Kepler vs Chronograph

For private equity firms, private credit funds and allocators who need answers that cut across every document and system the firm holds, with each number computed in code and cited to its source, Kepler is the stronger choice. Chronograph is a dedicated portfolio monitoring and reporting system for GPs and LPs, and it has added AI search, an MCP server and assistant plugins on top of the data clients keep in it. Kepler builds one living model of your firm from all of its documents and systems plus the public record, reads every page before anyone asks, and runs agents that act the moment a package or filing lands.

We build Kepler. Facts about Chronograph come from its website and announcements as of November 2026, linked inline.

The short version

• Chronograph collects portfolio company and fund data, runs valuations, benchmarking and LP reporting, and lets clients query that data through ChronoAI, Semantic Search and connectors for Claude, ChatGPT and Perplexity.
• Kepler is enterprise intelligence for investment firms. Our engineers build your firm’s ontology from every source you connect, extraction models read every page at ingest, a deterministic engine computes each figure with its formula and inputs, and agents keep watch on schedules and events.
• The choice turns on scope. A system of record answers well about the data collected into it. A model of the whole firm answers across the portfolio, the deal files, the inbox, the CRM and public markets at once.

Kepler vs Chronograph at a glance

	Kepler	Chronograph
Built around	One living model of your firm: funds, companies, deals, managers, people, documents	A system of record for portfolio monitoring, valuations and reporting
What it reads	Every document the firm connects, read in full at ingest	Portfolio and fund reporting plus the documents clients keep in Chronograph (board materials, LPAs, credit agreements)
Public markets	26M+ SEC filings, 40M+ IR documents and transcripts, market data, news	Not its focus
Collecting packages	Read from email, portals, SharePoint or data rooms in any layout	Collection workflows; PDF and Excel intake for credit, per Chronograph
Where each number comes from	Pulled from the source and computed by code	Collected, validated and versioned in what Chronograph calls a deterministic system of record; plugin analysis runs in Claude
What a citation opens	The exact page, line or cell, checked before you see it	Document and page for MCP answers, per Chronograph
Your firm’s definitions	Written into your ontology and applied in every answer	Configured KPIs, templates and reports
As filed vs restated	Either, on any date, public or private	Object-level versioning and point-in-time history, per Chronograph
Complete lists	Every matching document across the firm	Semantic Search across documents held in Chronograph
Work that runs on its own	Agents on schedules and events, acting as their owner	Collection workflows; AI skills run in Claude
Where it works	App, Excel, PowerPoint, Word, Claude, ChatGPT, Slack, email	Chronograph app, Office add-ins, Claude, ChatGPT, Perplexity
Build on it	API, MCP server, Python SDK, webhooks	GraphQL and REST APIs, MCP, Snowflake data share
Beyond monitoring	Research, diligence, screens, IC memos and LP letters on the same model	Valuations, benchmarking and LP reporting

What Chronograph is

Chronograph was founded in 2016 and sells two products. Chronograph GP automates portfolio company data collection, analytics, valuation, reporting and data warehousing across buyout, growth, venture, infrastructure, real assets and private credit. Chronograph LP gives fund-of-funds, pensions, insurers, OCIOs, endowments and family offices a view of their commitments down to the underlying companies. Its private credit product collects borrower financials and compliance certificates from PDFs and Excel without rigid templates, checks adjusted EBITDA add-backs against negotiated caps, and feeds Excel through the xConnect add-in and a GraphQL API. Snowbank ports client data into Snowflake.

The AI work sits on top of that data. ChronoAI lets LP clients work with the documents they keep in Chronograph, from quarterly reports and capital account statements to LPAs and AGM materials. Semantic Search reached LP clients in March 2026 and GP clients on September 30, matching meaning across board materials, operating reviews and credit agreements. Chronograph’s MCP server works with Claude, Claude Code, ChatGPT, Codex and Perplexity and cites the document and page behind an answer. On October 5 it released plugins for Claude: the GP plugin assembles quarterly review packs, flags companies off budget and attributes fund TVPI by company, and the LP plugin forecasts cash flows with the Takahashi-Alexander model, builds commitment pacing plans and scans look-through exposure. Chronograph’s product manager for ChronoAI and MCP put it this way in September 2026: “We have spent a decade making Chronograph a deterministic system of record, and anything we build on top of it inherits that.”

What Kepler is

A frontier model on its own brings two problems to a portfolio review. It writes numbers rather than pulling them, so a figure can be wrong with nothing on the page to say so. And it knows nothing about your firm: which entity in the board deck is which company in Fund II, how your team adjusts EBITDA, what the deal model assumed. We built Kepler to close both.

One model of the firm. Our forward-deployed engineers, who built this layer before at Palantir, build your firm’s ontology from everything you connect: the portfolio files, the deal folders in SharePoint, the inbox where packages arrive, the CRM, the data rooms, research notes and models. Every company, fund, vehicle, lender, sponsor, manager and person resolves to one record, every relationship points to the sentence that supports it, and your house definitions carry one meaning. The model keeps itself current and knows time, so you can ask what was reported, what was restated, or what was true at any quarter-end.

Every page read, public and private. Extraction models we train read every page at ingest, scans, tables and footnotes included, and keep each figure with its value, scale and unit beside its passage. The public record arrives already loaded: 26M+ SEC filings across 20+ form types, 40M+ investor relations documents and earnings call transcripts, market data and news for 14,000+ companies in 120+ countries. A portfolio company’s listed peers, its listed lenders and its customers’ filings sit in the same model as its monthly package.

Numbers from code. The model never writes a figure. Code pulls each input and computes every KPI, margin, leverage ratio and exposure, and the result opens to its formula and cited inputs. Each citation is checked against its source while the answer is written; one that doesn’t match is removed before you see it. When a board deck and a monthly package disagree, Kepler shows both and flags the conflict.

Agents, playbooks and every surface. Agents run on schedules and events (a package arriving, a covenant test date, a filing from a listed peer, news on a sponsor), acting exactly as the person who set them up and never steered by what they read. The quarterly portfolio review, the LP letter and the IC update are saved once in your format and applied every time, and each correction your team makes improves the firm’s models. Results arrive in the Kepler app, Excel, PowerPoint, Word, Claude, ChatGPT, Slack or email, and your engineers can build on the same engine through our API and MCP server.

Not the right fit if: your firm only wants a portal for collecting KPIs and publishing a fixed quarterly report, with no research, diligence or analysis on the same data.

“I can see exactly where every number and conclusion comes from, which means I can defend the work.” (Vice President, Global Investment Bank)

The same quarter, two ways

Take a request a deal partner might send in the week after quarter-end: “Which portfolio companies are more than 10% below budget on adjusted EBITDA, on our definition? What did each management team say about it, and how do their margins compare with listed peers?”

In Chronograph, the GP plugin in Claude can compare actuals against budget across the portfolio and rank the companies that are off plan, provided the financials and budgets were collected into Chronograph. Semantic Search can surface the board-deck commentary on why. Listed peers come from another source. In a trial, check whose adjusted EBITDA the comparison uses (the company’s reported figure or your fund’s definition with its own add-backs), what happens where the board deck and the reporting package disagree, and what a click on each figure opens.

In Kepler, the packages were read when they arrived. Each company is one record linked to its fund, its deal model in SharePoint and its lenders. Your adjusted EBITDA definition, with its list of permitted add-backs, lives in the ontology, so code computes the figure for every company from cited lines in the monthly financials and compares it with the budget in the deal model. Management’s explanation comes back as quotes from the board deck, each linked to its slide. Listed peers resolve from the public ontology, with each peer’s own adjusted EBITDA taken from the reconciliation in its earnings release and summed over four quarters in code. Where a company’s board deck shows a different EBITDA from its package, both appear with the conflict flagged. An agent reruns the screen as each new package lands and sends the deal team what changed, and the quarterly review drafts in your template.

On the LP side the pattern holds: “our look-through exposure to this company across every fund and co-investment” returns one number broken down by vehicle, every input cited to a capital account statement or schedule of investments. More on that work is in how LPs use AI on fund documents.

Setup, security and cost

Chronograph is a cloud platform sold through demos, and its website doesn’t list prices. Its MCP connectors run on the client’s own subscription with the AI provider, with no separate data contract, per Chronograph.

Kepler has the public record live the day you sign in. In the first week our engineers connect your sources read-only, read every document and build your ontology, then keep working on site as the use cases grow. We hold SOC 2 Type II, comply with GDPR and CCPA, isolate each firm’s data at the identity layer, check your own permissions on every read by a person or an agent, never let your data train models other firms use, and deploy in our cloud or yours (security).

When to choose Kepler

• Your questions cross systems. The answer needs the portfolio data, the deal model, the board deck, the credit agreement and the inbox at once.
• Packages arrive in every format. You want each one read where it lands, scans included, without asking companies to fill in a template.
• Your definitions matter. Adjusted EBITDA, net debt to EBITDA, net IRR and exposure the way your firm calculates them, in every answer.
• Numbers go to an IC or to LPs. Each figure needs its formula and source line, and the same answer next quarter.
• Public markets are part of the picture. Listed peers, listed lenders, customers’ filings and market data belong next to private reporting.
• Monitoring should run without a click. Agents that read each package on arrival and flag what moved, with the exact quote.
• One model should serve the whole firm. Diligence, monitoring, LP reporting and research, so the terms you diligenced are the terms you monitor.
• Engineers want to build. Your own agents and apps on Kepler’s data, ontology and engine through our API and SDK.

When Chronograph may fit

• A firm that wants a dedicated portfolio monitoring and valuation system of record, with standard LP reporting and benchmarking.
• An LP team whose main need is collecting GP data across many funds into one database for pacing and cash-flow forecasting.

Next steps

For the wider field, see our list of the best AI portfolio monitoring tools and the best AI tools for LP and fund-of-funds due diligence. For the reporting side, read how to automate LP reporting with AI, and for why entity resolution matters so much here, what an ontology is for an investment firm.

Try Kepler at app.kepler.ai, or book a demo and we’ll run last quarter’s packages through a model of your firm.

Frequently asked questions

What is the difference between Kepler and Chronograph?

Kepler builds one living model of your whole firm from every system and document it holds, plus the public record, reads every page in full, computes every number in code with a citation to the line, and runs agents on new packages and filings. Chronograph is a portfolio monitoring, valuation and reporting system for GPs and LPs, with AI search and assistant plugins over the data and documents held in it.

Can Kepler do portfolio monitoring?

Yes. Kepler reads each reporting package, board deck, compliance certificate and capital account statement as it arrives, in whatever layout the company or GP sends, and files every figure against the right company, fund and period. KPIs compute on your firm's own definitions, agents flag what moved, and the quarterly review and LP letter draft in your house format with every number cited.

Does Kepler work for LPs as well as GPs?

Yes. For allocators, Kepler reads every GP's quarterly reports, capital account statements, notices, LPAs and side letters, resolves each fund and underlying company to one record, and computes look-through exposure as one number broken down by vehicle. Capital calls and distributions reconcile against each statement, and key-person or MFN questions return the complete list of agreements.

Chronograph has Claude plugins. Does Kepler work inside Claude too?

Yes. Kepler runs inside Claude and ChatGPT, as well as Excel, PowerPoint, Word, Slack and email, and engineers reach the same ontology and engine through the Kepler API, MCP server, Python SDK and webhooks. Wherever the question is asked, each figure comes back computed in code and cited to its page, line or cell.

How long does Kepler take to set up?

Public company data is live the day you sign in. In the first week our forward-deployed engineers connect your portals, inbox, SharePoint, CRM and data rooms read-only, read every document in full and build your firm's ontology, so the first quarter's packages already land in a model of your firm.