Hebbia
Agentic AI document-intelligence platform for financial services — the Matrix product orchestrates multi-step research across long-document corpora (earnings transcripts, 10-Ks, deal documents, private-equity diligence binders) for 50+ Fortune 500 customers including a third of the top-50 US asset managers, plus marquee deployments at Centerview Partners, Charlesbank, Fenwick & West and Ropes & Gray. Founded 2020 in New York by George Sivulka (Stanford electrical-engineering PhD candidate; BS Mathematics, MS Applied Physics); backed by Andreessen Horowitz, Index Ventures and Google Ventures at $700M post-money on the July 2024 Series B.
The Business
Hebbia builds an agentic AI document-intelligence platform purpose-built for financial services — the Matrix product orchestrates multi-step research workflows across long-document corpora (10-Ks, earnings transcripts, deal documents, private-equity diligence binders) that buy-side and sell-side analysts work from, with FlashDocs as the synthesis surface that converts the multi-step research into the formatted output the analyst ships. The company was founded in 2020 in New York by George Sivulka, a Stanford electrical-engineering PhD dropout (BS Mathematics, MS Applied Physics) whose AI-research network anchors an unusually deep senior bench (approximately 30% of headcount ex-OpenAI / ex-Anthropic on Sivulka’s own framing). The product is grounded in customer-private long-document corpora rather than third-party-vendor corpus assets — the structural differentiator from AlphaSense (broker research and expert calls) and BloombergGPT (Bloomberg Terminal market data and news). By the July 2024 Series B close Hebbia had raised approximately $161M cumulatively at a $700M post-money valuation with Andreessen Horowitz leading and Index Ventures and Google Ventures co-investing; TechCrunch’s reporting at the round disclosed $13M ARR alongside a profitability claim — unusual at AI-application scale and a contributing input to the D4d capital-position sub-rubric score of 8. ARR is reported to have grown approximately 15x over the subsequent 18 months per Sacra and secondary coverage; we mark the implied range at $80-150M for mid-2026 but decline-to-publish a precise current figure pending primary disclosure.
Customers and Distribution
Hebbia serves 50+ Fortune 500 customers with a third of the top-50 US asset managers in the deployed-customer cohort, plus named flagship deployments at Centerview Partners (boutique investment-banking advisory), Charlesbank (middle-market private-equity), Fenwick & West (technology-and-life-sciences law firm) and Ropes & Gray (the emergent legal-vertical deployment that extends the platform reach beyond pure-play financial services). The customer roster skews buy-side hedge funds and asset managers plus corporate-development teams plus advisory-side investment-banking boutiques, with the asset-management concentration the standout GTM print that the P3c sub-rubric scores 8 on. The 1B-pages-processed milestone reported in 2026 is the data-gravity print on the financial-services deployment footprint — the structural-substitutability friction that the D1c sub-rubric scores 7 on, reflecting that Matrix query patterns plus custom workflows plus FlashDocs synthesis become sticky to Hebbia even though the underlying document corpus is portable in principle. Distribution rides direct enterprise GTM into investment-research, advisory and asset-management procurement; the UK and Europe team tripled by March 2026 per TechFundingNews coverage on the EMEA expansion. Revenue is not formally disclosed at the mid-2026 evidence-pass window; ARR implied by the 15x growth over 18 months from the $13M Series B base suggests $80-150M, but we decline-to-publish a precise figure pending primary disclosure.
Model Strategy
Hebbia’s defining technical asset is the Matrix multi-step orchestration layer plus the FlashDocs synthesis product surface — the agentic-workflow capability that converts a long-document corpus and a research prompt into a structured, formatted output through multi-step retrieval, reasoning and synthesis. Foundation-model strategy is multi-model orchestration across OpenAI and Anthropic routed through a named Microsoft Azure partnership disclosed in 2024-25 product and partnership coverage, with model selection inside Matrix driven by latency, cost and capability per orchestration step. The D4a supplier-diversity sub-rubric scores 5 on this evidence — the lowest score in the defensibility composite. Multi-model routing inside Matrix addresses the foundation-model-supplier concentration question in principle (capability can be re-routed inside the orchestrator as alternative providers ship competitive frontier capability), but the asymmetric overlay is that OpenAI itself competes at the workflow layer with ChatGPT Enterprise and the Morgan Stanley GPT bulge-bracket reference — the same indirect-competitor risk that Harvey carries in legal AI and Rogo carries in investment banking. The architectural cut is deliberate: the Azure partnership provides the enterprise-grade security perimeter that buy-side and sell-side procurement requires, and the multi-step Matrix orchestration over customer-private documents is the moat that makes the foundation-model layer commodity-input rather than capability-bottleneck for Hebbia’s product surface.
At A Glance
The Numbers
Annualised revenue
Headcount (FTE)
Funding to date
Leadership Team
Hebbia is founder-led with George Sivulka as the principal public voice and CEO. The senior bench is unusually concentrated in ex-OpenAI and ex-Anthropic research engineering experience (~30% of headcount on Sivulka’s framing), which is the structural source of the multi-step agentic-orchestration differentiation that Matrix delivers. No CRO, CFO or CTO has been publicly named as a separate appointment distinct from the founder-and-engineering bench at time of writing; founder concentration is meaningful at the $80-150M ARR implied stage and the D4e key-person dependency sub-rubric scores 5 on this evidence. Headcount sits in the ~100-200 range with the UK and Europe team tripled by March 2026; we decline-to-publish a precise headcount pending primary disclosure.
IM Framework Scoring
IM’s structured assessment of Hebbia’s competitive position. The summary below is the headline; expand “Show the full analyst-grade analysis” near the bottom for the per-dimension reasoning and evidence. Methodology →
Funding History
| Date | Round | Raised | Post-money | Lead investor(s) |
|---|---|---|---|---|
| Jul 2024 | Series B | $130M | $700M | Andreessen Horowitz |
| 2023 | Series A | ~$30M | — | Index Ventures |
| 2021 | Seed | ~$1M | — | Floodgate, Peter Thiel (named angel) |
Cumulative ~$161M raised through the July 2024 Series B at $700M post-money. The Series B was led by Andreessen Horowitz with Index Ventures and Google Ventures co-investing per the Hebbia press release and TechCrunch coverage, with the capital explicitly earmarked for product depth on Matrix orchestration plus international expansion into EMEA. TechCrunch’s reporting on the Series B disclosed $13M ARR at round close in mid-2024 alongside a profitability claim — unusual at AI-application scale and a contributing input to the D4d capital-position sub-rubric score of 8. ARR is reported to have grown approximately 15x over the 18 months following the Series B per Sacra and gitnux secondary coverage; we mark the implied range at $80-150M for mid-2026 but decline-to-publish a precise current ARR figure pending primary disclosure.
Competitive Landscape
Hebbia’s competitive set sits in three concentric rings: incumbent finance-research and market-intelligence platforms with proprietary corpus assets (AlphaSense with the Tegus expert-call corpus, BloombergGPT inside the Bloomberg Terminal), independent finance-AI vertical plays (Rogo as the closest agentic-platform rival), and foundation-model providers plus horizontal enterprise-search aggregators competing at the workflow layer (ChatGPT Enterprise with Morgan Stanley GPT as the named bulge-bracket reference, Glean on horizontal enterprise-search breadth). Hebbia’s distinguishing position in the set is the Matrix multi-step orchestration depth over long-document corpora plus FlashDocs synthesis plus the 1B+ pages processed milestone as the data-gravity moat in financial services.
| Competitor | Positioning | Distribution edge | Threat profile |
|---|---|---|---|
| AlphaSense | Market-intelligence and financial-research platform — AI-powered search and summarisation grounded in a proprietary corpus of broker research, expert calls, earnings transcripts and regulatory filings. Acquired Tegus in 2024 to deepen the expert-call corpus and consolidate the buy-side research-platform position. | Direct enterprise GTM into investment banking, asset management, corporate strategy and consulting; mature multi-thousand-customer installed base across the buy-side and sell-side. | High — the symmetric corpus-and-distribution incumbent on financial-research workflow; competes head-to-head on buy-side procurement where Hebbia is differentiating on Matrix multi-step orchestration and customer-data-grounded agentic-research rather than on third-party-corpus depth. |
| Rogo | Agentic AI research and analysis platform for investment banking and financial services — deep-research agents, financial-modelling support and earnings-call analysis deployed at 250+ institutions including Lazard, Jefferies, Rothschild, Moelis, Nomura, BofA and Wells Fargo. Closely matched independent finance-AI vertical play; the closest agentic-platform rival in the set. | Direct enterprise GTM into investment-banking IT and research procurement plus embedded engineering teams at marquee customer banks; $160M Series D at $2.0B post-money in April 2026. | High — the closest independent agentic-platform rival on the same agentic-research selection committee at marquee buy-side and sell-side accounts; competes head-to-head on capability and finance-vertical positioning with materially deeper capital position at the Series D stage. |
| BloombergGPT / Bloomberg AI (Bloomberg L.P.) |
Finance-domain large language model and AI features layered into the Bloomberg Terminal — the symmetric incumbent on terminal-distributed finance AI grounded in Bloomberg’s proprietary market-data and news corpus. | Lives inside the Bloomberg Terminal subscription footprint across the buy-side and sell-side — the deepest channel-control argument in financial services. | High — the corpus-owning incumbent with the strongest installed base in finance; the same channel-control logic that CoCounsel and Lexis+ AI carry in legal AI. |
| ChatGPT Enterprise (finance use cases) (OpenAI) |
General-purpose enterprise reasoning and research agent — not finance-specialised but credible at financial research and used at investment banks under enterprise contracts (Morgan Stanley GPT, named bulge-bracket deployments). | Available via OpenAI enterprise sales to any IT-approved customer; no finance-vertical procurement specialisation required. | High and asymmetric — foundation-model provider competing at the workflow layer; OpenAI is one of Hebbia’s named foundation-model suppliers and the workflow-layer competitor with the Morgan Stanley GPT bulge-bracket reference. The same indirect-competitor risk that Harvey faces in legal AI. |
| Glean (finance vertical use) | AI-powered enterprise search and workplace assistant with horizontal connector breadth across 100+ SaaS sources; named as a horizontal substitute in finance procurement where buyers want knowledge-graph breadth rather than long-document depth. | Direct enterprise sales into IT and knowledge-work buyers; 27+ country footprint with $200M ARR by Dec 2025 and Fortune 500 reference accounts. | Medium — flanking play from a horizontal enterprise-search angle; less direct on long-document agentic research where Hebbia leads, but credible on the broader knowledge-work procurement perimeter at the same Fortune 500 buyers. |
Pricing benchmark: enterprise finance-AI pricing in the buy-side and sell-side market is negotiated per-seat with reported figures clustering in the high-hundreds to low-thousands of dollars per user per month for the named-incumbent set; Hebbia has not publicly disclosed list pricing. The competitive cut is on long-document orchestration depth (Matrix and FlashDocs versus AlphaSense corpus, BloombergGPT terminal integration and Rogo’s embedded-engineering customisation) rather than headline per-seat price; Hebbia’s structural differentiator is the multi-step agentic-workflow capability over private deal documents and diligence binders that the customer brings, rather than over a third-party-corpus the vendor owns.
Potential Risks
The case for Hebbia at IM Framework 7.09 rests on the 50+ Fortune 500 footprint (33% top-50 US asset-manager penetration with named flagship deployments at Centerview Partners, Charlesbank, Fenwick & West and Ropes & Gray), the Matrix multi-step orchestration plus FlashDocs synthesis as the product moat, the 1B+ pages processed milestone as the data-gravity print, and the top-decile capital position at the $700M post-money Series B with the profitability claim attached. The case against splits into five risks of differing magnitude — with the foundation-model supplier concentration the most structural, the AlphaSense-and-BloombergGPT symmetric-incumbent head-to-head the most active competitive constraint, the Rogo capital-position gap the closest-independent-rival risk, the FINRA 2026 GenAI accountability regulatory perimeter binding on the customer base, and the founder-concentration plus undisclosed-CRO executive-bench depth the most distinctive governance pattern in the score.
Foundation-model supplier concentration — OpenAI plus Anthropic via Azure
Hebbia’s named foundation-model stack is OpenAI and Anthropic routed through a Microsoft Azure partnership disclosed in 2024-25 product and partnership coverage, with multi-model orchestration inside Matrix as the architectural mitigation. The D4a supplier-diversity sub-rubric scores 5 on this evidence — the lowest score in the defensibility composite. Multi-model routing addresses the foundation-model-supplier concentration question in principle (capability can be re-routed inside Matrix as alternative providers ship competitive frontier capability) but the asymmetric overlay is that OpenAI itself competes at the workflow layer with ChatGPT Enterprise and the Morgan Stanley GPT bulge-bracket reference. The same indirect-competitor risk that Harvey carries in legal AI and Rogo carries in investment banking.
Symmetric-incumbent head-to-head — AlphaSense and BloombergGPT on finance-research
On finance-research and market-intelligence workflow specifically, AlphaSense (deep corpus of broker research and expert calls with the Tegus acquisition behind it) and BloombergGPT inside the Bloomberg Terminal remain the deeper corpus-and-distribution incumbents on the buy-side and sell-side. Hebbia’s differentiation is the Matrix multi-step orchestration and FlashDocs synthesis over customer-private long-document corpora — not corpus depth that the vendor owns. The strategic risk is that buy-side and sell-side procurement defaults to the corpus-owning incumbent on terminal-distributed workflows, leaving Hebbia to compete on the customer-data-grounded agentic-research lane where the multi-step orchestration depth lands; the 50+ Fortune 500 footprint plus 33% top-50 asset-manager penetration is the offsetting evidence that the customer-data lane is durable.
Closest-independent rival pressure — Rogo’s $2.0B Series D capital position
Rogo (the closest independent agentic-platform rival) closed a $160M Series D at $2.0B post-money in April 2026, with Series C to D in under four months and 35,000 users across 250+ institutions including a marquee broker-dealer roster (Rothschild, Jefferies, Lazard, Moelis, Nomura, BofA, Wells Fargo) and JPMorgan Growth Equity as investor-and-customer. Hebbia’s capital position is materially smaller ($161M cumulative at the July 2024 $700M post-money Series B) and the gap-to-Series-C is now the active question. The competitive risk is not that Rogo beats Hebbia head-to-head — the Matrix-and-FlashDocs orchestration depth, the 1B+ pages processed milestone and the 33% top-50 asset-manager penetration are durable — but that the capital-position gap shapes the depth of product investment and embedded-delivery cadence over the next 12-18 months.
Regulatory exposure — FINRA 2026 GenAI accountability binding on customer base
FINRA’s 2026 Annual Regulatory Oversight Report (published December 2025) explicitly pivots to GenAI accountability and AI supervision in broker-dealer examinations, and SEC plus FINRA expectations on AI in Regulation Best Interest, Best Execution and vendor-oversight are tightening. Hebbia’s customer base (buy-side asset managers, hedge funds, corporate-development teams plus the emergent legal-services vertical via Ropes & Gray) is exactly the regulated universe in scope. The D4c regulatory-exposure sub-rubric scores 7 on the framing that Hebbia is beneficiary not target (agentic-workflow vendors supplying supervisory-documentation and audit-trail capability tend to win in tightening regulatory cycles) but the compliance burden lands on the customer surface and on Hebbia’s product roadmap concurrently.
Founder-concentration and executive-bench depth
Hebbia is founder-led with George Sivulka as the principal public voice and the AI-research recruiting flywheel anchored on his Stanford EE-PhD-program and ex-OpenAI / ex-Anthropic network (~30% of headcount on his own framing). No CRO, CFO or CTO has been publicly named as a separate appointment distinct from the founder-and-engineering bench at time of writing, and the D4e key-person dependency sub-rubric scores 5 on that basis. The bull case is that the AI-research bench is the structural source of the Matrix multi-step orchestration differentiation that the disruption composite scores against. The bear case is that scaling against Fortune 500 buy-side and sell-side anchor customers at the $80-150M ARR implied stage without a disclosed senior commercial layer is a known load-bearing risk; the executive-bench appointments and the Series C valuation print over the next 12 months are the material watch-items.
Recent IM Coverage
Show recent press coverage of Hebbia
- 2026 — Hebbia processes one billion pages as financial institutions deploy AI infrastructure at unprecedented scale.
- Jul 2024 — AI startup Hebbia raised $130M at a $700M valuation on $13 million of profitable revenue.
- Jul 2024 — Hebbia raises $130M Series B to scale Matrix — the agentic platform for knowledge work.
- 2025 — Hebbia company analysis — ARR trajectory, customer breakdown and Series B context.
- Mar 2026 — Hebbia triples UK and Europe team as EMEA expansion accelerates.
- Dec 2025 — FINRA publishes 2026 Regulatory Oversight Report — GenAI accountability and AI supervision in scope.
- 2025 — Hebbia Matrix product deep-dive — multi-step agentic-orchestration over long-document corpora.
Curated feed of named-source coverage — Hebbia’s own press releases and product blog, TechCrunch’s Series B coverage with the $13M profitable-revenue print, analyst coverage (Sacra’s company page), named-author technology and finance press (Global Banking & Finance Review on the 1B-pages-processed milestone, TechFundingNews on EMEA expansion), and FINRA’s regulatory oversight publication as the binding regulatory framing constraint. Excludes paywalled article bodies of The Information, WSJ, FT and Bloomberg (headline plus free-snippet only), PR-wire reposts of the same release and unsourced AI round-up pieces.
Show the source register for the figures on this page
IM operates a primary-source-where-possible discipline. The figures above come from:
- Customer base — 50+ Fortune 500 and 33% top-50 asset managers: Hebbia discloses 50+ Fortune 500 customers and a third of the top-50 US asset managers with named flagship deployments at Centerview Partners, Charlesbank, Fenwick & West and Ropes & Gray per Hebbia’s Series B announcement and Sacra’s company analysis. The 1B-pages-processed milestone in Global Banking & Finance Review is the data-gravity print on the financial-services deployment footprint.
- Funding to date — $161M cumulative at $700M post-money: $130M Series B at $700M post-money closed July 2024 with Andreessen Horowitz leading and Index Ventures and Google Ventures co-investing per TechCrunch’s coverage and Hebbia’s Series B blog. TechCrunch’s reporting disclosed $13M ARR with a profitability claim at the round close — unusual at AI-application scale. Series A and seed history per Sacra.
- Revenue trajectory — $13M to ~$80-150M implied: Hebbia reported $13M ARR at the July 2024 Series B close per TechCrunch, with secondary coverage in Sacra describing approximately 15x growth over the subsequent 18 months. IM triangulation implies an $80-150M range for mid-2026; we decline-to-publish a precise current ARR figure pending primary disclosure.
- Regulatory framing — FINRA 2026 GenAI accountability: FINRA’s 2026 Annual Regulatory Oversight Report (published Dec 2025) pivots explicitly to GenAI accountability and AI supervision in broker-dealer examinations. Hebbia’s buy-side asset-manager, hedge-fund and corporate-development customer base sits inside the regulated universe in scope; the D4c sub-rubric reflects Hebbia’s beneficiary-not-target framing on supervisory-documentation and audit-trail product surface.
Methodology & Disclaimer
For metric definitions, source-tier hierarchy, and decline-to-publish rules, see the tracker methodology. Confidence dots (• green / • amber / • red) follow the same convention as the AI Tracker.
Spotted a figure you believe is wrong? Send corrections to info@informationmatters.net.
Information Matters Framework scores are the considered opinion of the IM team — human and AI — applied to publicly-available evidence under a disclosed methodology. They are not statements of fact about the companies scored and they are not investment advice.
