acquisition record underneath it.
Security and governance.
Deal data is confidential by construction. What is in place is listed as in place; what is still in progress is labeled.
Encrypted in transit and at rest
In placeFiles and records are hosted on Google Cloud in the United States and encrypted in transit and at rest with Google-managed keys.
One workspace, one tenant
In placeEvery deal, document, calculation, and approval is scoped to your company workspace. Agent surfaces reach it over the same company-scoped rail.
Not used for learning unless you opt in
In placeNothing from your workspace enters a learning dataset unless a workspace admin opts in. Opted-in datasets stay per workspace and are never pooled; cross-firm model training is fail-closed.
Approvals stay human
In placeMaterial approvals are human actions. The agent-facing rail cannot self-approve governed artifacts, whichever model or surface did the work.
Agent runs are on the record
In placeRuns are recorded with provider, served model, tools called, and verdicts, and can be inspected span by span; harness runs can be replayed. A regulator export can be requested for a deal.
SOC 2 Type II — in progress
In progressThe audit is in progress with Vanta. Control status is published on the trust center; nothing here is claimed as certified until the report is issued.
Enterprise terms add SSO/SAML, multi-fund workspace separation, a VPC deployment option, an MSA, DPA, and Standard Contractual Clauses, and negotiated audit-retention requirements. Scope and availability are confirmed during procurement.
Pricing
One subscription. The Pre-LOI Screen and Full QoE are Pro capabilities, generated self-serve.
Pro
$199/ month
7-day free trial · no card
Start 7-day free trialIncludes:
Full web app plus MCP
Up to 3 active deals
25,000 credits / month
Memo, DCF, LBO, comps
Structured risk matrix (300+ checks)
Pre-LOI Screen and Full QoE PDFs
Team
$499/ month
2 seats included · +$199 / seat
Talk to usEverything in Pro, plus:
Up to 10 active deals
75,000 credits / month, shared
Shared data room
Comments, review, memo approval
Institutional learning and firm playbook
Scenario defense
Enterprise
Custom
Multi-fund platforms and regulated institutions
Contact salesEverything in Team, plus:
Unlimited seats and deals
Multi-fund workspace separation
VPC deployment option
SSO / SAML
CRM integration (DealCloud, Salesforce)
Dedicated CSM and custom SLA
Evaluating from Claude or Codex? Developer is free. One deal, 200 credits a month, no card.Install OloLand
FAQs
Is a Pre-LOI Screen or Full QoE an audit?
No. Analytical screening only — not an audit, attestation, or opinion of value. Prepared for the generating account. Liability is capped at fees paid.
What does “it says what it doesn’t know” mean in practice?
A claim without a source is named as such, a figure two documents disagree on becomes a discrepancy with both values preserved, and a subsystem without evidence is marked unevidenced rather than filled in.
Do I have to change where I work?
No. Use the web app, or the same tools and the same record from Claude Code, Claude Cowork, or Codex.
Is anything priced per deal?
No. One subscription. The Pre-LOI Screen and Full QoE are Pro capabilities, generated self-serve.
Make the next business easier to buy.
Start free, evaluate one business, and keep the record when the next opportunity arrives.