Brightwave Alternatives for Investment Research

For research teams moving off Brightwave, the best alternative is Kepler: a managed platform built around your firm, where our engineers build your firm’s ontology, every document is read in full before you ask, and every number is computed in code and cited to the exact page, line or cell. Brightwave’s homepage now reads “An agent infrastructure company”, and it released its research engine, rebuilt in Rust, as the open-source Tidebreak desktop app, so teams that want a hosted research platform are choosing again.

We build Kepler, so it’s first and labeled. Everything we say about Brightwave and the other tools is drawn from their own sites, blogs and press coverage, current to November 2026 and linked inline, and prices appear only when published by the vendor or reported in the press.

What happened to Brightwave

Brightwave was founded in 2023 to build an AI research assistant for investment professionals. On August 13, 2025 it launched Research Agents, a multi-agent system that drafted reports and evidence-linked tables, with an LLM judge reviewing each output before delivery. Through the first half of 2026 its blog logged Office add-ins, a Chrome extension, an Outlook add-in and desktop agents.

Today the company describes itself as building “the compliance-ready foundation for safely connecting AI agents to the systems where work happens.” The research engine lives on as Tidebreak, which Brightwave describes as “proven on three years of private equity diligence and rebuilt in Rust.” Tidebreak is Apache-2.0 and local-first. It tagged version 1.0.0 in September 2026 (its README still says pre-1.0), development now runs at github.com/naingthet/tidebreak, and new builds ship for macOS while Windows and Linux builds are paused. It turns folders of documents into versioned spreadsheets, decks, reports and memos, runs coding agents in isolated Git worktrees, and works with the model you bring: a ChatGPT plan, an API key from Anthropic, OpenAI, Google or others, or a local model through Ollama.

That leaves former users with a choice. Technical teams can run Tidebreak themselves and supply the model, the data, the connectors, the compliance review and the support. Teams that want a managed research platform with financial data and firm-wide controls are looking at the tools below.

What to look for in a replacement

• Where the numbers come from. Pulled from a filing and computed by code, or written by the model.
• What a citation opens. The exact page, line or cell, or a document in general.
• Your firm’s documents and definitions. Data rooms, models, notes and CRM, and your own way of calculating adjusted EBITDA or net IRR.
• Completeness. Whether a question across a data room returns every matching document. Our guide to reviewing a CIM with AI shows how to test it.
• Work that runs without a prompt. Agents on filings, schedules and events (see the best AI agents for finance).
• Who runs it. A vendor with engineers and an audited security program, or your own team.

Brightwave alternatives at a glance

Tool	Best for	Financial data included	Your own documents	How it shows sources
Kepler	Research built around your firm, with each figure computed and cited	26M+ SEC filings, 40M+ transcripts and IR documents, market data, news	Your firm’s ontology across SharePoint, Box, email, CRM, data rooms	Exact page, line or cell, checked before you see it
AlphaSense	Searching broker research and expert calls	500M+ documents, 300,000 expert interviews	Connectors to SharePoint, Box, Google Drive, Egnyte	Text snippets
Hebbia	One set of questions across many documents	Filings and transcripts; licensed feeds you connect	SharePoint, Box, Egnyte, data rooms	Sentence-level, per Hebbia
Rogo	Banker deliverables in Excel, PowerPoint and Word	Partner data such as PitchBook, FactSet, Daloopa	SharePoint, DealCloud, Datasite, Intralinks; Rogo Deal Room	Not published in detail
Claude or ChatGPT for Financial Services	A general assistant on financial data	Connected (Claude) or included (ChatGPT)	Connectors to file stores and warehouses	Links to the underlying data or passage
Perplexity	Quick cited lookups	Web and finance pages	Connected files	Linked web sources
Bigdata.com	Licensed content inside your own agents	News, transcripts, broker research, filings	Uploads	Ranked excerpts, each with source, date and URL

The best Brightwave alternatives

1. Kepler (our pick)

We build Kepler.

Best for: private equity, hedge fund and asset management teams that used Brightwave for diligence and research reports, and now want a managed platform built around their own firm, where every number is computed from source and cited.

Brightwave’s appeal was finished research with the evidence attached. A general model writes its figures as text, attaches a citation afterward, and reads only the passages its search pulls back, so a question across a 2,000-document data room returns the strongest hits and leaves the analyst wondering what was missed. It also doesn’t know your fund’s adjusted EBITDA, your deal code names or your IC memo template.

We built Kepler around the firm. Our forward-deployed engineers build your firm’s ontology, a living model in which every fund, company, deal, manager and document resolves to one record, each house term carries one definition, and every relationship points to the sentence it came from. Before a question is asked, models we train for extraction have read every page, scanned exhibits and tables included, across public data that’s ready on day one (26M+ SEC filings, 40M+ transcripts and IR documents, market data and news) and the data rooms, models, notes and email your firm connects.

Code computes the numbers. Statements build from XBRL as filed or as restated, every calculation runs over cited inputs with its formula visible, each citation is matched to its source as the answer is drafted, and where two documents disagree you see both, with the conflict flagged. Your playbooks are saved once: the IC memo in house style (how to write an IC memo with AI), the comps on your adjusted EBITDA, the quarterly portfolio review. Agents run them on schedules and events, such as a new document in the data room or a portfolio company’s 10-Q, and every run is recorded.

There’s nothing to host. Your team works with Kepler wherever it already works: the Kepler app, the Office suite, Claude, ChatGPT and Slack, plus our API and MCP server for your own engineers. We hold SOC 2 Type II, and we deploy in our cloud or yours. Our engineers built this layer before at Palantir, and in the first week they connect your sources and build your ontology.

Why teams moving off Brightwave pick Kepler

• Complete answers across a data room. Every document is read at ingest, so lists are whole.
• Numbers you can defend. Computed by code, identical every time you ask.
• Line-level sources. Every figure opens on its page, line or cell, and every derived figure lists the inputs behind it.
• Your firm’s ontology. Funds, holdings, deals and house definitions in one model of the firm.
• Playbooks and agents. Memos and reviews in your format, run on schedules and filings.
• Managed and secure. Engineers on site, your permissions mirrored, nothing read taken as an instruction.

Questions former Brightwave users ask Kepler

• “Every supplier agreement in the data room that lets the counterparty terminate on a change of control, with the clause quoted.”
• “Every related-party transaction the target’s parent disclosed over five years, each one cited to its filing.”
• “Draft the IC memo for this deal in our template, with every figure linked to its source.”

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

Not the right fit if: you want to self-host an open-source research app and wire up your own model, data and connectors.

Try it at app.kepler.ai or book a demo.

2. AlphaSense

Best for: teams whose research is mostly reading what brokers, experts and management have said.

Former Brightwave users who mostly wanted finished reports will find the nearest thing at AlphaSense in SuperAnalyst, an agent that AlphaSense opened to early-access users in June 2026 and launched on October 6, 2026, writing investment memos tied to source evidence. The draw is the library underneath: more than 500 million premium documents by AlphaSense’s count, broker research, and the Tegus expert-call archive (bought in 2024), now about 300,000 interviews. Your own files come in through connectors to SharePoint, Box, Google Drive and Egnyte. Prices are quoted privately, and SpendHound puts average AlphaSense spend at $10,131 a year for smaller customers and $139,455 for enterprise ones.

Compared with Brightwave: a far larger licensed library, built around search, with an agent that drafts from it. More in AlphaSense alternatives.

What users say: “for the generative AI, for a period, I struggled with potential hallucinations,” a G2 reviewer wrote in October 2026 (AWS Marketplace, G2 reviews).

3. Hebbia

Best for: diligence teams that want to put the same questions to every document in a data room.

Hebbia is the nearest hosted match for the document-heavy diligence Brightwave was known for. In its Matrix grid, every document gets a row, every question a column, and an agent answers each cell; Matrix 2.0, released in August 2026, turns a completed grid into a memo, deck or model, with a sign-off checkpoint before it’s final. Max, its assistant, reached a first group of users in July 2026 and gained memory across conversations in September, and Headless Hebbia opened the platform through an API and MCP server on October 1. Deal documents connect through Intralinks, SharePoint and Box, and deal records through Salesforce and DealCloud. Sacra estimates its Professional tier at $10,000 per seat a year.

Compared with Brightwave: hosted and supported, run question by question over the documents you select, with the model writing each answer. See Hebbia alternatives.

4. Rogo

Best for: investment banks and deal teams that want data-room review and finished Excel, PowerPoint and Word files from one vendor.

Rogo is organized around Felix, an agent that takes a request, by email if you like, and returns banker work product, alongside what Rogo describes as hundreds of purpose-built financial agents. For the data-room work many Brightwave users relied on, Rogo launched Rogo Deal Room in August 2026 and bought Rivanna to strengthen it, then acquired Arvo, a meeting assistant, in September. Its integrations page lists 51 sources, from PitchBook, Preqin, FactSet and Daloopa to SharePoint, DealCloud, Datasite and SS&C Intralinks.

Compared with Brightwave: stronger on formatted banker deliverables, now with a data-room product, and aimed at the sell side. See Rogo alternatives.

What users say: “can anyone make the case for Rogo being more than just a harness? Took a demo. It looked like basically a worse version of systems I’ve built myself,” a private equity VP wrote on Wall Street Oasis in 2026.

5. Claude for Financial Services or ChatGPT for Financial Services

Best for: firms that want a general frontier assistant with financial data attached.

Tidebreak lets you plug in a ChatGPT plan or an API key; these are the hosted, finance-specific versions of the same labs’ models. Claude for Financial Services reaches the data licenses a firm already pays for, such as FactSet, S&P Capital IQ, LSEG, PitchBook and Moody’s, and its add-ins for Excel, PowerPoint and Word are generally available. ChatGPT for Financial Services, which OpenAI opened to eligible institutions on September 10, 2026, bundles data into the plan instead: SEC filings, Quartr transcripts, Daloopa, Fiscal.ai, PitchBook Essentials, Crunchbase and LSEG News. Both point figures back to their sources, and neither trains on business data by default. Claude Opus 5.5 tops Vals AI’s Finance Agent benchmark at 67.85% (October 29, 2026 update).

Compared with Brightwave: managed, with data and enterprise controls, where Tidebreak leaves both to you. The model still writes the numbers. Kepler runs inside both; see ChatGPT vs Claude for financial analysis.

6. Perplexity

Best for: quick, cited orientation on a company, a market or a news event.

Before a deal team opens a data room, Perplexity Enterprise is a fast way to read up on the target and its market: it searches the web and any files you connect, and links each source in the answer. Its finance offering adds EDGAR filings and Quartr transcripts, and Perplexity Computer coordinates 20 models on longer projects.

Compared with Brightwave: faster for a first look, lighter for diligence. Figures headed for a model should still come from the filing.

7. Bigdata.com

Best for: engineering-led firms that run their own agents, Tidebreak included, and need licensed content with citations to feed them.

A team that takes the Tidebreak route still has to source its data, and Bigdata.com is one place to get it. RavenPack’s content service answers with ranked excerpts, each carrying its source, date and URL, drawn from earnings transcripts, more than 200 premium news sources, research from over 100 broker providers, filings covering 90,000+ companies and Economist Intelligence Unit analysis. Engineers call it through an API, and analysts reach it from connectors in Claude, ChatGPT and Microsoft Copilot.

Compared with Brightwave: content without the engine or the finished deliverables, for teams assembling their own stack.

Kepler vs Brightwave today

	Brightwave (Tidebreak)	Kepler
What it is	Open-source, local-first desktop app; the company now focuses on agent infrastructure	Enterprise intelligence platform for investment firms, deployed by our engineers
Models	You bring a ChatGPT plan, an API key or a local model	We route each task to the best model, with small models we train doing the reading
Financial data	Your files, web search and any MCP servers you connect	26M+ SEC filings, 40M+ transcripts and IR documents, market data and news, plus your firm’s systems
Your firm’s definitions	Skills and prompts your team writes	Your firm’s ontology, one definition per term
Where numbers come from	The model you choose writes them	Pulled from source and computed by code
Citations	Links to the supporting page, line, sheet or cell	Exact page, line or cell, checked before you see it, with formulas for calculated figures
Agents	Background agents in sandboxes you run	Agents on schedules and events, acting as their owner, every run recorded
Deployment and support	Self-run, Apache-2.0, macOS builds (Windows and Linux paused)	Forward-deployed engineers, SOC 2 Type II, our cloud or yours

How to choose

• You want to run your own engine and have engineers to support it: Tidebreak.
• Your research is mostly broker reports and expert calls: AlphaSense.
• You ask the same questions across big document sets: Hebbia.
• You produce banker deliverables: Rogo.
• You want a general assistant with financial data: Claude or ChatGPT for Financial Services.
• You need AI built around your firm, with complete answers and every number computed and cited: Kepler, which also runs inside Claude and ChatGPT.

For the wider field, see our best AI tools for investment research and our guide to how hedge funds use AI for research.

Try Kepler at app.kepler.ai, or book a demo and bring a data room you know well.

Frequently asked questions

What is the best Brightwave alternative for investment research?

Kepler, for firms that want a managed research platform built around their own documents and definitions. Our engineers build your firm's ontology and connect your data rooms, notes and models, Kepler reads every document in full before you ask, and code computes each figure and cites it to the exact page, line or cell. Agents run your playbooks, such as the IC memo in house style, on schedules and new filings.

What happened to Brightwave?

Brightwave, founded in 2023 as an AI research assistant for investment professionals, has shifted its focus to agent infrastructure. Its homepage now describes an agent infrastructure company building a compliance-ready foundation for connecting AI agents to business systems. It released its research engine, rebuilt in Rust, as Tidebreak, an open-source, local-first desktop app under the Apache-2.0 license.

Is Brightwave open source now?

Its research engine is. Brightwave released Tidebreak under the Apache-2.0 license, describing it as the research engine proven on three years of private equity diligence and rebuilt in Rust. Tidebreak runs locally, with new builds for macOS (Windows and Linux builds are paused), and you bring your own model through a ChatGPT plan, an API key or a local model. It tagged version 1.0.0 in September 2026.

Can I still use Tidebreak for financial research?

Yes, if your team is comfortable running its own setup. Tidebreak turns folders of documents into spreadsheets, decks and memos, with source links to the page, line, sheet or cell. Your team supplies the model, the data, the connectors, compliance review and support. Firms that want a hosted platform with financial data, firm-wide permissions and engineers on site usually choose a managed product instead.

How does Kepler compare to Brightwave?

Brightwave's engine now ships as a self-run desktop app where the model you choose writes the answer. Kepler is managed for you: in the first week our engineers connect your sources and build your firm's ontology, extraction models read every document at ingest, code computes and cites each figure, and agents run your playbooks on schedules and filings. It works in its own app, Excel, PowerPoint, Claude, ChatGPT, Slack and an API.