Kepler vs AlphaSense

For investment firms that need numbers they can defend, built on their own data and definitions, Kepler is the stronger choice. Our engineers build your firm’s ontology, every document is read in full before anyone asks, code computes and cites every number to the exact page, line or cell, and agents keep working on schedules and new filings. AlphaSense is built to search a large licensed library of filings, transcripts, broker research and expert calls.

We build Kepler. Facts about AlphaSense come from its website, press releases and public reporting as of November 2026, linked inline.

The short answer

• Choose Kepler when the answer becomes a number in a model, an IC memo or an LP report; when it depends on your firm’s own documents and definitions; when you need every matching document; and when you want agents watching the book overnight.
• Choose AlphaSense when the job is mostly reading broker research, expert-call transcripts and what management teams and competitors have said.

Kepler vs AlphaSense at a glance

	Kepler	AlphaSense
Built around	Your firm: its ontology, documents and definitions	A licensed library of 500M+ business documents
Core job	Answers, models and memos the firm decides on	Searching and summarizing what’s been said
Where each number comes from	Pulled from source and computed by code over your ontology, cited to the line	Standardized financials, Canalyst models, and agents that write and run code over the library
What a citation opens	The exact page, line or cell; formula and inputs for calculations	The text snippet behind the answer
Same question twice	The same number, with the same sources	No published reproducibility guarantee
Your firm’s definitions	Built into your ontology, applied every time	Written into prompts, Skills and templates
Your firm’s documents	Every page read at ingest and resolved into the ontology	Searched through 20+ connectors with mirrored permissions
Finding everything	Complete lists across every document	Search results ranked across the library
When sources disagree	Both shown, with sources, and the conflict flagged	Reconciled inside agent runs
As filed vs as restated	Either, on any date	Not a stated feature
Work that runs on its own	Agents on schedules and events, acting as their owner	SuperAnalyst monitoring and refreshed outputs
Where it works	App, Excel, PowerPoint, Word, Claude, ChatGPT, Slack, email	AlphaSense app, Excel and PowerPoint assistants
Build on it	API, MCP, Python SDK and webhooks on the same engine and ontology	Agent API with search and MCP tools; utility APIs

What Kepler is

We build enterprise intelligence for investment firms: one living model of your firm, its data, definitions and relationships, that AI can reason over, prove and act on under your firm’s own rules.

Your firm’s ontology. Our forward-deployed engineers, who built this layer before at Palantir, build it from your own documents and systems. Every company, fund, vehicle, manager, deal and person resolves to one record; your house terms carry one definition; every relationship points to the sentence that established it. It updates itself as documents arrive and keeps history, so you can ask what was reported originally, what was restated, or what was true on a given date. See the ontology.

Every document, read in full. Before your first login, the public record is already there: 26M+ SEC filings across 20+ form types and 40M+ IR documents and earnings call transcripts covering 14,000+ companies in 120+ countries, plus market data and news. SharePoint, Box, Egnyte, email, Slack, your CRM and data rooms join through read-only connections. At ingest, we read every page with our own extraction models and store each fact beside its passage, value, scale and unit; in held-out tests those models read tagged facts at the right scale 99.4% of the time (Stop Paying to Forget).

A deterministic engine. No number in Kepler is written by a model. Financial statements come straight from each filing’s XBRL, every calculation runs as code over cited inputs with its formula visible, and each citation is tested against the passage it names before you see it.

Your playbooks, on the right models. The earnings update the way your PM wants it, comps on your own adjusted EBITDA and the IC memo in house style are saved once and run the same way every time, and each correction your team makes improves the firm’s models (how to get AI to use your firm’s definitions). We route each task to the best model for it: small models we train do the reading, frontier models handle the question and the write-up, and every better model that ships makes Kepler better.

Agents, everywhere, under your rules. Agents start on a schedule or a trigger and act with exactly their owner’s access. You can work with Kepler in its own app, in Office, inside Claude and ChatGPT, or in Slack and email, and your engineers can build on our API and MCP server. One access check governs every read, every query is logged with its lineage, and we hold SOC 2 Type II.

Not the right fit if: your analysts mostly read expert calls and broker notes and rarely turn that reading into numbers the firm acts on.

What AlphaSense is

AlphaSense is a market intelligence platform, and its center of gravity is content. It says it searches more than 500 million premium business documents, and the 2024 Tegus acquisition brought Canalyst models, BamSEC and expert transcripts in a library AlphaSense now puts at 300,000 interviews. Generative Search answers a question with a citation to the exact snippet. SuperAnalyst, the company’s agent, opened in early access in June 2026 and launched on October 6, 2026: it takes one objective, plans the work, reconciles evidence, runs code and produces memos, models and dashboards that refresh themselves. Its Context Graph turns library content into structured data points linked to source, and Skills package workflows such as an earnings preview or competitive mapping.

On the firm’s side of the ledger, AlphaSense added native Excel and PowerPoint assistants in July 2026, and Enterprise Intelligence connectors pull SharePoint, Box, Google Drive, Egnyte and other systems into the same search as the library. For how it compares with the Terminal’s AI, see AlphaSense vs Bloomberg ASKB.

What users say: “When I check whether an agent or routine followed the process I specified, I consistently see errors,” a G2 reviewer at a private equity firm wrote in October 2026 (AWS Marketplace, G2 reviews).

How each one answers the same question

Take a request a sector PM might send on a Monday: “Segment operating margin for our eight semiconductor names, last 12 quarters, calendarized to our quarter-ends, next to adjusted operating margin on our definition (GAAP operating income plus stock-based comp, acquired-intangible amortization and restructuring). Flag any segment recasts.”

In AlphaSense, an analyst can hand this to SuperAnalyst as one objective. It plans the steps, finds the segment disclosures and add-backs across filings, transcripts and Canalyst models, runs code to compute margins, and returns a table with each data point linked to its source snippet. The house definition and the calendar convention go into the prompt or a custom Skill. Before the table goes to the PM, the analyst should confirm which version of each segment table the inputs came from where a company recast its segments, check that year-to-date cash flow figures were turned into single quarters, open the source behind each add-back, and rerun the request to see whether the figures hold.

In Kepler, the request starts from what the firm already knows. The eight names are a coverage list in your ontology, and “adjusted operating margin” and your quarter-end convention already carry your definitions. The engine pulls segment revenue and segment operating income from the segment note of each 10-Q and 10-K, takes operating income, restructuring and acquired-intangible amortization from the statements and notes, and takes stock-based compensation from the cash flow statement, turning year-to-date figures into quarters. It maps each fiscal quarter to your calendar with the rule written into the formula, then computes every margin for all 12 quarters in code and fills the table. Every cell opens the line it came from; every margin opens a small sheet with the formula and its inputs. Where a company recast its segments, we show the as-filed and restated figures side by side and flag the change. Ask again next month and the history is identical, with the new quarter added. An agent can rerun the table the morning each new 10-Q lands and send the PM the changes, with quotes from the call.

The same pattern holds on the private side. “Every LPA in our fund documents with a key-person clause, and the named individuals” returns the complete list from Kepler, because every document was read at ingest and each clause is a record in your ontology.

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

Setup, security and cost

AlphaSense is enterprise SaaS with a private cloud option, ready once seats are provisioned. Bringing internal content into the search requires the Enterprise Intelligence offering and connector setup. AlphaSense says customer content never trains its models. Prices aren’t published: SpendHound reports average spend of $10,131 a year for smaller customers and $139,455 for enterprise customers, across 160 AlphaSense customers, and Vendr reports that contracts commonly auto-renew unless canceled 60 to 90 days ahead.

Kepler is live on public companies the day you sign in. In the first week our engineers connect your sources read-only, index every document and build your ontology, then keep working on site with your team as the use cases grow. Anything a second firm would need goes back into the platform, so each new firm starts further ahead. We hold SOC 2 Type II, comply with GDPR and CCPA, isolate each firm’s data at the identity layer, keep your data out of any model other firms use, and deploy in our cloud or yours. Details are on our security page.

To compare the two on your own work, put the same request to both and run the tests in our guide to checking AI numbers.

When to choose Kepler

• The number is going somewhere. Models, IC memos, LP letters, client decks: anything a person will be asked to defend.
• Your firm’s data matters. Fund documents, data rooms, CRM, portfolio data and internal models belong in the answer alongside public filings.
• Your definitions differ from the vendor’s. Adjusted EBITDA, net IRR, look-through exposure and peer sets the way your firm defines them, every time.
• You need the complete set. Every 10-K with a going-concern paragraph, every agreement with a change-of-control clause, every holding exposed to a supplier.
• Restatements and recasts matter. As filed or as restated, on any date, with conflicts flagged.
• Work should run without a click. Overnight agents on the book, guidance-change alerts with the exact quote, digests before the morning meeting.
• Your team lives in other tools. Excel, PowerPoint, Claude, ChatGPT, Slack, or your own apps through our API.
• Engineering wants to build. Firms with engineers build their edge on Kepler’s data, ontology and engine, and skip the data layer that took us 17 months and 109 releases to build.
• Compliance needs the trail. Every run recorded and attributable, one permission check on every read, walls between deals.

When AlphaSense may fit

• Expert calls are central to your research process.
• Broker research is something your analysts read daily in one place.
• Broad qualitative search across news, transcripts and commentary is the main job.

Expert-call transcripts and broker research are licensed content outside Kepler, so teams that read them daily keep a library for that. Our Tegus alternatives guide covers the expert-call options.

Next steps

For other comparisons, see Kepler vs Hebbia, Kepler vs Rogo, the full list of AlphaSense alternatives, our ranking of the best AI tools for investment research and our compare page.

Put the same question to both. Start with Kepler at app.kepler.ai, or book a demo on your own coverage list and fund documents.

Frequently asked questions

What is the difference between Kepler and AlphaSense?

Kepler is enterprise intelligence built around your firm: our engineers build your firm's ontology, every document is read in full before you ask, and code computes and cites every number to the exact page, line or cell. AlphaSense is a market intelligence platform built to search a licensed library of more than 500 million documents, including broker research and 300,000 expert interviews. For numbers the firm decides on, built on its own data and definitions, Kepler is the stronger choice.

Can Kepler replace AlphaSense?

For the work the firm decides on, yes. Kepler covers filings, transcripts, IR materials, market data, news and your own documents, and adds your firm's ontology, numbers computed in code and agents that run on schedules and new filings. Models, IC memos, LP reports, screens and monitoring all run on it, with every figure cited to its source and the same answer every time you ask.

Does Kepler work with my firm's own documents?

Yes. In the first week our engineers connect your SharePoint, OneDrive, Box, Egnyte, email, Slack, CRM, data rooms and research notes, read-only, and extraction models we train read every page. Each fund, holding, manager, deal and document resolves to one record in your firm's ontology, and your definitions of terms like adjusted EBITDA or net IRR are applied in every answer.

Which is better for numbers that go into a model or IC memo?

Kepler. Every figure is pulled from its source and computed by code, never written by the model, and every citation is checked against its source before the answer reaches you. Asking the same question twice returns the same number with the same sources, as filed or as restated, which is what an investment committee or an LP report needs.

How long does it take to move from AlphaSense to Kepler?

Public companies are live in Kepler the day you sign in. In the first week our engineers connect your sources read-only, index every document and build your firm's ontology. The longer clock is usually the AlphaSense contract: Vendr reports that contracts commonly auto-renew unless canceled 60 to 90 days ahead, so check the notice window before you plan a trial.