Brightwave
Brightwave is a New York AI research platform for investment professionals, founded by engineers from artificial intelligence and financial data infrastructure backgrounds. Its systems synthesise analysis across a corpus of hundreds of millions of documents including securities filings, earnings call transcripts, sell side research, breaking news, market data, expert network calls, investment memos and a customer's own internal knowledge base content, then distil the findings into written narratives structured the way an analyst would write them rather than returning a ranked list of passages. Users can also question primary sources directly.
Buyers span the buy side, from owner operated registered investment advisers to crossover hedge funds, with a customer base reported at over $120 billion of assets under management, and wealth management firms use it to prepare client facing market commentary. Funding totals $21 million across a seed round in June 2024 led by Decibel Partners with Point72 Ventures, and a $15 million Series A in October 2024 led by Decibel with OMERS Ventures.
Capability Axes
Capability grades
15 of 15 axes rated · 3 graded A or B
Strip the inference and nothing remains but a document corpus the customer could not use. There is no workflow layer, system of record or data asset underneath that would survive. The models are the entire product. The company describes itself as building and training purpose built systems that reason over bodies of information exceeding what a person can hold. The deliverable is generated analysis rather than retrieved documents, the assistant native shape at its clearest.
The output is analysis handed to an investment professional who decides. The vendor explicitly positions the product as a partner in thought and a force multiplier rather than a decision maker, a real scope statement rather than reassurance. Users can interrogate the primary sources behind an answer, which puts a check between generation and reliance. No approval gate, confidence threshold or escalation rule is published. The system is described as autonomously reasoning, with no statement of what it is prevented from doing.
No accuracy rate, validation approach or benchmark is published, and there is no evaluation method or drift policy. Trustworthiness, insight and precision are asserted as qualities of the output. Quality controls are referred to without being described. The ability to question primary sources is the nearest thing to a provenance property. It is offered as a feature for exploring the corpus, not as a guarantee that every generated assertion traces to a cited source.
No customer is named, and every figure is self reported. The published numbers describe the vendor's own business rather than a customer outcome: fourfold revenue growth over four months, and an aggregate assets under management figure for the customer base. The chief executive is quoted, but in the company's own funding announcements rather than as an independent party. An aggregate assets figure across unnamed clients is a scale claim, not evidence that the product produced a result for any of them.
Storage isolation is stated and training isolation is not. Customer data is stated to be stored in isolated sections of the infrastructure, segregated from other customers. That is a tenancy statement about storage, not about inference or training. The company separately and affirmatively describes itself as building and training proprietary systems without saying what they are trained on. The two are different commitments, and a reader who conflates them would credit a protection that has not been offered.
A privacy policy and terms of use are published, and nothing beyond them. There are no processing terms, no sub processor list and no retention schedule. Nothing addresses the handling of client identifying material that may sit inside the investment memos and internal documents customers upload.
No certification, audit type, report or trust portal is published. In their place is an adjective: security described as exchange grade and built by financial services veterans. It is the same kind of claim as institutional grade, moved from the buyer question to the security question. Exchange grade is a claim about quality that a reader cannot check. A dedicated safety page exists and carries no credential, so the absence is documented, not assumed.
No licence, registration, supervised status or regulator relationship, and none would be expected: the product generates research for regulated firms rather than performing a licensed activity. Worth noting for a buyer that output used to produce client facing market commentary may fall inside the adviser's own marketing and communications rules, and the vendor publishes nothing addressing that.
Nothing published on evaluation for bias, on how the system weights or selects among conflicting sources, or on governance of the generation process. Relevant in a different way than in credit: the risk here is not protected class discrimination but systematic slant in which sources are surfaced and which are omitted when synthesising a thesis from a corpus the user cannot inspect in full.
No liability position, indemnity, accuracy warranty or remediation route published for generated analysis. The exposure is concrete rather than theoretical: the output is intended to inform investment theses and client facing commentary at firms managing money, and nothing published addresses what happens when a synthesised fact pattern is wrong.
The systems are described as purpose built and proprietary. Not one model, provider or model family is named, and neither is a version or hosting arrangement. Vendors that license capability tend to name their suppliers, while vendors that claim to build their own usually name none. No model provider publishes this vendor as a customer either.
The platform ingests a customer's internal documents, investment memos and proprietary knowledge base content, which is genuine document level integration, but no core system connectors are named: no order management, portfolio accounting, customer relationship or research management system is listed, and no application interface or protocol server is described. It sits alongside the systems its users work in rather than inside them.
Nothing published on hosting model, region, residency options or whether any single tenant or private deployment exists. Consequential for a product ingesting a fund's internal investment memos and knowledge base content, where a European or Asian customer would need to know where that material comes to rest.
No price, tier, seat rate or unit of charge published, and no pricing page. Independent comparison coverage of this category describes the vendor's pricing as enterprise quoted and unpublished, which corroborates absence rather than relieving it.
Buy side and wealth management, with real spread inside that: owner operated registered investment advisers at one end and crossover hedge funds at the other, plus asset managers using it for public equity work, private markets diligence and thesis development. Aggregate customer assets under management are published at over $120 billion. The institution types are confined to the buy side and adjacent wealth firms, with no banking, insurance or corporate segment addressed, and no geographic breadth published.
Alternatives to Brightwave
Vendors in the same categories as Brightwave whose profiles look most alike, each with a line on what the product does. Similar is not the same: check that each one does the job you need. No vendor pays for placement.
Farseer sells an AI monitoring and decision support platform to banks, listed companies and government bodies, mainly in Hong Kong and mainland China.
Hudson Labs, founded in 2019 in Toronto and formerly called Bedrock AI, builds finance specific language models for institutional equity research.
Linvest21 builds AlphaCopilot, an investment platform run by specialized agents including equity analysts, portfolio managers and virtual chief investment officers, sold to investment…
KelAI runs the whole systematic investment research loop autonomously for hedge funds and institutional investors, taking idea generation through data analysis, backtesting, signal…
TeamSec runs a cloud securitization as a service platform for banks, investment banks, brokerage houses and asset originators.
Artian builds multi agent AI for financial institutions.
The index publishes no overall score or ranking. See what we assess and the methodology.
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.
No pricing data has been verified for this vendor. Pricing information will be published here once confirmed through vendor disclosure or third-party estimation.