Lantern
Lantern is a data assurance platform for private markets, used by general partners, limited partners and fund administrators. It sits alongside existing administrators and accounting systems rather than replacing them, collecting fund, portfolio company and investor data, running about 2 million tests a day on every number, and delivering verified data to reporting tools, warehouses and downstream systems, with each figure traceable to its source. Founder Edward Moore previously built the fund administrator Aztec Group, and the team includes people from eFront, Allvue, iLevel and Altvia.
Document extraction comes from Freyda, which Lantern acquired in June 2026 and now runs inside the platform. It turns capital calls, quarterly reports, capital account statements, limited partnership agreements and side letters into structured data, checking each figure against other parts of the same document. Custom rules then test values across systems, exceptions go to the operations team, and validated records flow to Snowflake, Excel, eFront, Investran or Allvue through APIs, secure file transfer and a Model Context Protocol interface. In November 2026 Lantern bought Linga AI to add AI teammates for repetitive data work.
Anonymized case studies cover a €40 billion private equity manager consolidating data across several servicers, a €2 billion growth equity firm and a Nordic venture firm with more than €300 million under management.
Capability Axes
Capability grades
15 of 15 axes rated · 9 graded A or B
AI carries document extraction, where Freyda reads capital calls, statements, quarterly reports and legal documents and turns them into structured data, the workflow limited partners rely on to process what their managers send. Verification runs on configured rules, roll forward checks and cross source comparisons, so general partners and administrators can validate and deliver data that already arrives structured without the AI step.
Incoming figures are checked against configured rules, and each record keeps its source, its verification history and every change. Issues the checks cannot resolve go to private markets specialists for review, and teams can investigate discrepancies, amend values and trace overrides before validated data moves on to downstream reporting.
Each extracted figure links back to the original document, with the checks applied and any adjustment recorded, and a value can be compared against its source and other linked values. Staff can rerun verification and work through exceptions. A reproducible accuracy benchmark for extraction or the AI teammates, or the method for testing performance across fund document types, is not released.
Ronan Shally, a partner and chief financial officer whose firm is not named, is quoted on the platform. Case studies describe a €40 billion private equity manager consolidating fund data from servicers on different accounting systems, a €2 billion growth equity firm and a Nordic venture firm with more than €300 million under management, each anonymized and focused on process changes rather than measured results.
Under the data processing addendum, Lantern may retain, use and disclose customer personal data for product improvement, and it treats product development as one of its own purposes. Whether extracted fund data can train AI models, whether a client can opt out of that use, and how long the two language model providers keep requests are not stated.
A data processing addendum, last updated March 2026 and open to read, sets out customer and Lantern roles, permitted purposes, transfers and incident duties, with a subprocessor annex naming each provider by function and location. Customers can ask for their data back for up to 90 days after termination, after which remaining copies are deleted subject to legal retention, and they can object to subprocessor changes.
Lantern has completed a SOC 2 Type II audit of its controls over access and data handling, with the report and further security detail available on request. The data processing terms add confidentiality duties and notice after a personal data breach.
Lantern sells data verification technology that helps firms prepare and reconcile data for investors and auditors, including amounts from different administrators, and each firm decides how the output serves its own obligations. It works as a technology supplier rather than a registered investment manager or fund administrator, and regulatory filings stay with its clients.
Record lineage, timestamped changes and logged overrides let managers see who changed a figure and why, and unusual data issues go to human review. A named AI management program, an independent assessment of the extraction models, or measured performance across document types is not offered.
A team can trace a wrong figure back to its source, inspect the checks applied and see each change or override, and Lantern helps clients with personal data requests. A path for an investor to challenge a generated figure, or compensation for a reporting error that began in AI extraction, is not offered.
Anthropic and OpenAI supply the language models, Amazon Web Services and Google Cloud the cloud infrastructure, Microsoft Azure hosting and iMerit data annotation, each named with its location in the subprocessor annex to the data processing addendum, beside the analytics and sales tools. Freyda, now part of Lantern, points to the same addendum. Which model handles each document type is not stated.
Connections cover eFront, Investran, Allvue, iLevel and Chronograph alongside Excel. Records from several administrators are mapped to a common format and delivered through Snowflake, APIs, secure file transfer and a Model Context Protocol interface, so validation and reporting run on existing records without a replacement ledger.
Lantern runs as a hosted service, with Microsoft Azure hosting in the United States and the EU, Amazon Web Services and Google Cloud as worldwide cloud infrastructure, and both language model providers processing worldwide. A committed region for a client's data and AI processing, or an on premises version, is not offered.
Firms start with a demonstration and scope which administrators, internal systems and document types to connect, and document extraction can also be bought as a managed service through Freyda. Prices per fund, document or user, and any minimum volume, are not listed.
General partners verify portfolio company and fund reporting, limited partners extract data from the notices and statements their managers send, and fund administrators standardize the data they receive and distribute. Case material spans buyout, healthcare and infrastructure private equity, growth equity and venture capital, and firms with several administrators on different accounting systems, all on the same verification layer.
Alternatives to Lantern
Other products that do the same kind of work as Lantern, and what each one does.
Freyda reads private markets documents for institutional investors and fund administrators. It retrieves files from general partner portals and can take in documents from cloud storage, email, file transfer or an API.
Juniper Square is a San Francisco based fund operations partner to more than 2,300 private markets general partners, combining software with an outsourced fund administration business, valued at $1.1 billion after a $130 million Series D led by Ribbit Capital.
Accelex automates the extraction of investment data from private markets fund documents for asset owners, allocators, institutional investors and the asset servicers acting for them.
Allvue Systems sells the operating software behind private equity firms, private credit and CLO managers, fund administrators and banks.
Carta is a San Francisco platform for private capital operations, originally a capitalization table system and now positioned as a vertical resource planning system for private markets, spanning fund administration, fund accounting, tax, compliance, investor reporting and a limited…
CEPRES provides private markets data and analytics. It began in 2001 as the Center of Private Equity Research, collecting fund, deal and portfolio company data directly from general partners, became independent in 2012 and has had a New York office since 2014.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | Data Protection Terms | Implementation | Source |
|---|---|---|---|---|
|
Quote from Lantern
|
Pricing follows the administrators, systems, documents and validation scope a firm connects, set after a demonstration. | A public data processing addendum, last updated March 2026, names every subprocessor, including Anthropic and OpenAI for language models. | Setup covers mapping administrators and systems, writing custom rules and configuring delivery feeds. A standard setup fee is not listed. | Vendor Published |
Lantern sells by quote after a demonstration, with managed document extraction through Freyda as an option.