MongoDB Investor Day 2026: In Good Hands!

The speakers were Dev Ittycheria (Interim CEO), Mike Berry (CFO), Ben Cefalo (CPO, Core), Pablo Stern-Plaza (AI and emerging products), Ryan Ban (CRO), Erica Volini (Chief Customer Officer), plus customers from Citizens Bank and Emergent.

The bottom line

Management had three jobs: show the CEO change is a non-event, launch three products, and raise the long-term model. The main financial news:

• Long-term model shortened from 3–5 years to 3 years and raised:
  • Total revenue growth: from high teens to 20%+
  • Atlas growth: from 20%+ to mid-20s
  • EA: from mid-single digits to 10%+, measured on ARR
  • GAAP profitable from FY27 onward
• An extra $1bn buyback authorization.
• New disclosures: AI-native revenue share, total subscription ARR in dollars, the >$5m customer cohort, and ARR by customer size.

The underlying message is “solve forward.” Management says the business doesn’t need fixing. The product and go-to-market work belongs to a team of about 5,700 people, not one CEO.

Also we believe Dev Ittycheria’s return is ratter positive and we keep our positive view on the company’s strategy for the Age of AI!

The CEO transition

• CJ resigned “last Friday” and is going to Meta. Dev Ittycheria is Interim President and CEO, and the Board has started a search.
• Berry said they “can’t say anything more than is in the 8-K.”
• The rest of the management team and four directors (Roelof, Hope, Chip, Tom) stood up in the room to show continuity.

Q&A on CEO Exit

• What CJ did well (Keith Bachman, BMO): Dev said CJ moved selling up to the C-suite and used his ServiceNow contacts to close key accounts faster. Dev also said the go-to-market team’s groundwork did most of the work, so it “wasn’t a one-man show.”
• Dev’s priorities: keep the momentum, execute sharply on product and go-to-market, protect the culture, and hand over cleanly.
• The next CEO: someone who understands both technology and how to win and keep customers, not “one-dimensional.” The Board and executives will both evaluate candidates. Dev said his “phone has been ringing off the hook.”
• Managing change (Sanjit Singh, MS): Dev said MongoDB has handled many executive changes before. He replaced the CRO right after the IPO and was acting CRO for almost a year. Some recent departures were simply people who were “tired” after a long run.
• The EA strategy (Rishi Jaluria, RBC): Dev said the decision to invest in EA (Enterprise Advanced, the self-managed product) was made while he was still CEO, not by CJ. Large financial institutions asked for it, and AI is pushing buyers toward on-prem, neoclouds and open-source models for IP protection.

Management worked hard to say the strategy existed before CJ, especially the EA reinvestment. That supports continuity, but it also plays down what CJ changed. Berry later admitted the company “didn’t have a CRO for 6 months” and CJ “had to jump into that void.” That was a real execution gap, now filled by Ryan Ban.

Dev Ittycheria: market and strategy

• The pitch: MongoDB is becoming an “intelligent data platform.” Dev described three stages:
  1. Prove a document database can run mission-critical work.
  2. Win in the cloud against AWS and GCP.
  3. Broaden the platform: Vector Search, Full-Text Search, and Voyage embeddings and reranking.
• Scale: the market is expected to reach about $200bn by 2030, roughly double 2024. MongoDB revenue is up 5x since 2020 and on guidance reaches about $3bn this year. Customer count is up 3x over the same period, so a lot of growth comes from expanding within existing customers.

• Three reasons AI helps MongoDB:
  1. More software gets written, so more databases are needed.
  2. New kinds of workloads: agents, memory, knowledge work.
  3. New customers: frontier labs and AI natives.
• What AI workloads need: live, real-time data; precise retrieval; elastic scale for unpredictable agent traffic; trust (security, governance, availability); and the ability to run anywhere.
• The swipe at Postgres: moving between Postgres versions (AlloyDB to Aurora) requires a lot of code changes, so “Postgres is portable” is a misconception.
• Customer examples:
  • A large healthcare provider: spend up 22x in 3.5 years. Its claims platform handles 15m transactions a day at under 200ms. A chatbot built on top cut total cost of ownership by about 80% and improved answer accuracy by about 60%.
  • A media company: ARR up 3x in 12 months, growing from one app to 20 streaming services, now adding scene-level video search.
  • A European retailer: spend up 7x in 3 years. It moved an ERP application off Oracle, replaced Elastic with Vector Search, and added an AI chatbot.
• Frontier labs: not named, but they use MongoDB for model training, chat systems, feature testing, experimentation and analytics. About two-thirds of AI natives use more than one product.
• Managed MCP connector: launched about a month ago. It gives one-click connection to Claude Code, Codex and Grok Build, and interest is up 8x in 6–7 weeks. It also gives MongoDB its first view of agent traffic versus human traffic on Atlas.

Ben Cefalo: core products

Why the document model matters

Ben argued that JSON is “the default language of AI.” Native JSON beats Postgres JSONB because:

• large documents are handled better
• updates don’t rewrite the whole stored object
• the query language is designed for JSON

He called JSONB “forcing flexible data into a rigid table.”

Pillar 1: Run Anywhere, with EA reinvestment

• Why customers want it: regulation and data sovereignty, protecting IP and data from AI training, and resilience against outages or vendor changes.
• Examples: Nationwide UK runs both EA and Atlas. A major US bank runs more than 100 production apps on EA. Zoom runs EA in dozens of environments.
• Atlas statistics: 10 years old; 130 cloud regions across AWS, Azure and GCP; clusters can span several clouds; about 20m users; more than 3 trillion queries a day (about 35m per second, more than double a few years ago).
• Developer funnel: about 500k free-tier clusters a quarter and 150m community downloads. Ben’s rebuttal: “Not everyone starts with Postgres.”
• New on EA: Search and Vector Search became generally available (GA) for self-managed customers in June. Demand came immediately, for example a multinational bank combining operational data, vectors, embeddings and full-text search on EA.

Pillar 2: MongoDB 9.0

• Gains versus 8.0: up to 20% more transactional throughput, 35% faster reads, 30% faster updates, and 2x throughput on larger instances. It is about 100% faster than releases from before 2023.
• 8.0 adoption: about 80% of the Atlas fleet runs 8.0, and customer satisfaction and retention went up after it launched.
• The Jevons argument: better price-performance has historically led to more consumption, not less. Coinbase tested 9.0 at production scale.
• Queryable Encryption (only MongoDB has it): data and queries stay encrypted. Examples were French law-enforcement IT and a financial institution protecting customer personal data.

Pillar 3: Atlas Infinite

• Atlas’s biggest architecture change since 2016: storage and compute are separated so each scales on its own. It was more than 2 years in development and targets bursty, unpredictable agent traffic.
• It is a deployment option inside Atlas, so customers don’t change application code. The existing product is renamed “Atlas core database” and will keep getting investment.
• Security: SOC 2, ISO and PCI at launch. Data is encrypted before it leaves the compute layer, and “neither the cloud providers nor we can see” customer data.
• Private preview results: Icon Solutions processed more transactions per second, PicPay ran 4x its normal peak, and Carrier gave a testimonial. Okta presents at .local.
• Pricing stance: “this is not about spending less.” Customers stop paying for idle capacity.

Pablo Stern-Plaza: AI and Atlas Agent Engine

Getting picked by builders

• Answer engines: accuracy on prompts where MongoDB is a strong fit improved more than 50% in a year.
• Frameworks: LangChain and Mastra.
• Coding agents: marketplace plug-ins for Codex, Claude, Grok Build, Devin and Cursor launched mid-August.
• AI platforms: Vercel v0 now, with Monk, Retool and Emergent announced at .local.

Retrieval

• The stack is Full-Text Search, then Vector Search, then Voyage embeddings and rerankers (Voyage was acquired in Feb 2025).
• Voyage models “consistently rank at the top” of Hugging Face retrieval benchmarks, ahead of Google, Cohere and OpenAI.
• Vector Search customer count is up 4x in 2 years. The Voyage customer base has doubled quarter over quarter for 2 quarters.
• Examples: the Financial Times uses auto-embeddings, launched 7 weeks ago. Gravity AI picked Voyage after its own benchmarks.
• Pablo described a cross-sell loop: database customers adopt Vector and Voyage, and Voyage customers lead to database conversations.

Atlas Agent Engine

Pablo called it “one of the largest expansions of our platform in company history.” It answers three gaps customers raised:

1. Memory: managed long-term and short-term memory that the customer owns. In the Q&A he said long-term memory is broken into four types: procedural, episodic, taxonomic and semantic, with Voyage and Vector retrieval underneath. It is offered as an SDK and through MCP.
2. Security: built on Atlas’s enterprise security, with governance and guardrails designed in from the start.
3. Openness: works with any cloud or self-managed setup, any model, any framework, any observability tool and any data source.

• Design partners: Paysafe (agents that investigate payment errors) and CDK (agents that extract auto-deal terms). Accenture gave a testimonial.
• Commercial point: every Atlas customer already has access in public preview, so there’s no new contract or sales motion. That means “tens of thousands” of customers can use it now. Revenue comes through runtime and memory, plus more database consumption.

Customer panel: Citizens Bank (Vikas Agarwal, Head of Data Platforms)

• Five years on MongoDB; many Tier-1 apps run on it.
• Built a patent-pending multi-agent framework. Agents watch platform events, diagnose problems together and hand a fix to a human, which reduced incidents.
• Insists on being LLM-agnostic: “3 months back… Claude is the best… today… Codex.” The bank also builds small language models for routine tasks.
• Liked memory (the bank built something similar on MongoDB about 4 years ago, and it’s faster than calling an LLM), platform neutrality, and running air-gapped behind three layers of guardrails. Infosys introduced the bank to MongoDB.

Customer panel: Emergent (Madhav Jha, CTO)

• No-code app builder with 15m users, 18m apps built, and about 150k active paid apps, all deployed on MongoDB.
• Started on Postgres in its research-lab phase. Schema migrations broke its coding agent (on Claude 3.5), so it moved to MongoDB after running evals.
• Its main point: users who aren’t programmers can’t fix a failed migration, so a flexible schema matters. Some of its users’ apps grow about 5GB a week and hold 15m documents.
• Wants from MongoDB: uptime, scale, cost, and scale-to-zero.

Ryan Ban (CRO): go-to-market

• Why he joined: people, technology and timing. AI coding tools (Cognition) now make Oracle, SQL and relational migrations take “days, not weeks and years.” He said this opens a $10bn+ new TAM. He compared MongoDB with Confluent, where he was CRO before IBM bought it, and called this “a much larger opportunity.”
• Recent results:
  • 6 straight quarters of Atlas growth at about 29%.
  • Self-serve adds about 2,500–3,000 customers a quarter. The transcript reads “25,00,” which is likely a typo.
  • Record net new customers in Q2, for more than 70,000 in total.
  • +17% year over year in both the $100k+ and $1m+ ARR customer groups.
• Three investments:
  1. Enterprise and executive buyers, plus AI and digital natives. He called this “an and, not a replacement” for the developer-led motion. It includes tighter segmentation, executive briefings, earlier investment in new products, higher seller incentives for large deals, and depth in financial services, healthcare and the public sector.
     • US Fed: pursuing FedRAMP High, only “10% penetration,” hiring a new General Manager, and building on the “recent acquisition of Clarity.”
     • Japan: new legal entity and country leader.
  2. Product specialist teams so new products can scale without slowing core Atlas and EA selling. The model is already used for vector, embeddings and reranking, and is being extended to Agent Engine.
  3. Modernization and migration through AMP (MongoDB’s app modernization platform) and Cognition.
• He called this the “path to $5bn+.”

Erica Volini (Chief Customer Officer): partners

She came from 20+ years at Deloitte and then ServiceNow (during its $5bn to $10bn phase). She described four partner levers:

• The AI ecosystem: coding agents, AI platforms, answer engines and frameworks.
  • Devin (Cognition) for legacy code.
  • Vercel v0: Atlas provisions in one click through managed MCP.
  • Mastra runs all its retrieval and memory on one Atlas cluster.
  • LangChain-linked revenue is up more than 460% year over year, and builders pick MongoDB inside LangChain 220k times a week.

• Systems integrators: the top 6 SIs named MongoDB a strategic partner for agentic delivery and use Agent Engine internally. One saw 984% faster queries and a 63% drop in latency against sovereign AI offerings. Infosys’s Topaz framework embeds MongoDB. Gartner expects SI AI engagements to rise 15% by 2029.

• Hyperscalers and neoclouds:
  • About half of MongoDB revenue is supported by a hyperscaler partner. Marketplace revenue grows faster than total revenue.
  • Partner of the Year at AWS, GCP (7 years running) and Microsoft; a top-3 Azure partner.
  • Talking with AWS and Microsoft about putting Agent Engine memory into their AI platforms. DigitalOcean was given as a neocloud example.
  • Nationwide: £1.5bn moved every 24 hours, 3m payments at peak, from zero code to production in under 8 months, with Accenture and a cluster stretched across several clouds.

• Industry software vendors (ISVs): BigID, a health-tech provider, an AI support line using Queryable Encryption, and Emergent. She said Emergent produced “nearly 2 million” apps in 4 months with 50k+ in production. That doesn’t match Emergent’s own figures of 18m apps and about 150k paid apps, probably because the two counts measure different things.

Mike Berry (CFO): financials

Last year’s commitments versus results

Growth

• TAM: $200bn by 2030 at a 12% CAGR (IDC), covering both transactional databases (OLTP) and analytical ones (OLAP). MongoDB’s share is about 2.2%.

• 5-year CAGRs: total revenue 28%, Atlas 36%, EA and other (including services) 14%. Net dollar retention is 122%.
• Incremental Atlas dollars: in FY25 the extra dollars added each year had flattened. They have now grown for 6 straight quarters, which is what Berry means by acceleration.
  • Growth is led by large US and EMEA enterprises. Self-serve and sales-led are both growing.
  • Berry said Atlas growth has not been hurt by EA’s recovery. The transcript says “VA,” which reads as EA.

• EA inflected in 3Q FY26. ARR has grown double digits for 3 straight quarters, and this happened before Search and Vector reached EA. Technology companies are now adopting EA, not just regulated industries.
  • About two-thirds of EA customers with $100k+ ARR also use Atlas, and that group grows faster than the total. Berry used this to argue there’s no cannibalization.
• Customer cohort CAGRs: total customers 23%, $100k+ roughly in line, $1m+ 34%, $5m+ 57% (new disclosure).
  • A new chart showed ARR by customer size back to Q2 FY18.
  • Berry’s message: “a customer play, not a workload-by-workload play.” Individual workloads can plateau, and expansion makes up for it.
• Total subscription ARR for Q2 FY25, FY26 and FY27 is newly disclosed once a year. Both Atlas and EA ARR accelerated each year. The dollar figures were on the slide but are not in the transcript.

Margin, GAAP and cash

• OpEx: grew mid-teens last year and is guided up about 20% this year, with R&D up about 30%. Future savings could come from AI and offshoring.
• GAAP: GAAP profitable at $0.77 EPS at the high end of guidance.
• Stock comp: down from 30% of revenue five years ago to 20% year to date. No target was given, but Berry expects it to keep falling.
• Cash conversion used to be about 50%. It improved because the shift from EA to Atlas billing has normalized and working capital is managed better (for example, paying vendors more slowly). Free cash flow was close to $500m last year.

AI metrics (one-time definitions, which may change)

• Definition of an AI native: frontier labs, AI infrastructure companies, and AI-native software that “would not exist without” generative AI.
• Multi-product use among $100k+ customers rose from 34% to 48%.
• Atlas ARR customers with at least one AI use case rose from 30% to 40%.
• Top 3 enterprise use cases: real-time context from company data, autonomous agents, and AI customer support.
• AI natives are under 5% of Atlas revenue, about 3–5%. Their Atlas revenue grew about 67%, then 129%, then 165% year over year across the three periods shown. Their customer count doubled in 2 years.

When the new products start contributing revenue

Product	Monetization	Timing
EA Search and Vector	new licenses	small in FY27, building in FY28
Atlas Infinite	Atlas consumption	GA late FY27 or early FY28; meaningful revenue only in FY29
Atlas Agent Engine	drives Atlas consumption	GA late FY27 or early FY28; momentum in FY28

Berry said EA is the line that will move most in FY28.

New 3-year model

• Total revenue 20%+.
• Atlas mid-20s, using the same cautious guidance framework.
• EA ARR 10%+. Targets are based on ARR to strip out lumpy multi-year deal revenue. Berry: “I’m not going to… bet on a multiyear deal.”
• Operating margin +100–200bp a year, “not a ceiling.”
• FCF conversion 80–100%.
• GAAP profitable in FY27 and after.
• Rule of 40: guided to 44 this year. The aim, not a commitment, is to be close to 50 by the end of the 3 years.
• Possible upside to Atlas: enterprise AI adoption, a larger AI-native share, faster new-product uptake, app modernization (AMP and Cognition), and partners.
• Capital allocation: more than $2bn of cash, targeted M&A (technology and talent, or adjacent markets, like Voyage), plus $1bn of new buybacks, and buybacks to partly offset dilution.

Final Q&A: main exchanges

• On Atlas Agent Engine’s competition: Pablo said customers are currently piecing together building blocks from hyperscalers, LLM providers and vertical SaaS tools, which puts “a lot of onus on the builders.” He named three differences: native, configurable memory; no lock-in to a model or cloud; and security designed in from the start: “we took security as the first principle and actually built it into the system. So you don’t have to actually build security on top or bolt it on top.” He said that matters most to regulated customers like Citizens Bank.

• On consumer-facing agents (like Meta Muse, which runs on Postgres): Dev listed four things agents need. First, fresh data: “the freshest data in your system is live operational data.” Second, precise retrieval, which he compared to sending someone to the right section of a library instead of making them read every book. Third, elastic scale, because “agents could be working all night in an unpredictable manner.” Fourth, state, now addressed by the new memory capabilities. Pablo added that frontier labs reaching production scale are “moving from solutions like PostgreS onto us, both across pre-training and post-training.” He also said Agent Engine components can plug into agents from other vendors.

• On analytics databases (OLAP): Mike said the focus stays on transactional databases (OLTP). Any M&A would mostly speed up the roadmap. Analytics is becoming “a little bit of a gray area,” but “we also don’t want to get diffused outside of our core focus,” including keeping sellers focused. Ben was more dismissive: “I describe this as cute that OLAP players think they could be credible in the OLTP space.” He added that OLAP vendors moving into OLTP just confirms MongoDB’s view that “OLTP was the higher ground for AI.”

• On confidence in the Atlas outlook: Mike pointed to expansion inside existing customers: “yes, workloads will plateau. But the team… have done such a wonderful job of cross-selling and upselling.” He said multi-product adoption among $100k+ customers is still below half and doesn’t yet generate much revenue, so it’s upside. He also cited Japan, US Fed and Voyage. He then acknowledged the leadership gap: “we didn’t have a CRO for 6 months, and now we do.” CJ had to fill that role, and with a CRO in place, Mike said go-to-market is in better shape than a year ago.

• On Atlas Infinite cannibalization: Ben noted Infinite is still early: “what we announced today is preview, right? Just on AWS.” He expects no rush to switch: “I don’t see a mass movement of existing customers moving to Atlas Infinite. They’re happy where they are. It’s a predictable pricing model.” He thinks new agent workloads will start on Infinite. He framed the flexibility to move between Atlas, Infinite and self-managed as a way to let customers prototype more, not cut spend.

• What’s the next big thing?: Mike said MongoDB has gone from one line of business to “a seat at the table at a more executive level,” across several business lines. Customers like the product roadmap. Years of technical debt are finally being addressed because AI tools make “being able to migrate in literally days and weeks instead of years” possible.

• On pricing across segments: Ryan said “it starts by listening” and they’ll adjust over time “keeping an eye on margins,” helped by having both Atlas and EA as hybrid options. Ben contrasted MongoDB with peers that “bill customers… by read units and write units. We don’t do that.” Atlas Dedicated customers pick capacity and storage, and the bill stays flat unless they go over it. So a weekend agent experiment that “throws 100,000 queries at it” costs nothing extra, and enterprises are “not worried about the runaway bills.”

• On winning back workloads: Pablo said the starting point is showing up for builders in answer engines and marketplaces. He gave a concrete case: a “frontier lab that migrated to us in 4 weeks off of a PostgreS database for conversation history.” Dev cited ElevenLabs, which “started on Firestore” and moved to MongoDB once it hit scale. Ben was candid about the funnel problem. AI models model data in JSON, which helps, but “the mistake that the models make is that it picks PostgreS on JSONB most of the time.” Once those apps scale, he said, “we’ll migrate them every single time, no problem.”

• On application modernization (AMP): Dev admitted the delay: “what we underestimated, candidly, was that the maturity of the AI models took longer than we had expected.” He said Mike told the team not to talk about AMP again until they had something solid. Cognition, announced that morning, now handles the hard part of rewriting application logic, while MongoDB supplies data-migration tooling and guardrails. He added: “we didn’t want to kind of overplay our hand today.” AMP sits among the upside cases in the model.

• Sanjit Singh, Morgan Stanley: if developers stop writing code, does MongoDB’s developer-first pitch weaken? Dev agreed with the premise: within 3, 5 or 10 years, “humans will no longer write code.” He argued scalability and JSON (“we speak the language of AI”) still matter. He conceded: “we may have been a little slow like with other companies like Supabase… Those are easy gaps for us to fill.”

• On memory, Ben said that until now “it was just another database workload for us.” Pablo explained the new memory is a composable product. It breaks long-term memory into procedural, episodic, taxonomic and semantic types, and retrieves it through Vector Search and Voyage for fewer turns and lower token cost. “We provided as an SDK, and we’re externalizing also through MCP.”

• On the mid-20s Atlas target : The drivers are the move upmarket, geographic expansion, new products and AI natives. The core business “as a percentage continues to… slow down, but the dollars continue to grow.” He added: “we really haven’t baked in a lot of upside for enterprise AI.” Dev cited the 8.0 precedent. Faster performance didn’t shrink bills, because customers “wanted to drive that car more because they put more workloads on it.” He expects 9.0 to do the same.

• On Infinite cannibalization he answered “No.” MongoDB was seen as a “Ferrari” for specialized workloads, and “by increasing the price performance, we make it just that much easier for people to put more and more workloads on MongoDB.”

• Steven Koenig, Macquarie: coding agents default to Postgres, so can MongoDB get into enterprise coding agents from the top down? Pablo said MongoDB does well on enterprise approved-vendor lists but hadn’t shown up well inside coding agents. In large companies, “they’re not going to let developers build production scale applications using any single solution. Those typically go through the architecture team.” Through Cognition, he said, “the coding agent and the platform is being made together.”

• On hands-on support: Erica said MongoDB is building forward-deployed engineers (FDEs), starting in product and moving to professional services. Customers who set up schemas and indexes correctly get more value “and then they want to consume more.” Pablo added that FDEs already work on AMP and Agent Engine and help “reduce the cycle time for new innovation.”

What stands out

Strengths

• Two growth engines. EA is growing double digits again, its inflection came before the new features, and it overlaps heavily with Atlas. That weakens the “self-managed is in decline” argument and supports a 20%+ total revenue target that doesn’t depend on Atlas alone.
• Growth from moving upmarket. The $5m+ cohort grew at 57% and $1m+ at 34%, against 23% for all customers. Combined with 6 quarters of rising incremental Atlas dollars, this is the clearest evidence behind the higher targets.
• Delivery record. Every 2025 commitment was beaten: margin expansion of +360bp and +250bp against a 100–200bp promise, and FCF conversion of 108%. Berry repeats that the new targets use the same cautious framework.
• A consistent product story. 9.0, Infinite and Agent Engine all serve one thesis: agents need live operational data, precise retrieval, elastic scale and state. Run Anywhere and model neutrality are the main differences from hyperscalers and LLM providers.

Tensions and open questions

• The mid-20s Atlas target sits below current growth. FY26 was about 29% and the FY27 guide is about 27%, so the target leaves room to slow down. It still drew the most pushback. Management says it bakes in “little enterprise-AI upside.”
• AI natives are a small slice. At 3–5% of Atlas revenue, against a peer about 3x higher, MongoDB is under-represented where AI demand is most concentrated today. Growth is fast but from a small base.
• Management admitted to competitive gaps:
  • coding agents usually pick Postgres JSONB
  • MongoDB was slow compared with Supabase on agent-friendliness
  • AMP was delayed because AI models matured more slowly than expected
  The plan is to win workloads back when customers reach scale (Emergent, ElevenLabs, a frontier lab), which is reactive rather than winning at the start.
• New products pay off late. Infinite produces meaningful revenue only in FY29, Agent Engine gains momentum in FY28, and EA search builds in FY28. Most of the new launches are upside to the 3-year plan rather than its foundation.
• Leadership. An interim CEO, a CRO in the job only a short time, and a new CCO all lead the execution phase. Management says the team is aligned, but the strategy will outlast the people currently presenting it.

Thank you!

CoreValue Research Team