리서치 솔루션스(RSSS) 2026 회계연도 4분기 실적 발표 콘퍼런스 콜: B2B ARR 14% 증가, AI ARR 80만 달러 달성

핵심 요약

• 2026 회계연도 4분기 매출은 플랫폼 구독 매출 증가에도 불구하고 트랜잭션 매출이 6.7% 감소함에 따라 전년 동기 1,240만 달러에서 1,210만 달러로 감소했습니다.
• 총 연간 반복 매출(ARR)은 전년 동기 대비 7.8% 증가한 2,250만 달러를 기록했습니다. B2B ARR은 80만 달러의 AI 관련 ARR을 포함해 14.1% 증가한 약 1,620만 달러를 기록했습니다.
• 2026 회계연도 전체 총이익률은 260베이시스포인트(bp) 확대된 51.9%를 기록했습니다. 전체 매출 감소에도 불구하고 매출총이익은 3.6% 증가한 2,510만 달러를 나타냈습니다.
• AI 도입으로 연구 활동이 에이전트 기반 인터페이스로 이동하고 있습니다. 사이트(Scite)는 1,800만 건 이상의 AI 리드(read)를 기록했으며, MCP 액세스는 해당 연도에 체결된 계약의 평균 계약 가치를 대략 두 배로 늘렸습니다.
• 경영진은 2027 회계연도에 플랫폼 구독 성장, 고객 유지율 개선, 추가적인 트랜잭션 안정화, 조정 EBITDA 증가 및 견조한 현금 창출을 예상했으나, 공식적인 매출 가이던스는 제시하지 않았습니다.
• 회사는 1,260만 달러의 현금을 보유하고 리볼빙 신용 차입금은 없으며, 2027 회계연도에 지급 예정인 사이트(Scite) 어닝아웃(earn-out) 지급금 3회가 남아있는 상태로 2026 회계연도를 마무리했습니다.

핵심 재무 실적

전년 동기(4분기) 순이익 결과에는 사이트(Scite)의 최종 어닝아웃 결정과 관련된 유리한 조정 항목이 포함되어 있었습니다. 경영진은 회계연도 전체 영업활동 현금흐름 감소의 주요 원인이 근본적인 수익력이나 매출채권 회수가능성보다는 운전자본 타이밍 문제 때문이라고 설명했습니다.

사업 및 영업 성과

리서치 솔루션즈(Research Solutions)는 4분기에 29개, 연간 105개의 플랫폼 도입 건수를 추가하며 2026 회계연도를 총 1,276개의 플랫폼 도입 건수로 마감했습니다. 업셀링 및 대형 계약에 집중한 보다 체계적인 영업 프로세스에 힘입어 평균 판매 가격은 4.7% 상승했습니다.

플랫폼 매출은 전체 매출의 43% 이상을 차지하며 전사적인 이익률 확대를 뒷받침했습니다. 4분기 플랫폼 총이익률은 87.3%로 경영진의 목표 범위인 80%대 중반에 부합했습니다. 트랜잭션 총이익률은 우호적인 출판사 구성 및 저작권 콘텐츠 가격 정책을 반영하여 24.1%에서 26.0%로 개선되었습니다.

연말 기준 총 ARR은 2,250만 달러로, 약 1,620만 달러의 B2B ARR과 630만 달러의 정상화된 사이트(Scite) B2C ARR로 구성되었습니다. 2026 회계연도 동안 B2B ARR은 200만 달러(14.1%) 증가한 반면, 정상화된 B2C ARR은 약 38만 달러 감소했습니다.

연간 트랜잭션 매출은 소수 대형 고객의 물량 감소 및 플랫폼 구독에 포함된 트랜잭션 혜택의 영향으로 8.7% 감소했습니다. 다만 4분기 감소율은 3분기의 11%에서 6.7%로 완화되었으며, 경영진은 이를 안정화 추세에 부합한다고 설명했습니다.

AI 관련 ARR은 전분기 대비 125% 증가하며 80만 달러에 달했으며, 1년 전 거의 0에 가까웠던 수준에서 크게 늘었습니다. 사이트(Scite) 및 아티클 갤럭시(Article Galaxy) 커넥터는 챗GPT(ChatGPT), 클로드(Claude), 코파일럿(Copilot)을 통해 이용할 수 있습니다. 경영진에 따르면 MCP 이용량의 약 4분의 3은 유료 요금제에서 발생하며, MCP를 도입한 고객은 현저히 높은 고객 유지율을 나타냈습니다.

리서치 솔루션즈는 출판사용 MCP 게이트웨이(MCP Gateway)도 출시했습니다. 약 40개 출판사가 인덱싱되었으며, 첫 두 건의 게이트웨이 계약이 체결되었습니다. 경영진은 해당 제품이 아직 상업화 초기 단계에 있다고 밝혔습니다.

경영진 전망

리서치 솔루션즈는 공식적인 2027 회계연도 가이던스를 제시하지 않았습니다. 그럼에도 경영진은 다음과 같은 전망을 제시했습니다.

• 확대된 영업 조직과 신규 AI 제품에 힘입은 지속적인 플랫폼 구독 성장 및 견조한 기업/학술 B2B 매출.
• 고객 상태 모니터링, 팀 구조 개편 및 MCP 도입을 통한 참여도 개선으로 갱신율 및 업셀링 향상.
• 경쟁 심화 및 가격에 민감한 학술 고객으로 인해 B2C 사업은 전반적으로 보합세 유지.
• 트랜잭션 및 문서 전달 매출의 전년 대비 한 자릿수 낮은 비율(low-single-digit) 감소.
• 절제된 비용 관리를 바탕으로 한 또 한 번의 조정 EBITDA 성장 및 견조한 현금 창출.

경영진은 AI 제품 파이프라인이 이전에 언급된 100만 달러 수준을 훨씬 상회한다고 말했으나 정확한 수치는 제시하지 않았습니다. 회사는 인수, 자사주 매입 및 기타 잠재적 현금 활용 방안을 검토 중이지만 단기적으로 예정된 인수는 없다고 밝혔습니다.

리스크 및 주요 관전 포인트

• 연구 활동이 AI 인터페이스로 이동함에 따라 기존의 논문 단건(per-article) 트랜잭션 물량은 계속해서 압박을 받고 있습니다.
• 개인 학술 연구자들이 비용에 민감하고 경쟁 제품이 늘어남에 따라 B2C 성장은 여전히 어려운 상황입니다.
• 4분기 실적 개선과 최근의 조직 개편에도 불구하고 갱신 및 업셀링 실행에는 여전히 주의가 필요합니다.
• 출판사 게이트웨이의 상업화는 아직 초기 단계이며, 경영진은 AI 라이선싱 모델이 여전히 발전하는 과정에 있다고 설명했습니다.
• 전체 활성 고객 수는 1년 전 1,338명에서 1,323명으로 감소했습니다.

애널리스트 Q&A 주요 내용

경영진은 80만 달러의 AI 관련 ARR에 신규 매출과 업셀링 매출이 모두 포함되어 있다고 설명했습니다. 회사는 향후 AI 관련 매출을 공개할 계획이지만 신규 고객(new-logo) 기여분과 기존 고객 확장 매출을 구분하여 공개할지 여부는 확정하지 않았습니다.

MCP 가격 책정은 일반적으로 이용 한도 및 한도 초과 시 추가 요금이 부과되는 시트(seat) 기반 또는 엔터프라이즈 계약을 활용합니다. 현재 사이트 MCP(Scite MCP) 및 아티클 갤럭시 MCP(Article Galaxy MCP)는 해당 핵심 제품의 라이선스를 필요로 하지만, 리서치 솔루션즈는 대체 모델을 테스트 중입니다.

사이트(Scite) 이용은 회사의 어시스턴트 및 검색 인터페이스에서 MCP로 이동하고 있으며, AI 리드 건수가 가속화되고 있습니다. 아티클 갤럭시 MCP 도입은 보다 초기 단계에 있으며, 경영진은 플랫폼과 MCP 채널 간 논문 구매 및 이용 간의 관계를 모니터링하고 있습니다.

실적 발표 컨퍼런스 콜 전문

전체 실적 발표 컨퍼런스 콜 녹취록

경영진 발표

Operator

Good afternoon, everyone, and thank you for participating in today's conference call to discuss Research Solutions financial and operating results for its fiscal fourth quarter and full year ended June 30, 2026. As a reminder, this conference is being recorded.

I would like to now turn the conference over to your host, John Beisler, Investor Relations.

John Beisler

Thank you, operator. Good afternoon, everyone, and welcome to the Research Solutions Fourth Quarter and Full Fiscal Year 2026 Earnings Call. On the call today are Roy W. Olivier, Chairman and Chief Executive Officer; Dave Kutil, Chief Financial Officer; and Josh Nicholson, Chief Strategy Officer.

After the market closed this afternoon, the company issued a press release announcing its results for the fourth quarter of full year fiscal 2026. The release is available on the company's website at researchsolutions.com.

Before management begins their prepared remarks, I would like to remind you that some of the statements made today will be forward-looking and are made under the Private Securities Litigation Reform Act of 1995. Actual results may differ materially from those expressed or implied due to a variety of factors. We refer you to Research Solutions' recent filings with the SEC for a more detailed discussion of the risks that could impact today's -- that could impact the company's future operating results and financial condition.

Also on today's call, management will reference certain non-GAAP financial measures, which we believe provide useful information for investors. A reconciliation of those measures to GAAP measures is included in the earnings press release issued this afternoon.

Finally, I would like to remind everyone that this call will be recorded and made available for replay via a link on the company's website.

I would now like to turn the call over to Roy W. Olivier. Roy?

Roy Olivier

Thank you, John. It was a busy year at Research Solutions in all respects. The corporate and academic sales teams did a nice job growing year-over-year while upgrading our sales process and sales team.

We turned over almost 50% of our sales team during the year and have expanded sales head count going into FY '27.

We continue to transition to a more structured sales process that has resulted in closing larger deals than we have in the past, raised our average sales price or ASP on both products and resulted in 105 net new deployments for the year.

While the new sales teams are doing well, and we feel very good about that continuing in FY '27, the upsell and renewal teams continue to need focus. We appointed a new leader to run that team, strengthen and expanded the team and realigned the structure around customer size and location.

We also implemented a tool to do a better job measuring customer health and kicking off automated and manual workflows based on those results. For example, we can kick off an engagement workflow to a specific cohort of users who have not used the software or used what we know are high-value features.

We started all this in early FY '26 and saw a nice improvement in renewal rates in Q4 of FY '26. We expect this more to positively impact net ARR growth as we go into and through FY '27.

We did report a 14% B2B ARR growth during the year, raising our high gross margin platform revenue from 39% of total revenue last year to over 43% this year, which translated into another year of positive results in terms of operating income, net income, EBITDA and cash flow.

In addition to increased spend in sales, we made some additional investments in product development and software engineering, which helped us release 2 new key AI products in addition to increasing development velocity on our core Scite and Article Galaxy products.

In addition, we made several internal changes to improve productivity and output, including using AI to help us write and test code.

All these improvements drove a large increase in development productivity. In fact, in the third quarter of FY '25, we were doing about 50 software updates a month. In June of '26, we did 200, a 4x improvement. Most importantly, we released 2 AI products that extend the unique capability of Scite and Article Galaxy to be accessible in ChatGPT, Claude or Copilot. This is part of our intent to "be where the researchers are working," and those products have been well received by our customers. As noted in our press release, we did about $800,000 in AI-related bookings in Q4, and we have built a strong pipeline of interest in those products that we expect we'll close in FY '27.

I think much of what we did in FY '27 will set us up nicely to grow the business profitably in FY '27.

I'd like to pass the call over to Dave to walk you through the fiscal fourth quarter and full year FY '26 financial results in detail. And then I'll discuss our goals in FY '27 and wrap up with some comments and the outlook for the year ahead. Dave?

Dave Kutil

Thank you, Roy, and good afternoon, everyone. I'll start my comments with a recap of our fourth quarter, followed by a summary of fiscal 2026 results.

Total revenue for the fourth quarter of fiscal 2026 was $12.1 million compared to $12.4 million in the fourth quarter of fiscal 2025, as increased platform revenue was more than offset by a decrease in transaction revenue.

Our platform subscription revenue was $5.3 million compared to $5.2 million in the prior year quarter. The growth was driven by an increase in B2B platform ARR due to a mix of new logo generation and upsells and cross-sells into our existing customer base, partially offset by a decline in B2C ARR.

We added 29 net new platform deployments in the quarter, bringing total deployments to 1,276 at year-end. We ended the quarter with $22.5 million in annual recurring revenue, up 7.8% year-over-year, which breaks down as approximately $16.2 million in B2B ARR and approximately $6.3 million in normalized ARR associated with sites B2C subscribers.

B2B ARR grew $2 million or 14.1% versus the year ago period and included AI-related ARR of $800,000, which grew 125% sequentially from the third quarter of fiscal '26. Please see today's press release for how we define and use annual recurring revenue and other non-GAAP terms.

Transaction revenue for the fourth quarter was approximately $6.8 million compared to $7.3 million in the prior year quarter, a decline of approximately 6.7%. That result represents a meaningful improvement from the 11% year-over-year decline we reported in the third quarter and is consistent with the stabilization trend we discussed on last quarter's call.

Our total active customer count for the quarter was 1,323 compared to 1,338 in the same period a year ago.

Gross margin for the fourth quarter was 53%, a 200 basis point improvement over the fourth quarter of 2025 and a new quarterly record for the company. The increase was due to the ongoing revenue mix shift towards our higher-margin platforms business which now represents 43% of our total revenue.

The Platform business recorded gross margin of 87.3% compared to an all-time high of 88.5% in the prior year quarter, but still well within our target range of mid-80% gross margin.

Gross margin in our Transactions business was 26% compared to 24.1% in our prior year quarter. The increase was primarily attributable to improved margins on our copyright content, reflecting favorable publisher mix and pricing partially offset by lower service fee margins.

Total operating expenses in the quarter were $5.6 million compared to $5.1 million in the prior year quarter, as increased sales and marketing expenses and upfront investments in AI were partially offset by lower general and administrative costs compared to the fourth quarter of fiscal 2025.

Other expense for the quarter was $135,000 compared to income of $1.2 million in the prior year quarter. The prior year result was primarily attributable to a favorable adjustments, the final earn-out determination for sites in the fourth quarter of fiscal 2025.

As a reminder, as of August 2, we have completed 5 quarters of earn-out payments with 3 additional payments remaining in fiscal 2027.

Net income for the quarter was $666,000 or $0.02 per diluted share compared to $2.4 million or $0.09 per diluted share in the prior year quarter.

Adjusted EBITDA for the quarter was $1.4 million compared to $1.6 million for the fourth quarter of last year.

Now let me turn to the full year fiscal 2026 results. Total revenue for fiscal 2026 was approximately $48.3 million compared to $49.1 million in fiscal 2025. Platform subscription revenue increased roughly 10% to $20.8 million. Total deployments at year-end were 1,276 and net increase of 105 the deployments from the end of fiscal 2025, and our average sales price increased 4.7% due to the upselling efforts mentioned earlier.

From an ARR perspective, we added approximately $2 million of net B2B ARR during the fiscal year, while normalized B2C ARR declined by approximately $380,000 for the full year.

Transaction revenue for fiscal 2026 was $27.5 million compared to $30.1 million in the prior year, a decline of approximately 8.7% as transaction purchases for new customers were more than offset by lower volumes from a small number of large customers and the transactions-related benefits offered within our Platform subscriptions.

As I noted earlier, the year-over-year trend improved as we exited the fiscal year. Gross margin for fiscal 2026 was 51.9%, a 260 basis point improvement over fiscal '25. The increase is primarily related to our continued revenue shift towards our higher-margin Platform business and it drove a 3.6% increase in gross profit dollars to $25.1 million despite the lower revenue base.

Total operating expenses in fiscal 2026 were $21.5 million compared to $21.7 million in the prior year. Lower general and administrative and stock compensation expense were partially offset by higher sales and marketing expenses and product development costs.

Other expense for the year was $724,000 and $1.2 million in fiscal 2025. Last year included $1.7 million to reflect the adjustments made to the Scite earn-out finalization.

Net income for fiscal 2026 was $2.8 million or $0.08 per diluted share compared to $1.3 million or $0.04 per diluted share in the prior year.

Adjusted EBITDA for the year was $5.8 million compared to $5.3 million in fiscal 2025.

Turning to cash flow. Cash flow from operations for the fourth quarter was approximately $1.8 million compared to $2.3 million in the prior year quarter. For the full year, we generated approximately $5.3 million in cash flow from operations compared to approximately $7 million in fiscal 2025.

As we discussed on last quarter's call, the year-over-year decline primarily reflects the timing of working capital payments rather than a change in the underlying earnings power of the business or the collectibility of our receivables.

Turning to our balance sheet. Cash and cash equivalents as of June 30, 2026, were $12.6 million compared to $12.2 million on June 30, 2025. The end of the fiscal year with a higher cash balance than a year ago, even after funding 4 quarters of Scite earn-out payments, consistent with the expectations that we laid out on last year's fourth quarter call. There were no outstanding borrowings under our revolving line of credit. And with a growing cash position and no debt, our current balance sheet provides the flexibility to complete the remaining Scite earn-out payments in fiscal 2027, while remaining opportunistic in regard to strategic alternatives.

As we look forward to fiscal 2027, we expect continued Platform subscription growth, improving retention and further stabilization in our transactions business. Paired with disciplined expense management, we believe that we are also positioned to deliver another year of adjusted EBITDA growth and strong cash generation.

I'll now turn the call over to Josh to talk about our products. Josh?

Josh Nicholson

Yes. Thanks, Dave, and hello, everyone. I missed the last call as my wife and I welcomed our daughter to the world, really 1 of the best things ever, but I'm happy to be back with you guys today.

So today, I want to cover 3 things. What is AI actually doing to our business, what we shipped in response and why we think we are on the right side of the shift.

First, the impact of AI on the business, because I know it continues to be the question on everyone's mind and the question we get a lot. Our answer is that AI is already reshaping how research gets read, and we can see it directly in our own data.

Since we launched the Article Galaxy and Scite MCP connectors in February, AI agents have performed more than 16 million scholarly reads through Scite. Two details in that data matter for how you think about research solutions: One, in June, 4 months after launch; two, calls from AI agents overtook the number of queries in our own Scite assistant interface.

Usage didn't shrink, it moved. Researchers are still asking the same questions. They're increasingly asking them inside Claude, ChatGPT and Copilot instead of inside a browser tab and more corporate and academic customers are taking licenses with these tools.

Our job is to recognize our differentiators, leverage them and be where the users are and where we are. Scite and Article Galaxy connectors are in the official connector directories of all 3 of these platforms. Copilot, ChatGPT and Claude. Article Galaxy is the only DocDel tool with a connector that integrates directly into LLMs. Scite is 1 of the very few scientific article search connectors, and it adds capability that platforms themselves do not have and other search tools do not have either. I can tell you whether a claim has been supported or contradicted by later research because we have classified over 1.6 billion citations statements from the full text of the literature. It can also search inside Paywalled full text through our publisher agreement.

The second point roughly 43% of what AI agents reach for sits behind a Paywall, that is the unique value that AI and research solutions can deliver together. An agent finds a paper in Scite and about 4 times in 10, it cannot read the full text. Article Galaxy is a compliant rights cleared way to get it. 20 years of publisher relationships and right settlement do not get disrupted by a language model. It becomes the thing that the language model needs.

So the honest answer on AI is this. Legacy per article volume is under pressure, and Dave walked you through the transaction numbers, but the same force that pressures document delivery is creating demand for verified search, verification and rights cleared access, and that demand is landing on our recurring platform business.

Second, I want to talk about what we shipped. In fiscal 2026, we put 3 major things into production. The Gateway, MCP access to Scite and Article Galaxy with admin controls, purchasing and ordering and coverage that now extends beyond papers to patent grant clinical trials and drug and device data. The Meter, metered billing on agent usage, Pro and Teams plans on Scite and pooled usage across an organization. So that the agentic access is something we price for rather than give away. Today, roughly 3/4 of MCP usage comes from paid plans, not free ones. And then self-serve and API console launched in August, so a developer or a pharma data science team or an agent can get started ASAP.

Third, why is this working commercially? MCP is behaving as an expansion engine. MCP deals that we closed this year were almost entirely upsells to existing Scite customers, and on average, adding agentic access roughly doubled the contract. The first article Galaxy MCP deals landed as new logos, a large pipeline Roy continues to grow and the mix is shifting towards larger corporate deals. That shows up in deal size. The average value of new opportunities we create each quarter has nearly doubled over the past 2 years, with the sharpest step-up in the 2 quarters since the MCP launch.

It also shows up in retention. Customers who adopt MCP retain at far higher rates than those who do not. The part of the retention fix and the AI strategy are the same work, get customers using agentic access early and they stay.

The last piece is the supply side. We launched a publisher MCP Gateway that makes a publisher's content discoverable to AI agents while keeping Paywalled full text protected and that gives the publisher a clean read on what AI demand for their catalog looks like. Around 40 publishers are indexed and the first 2 Gateway agreements are signed. Every publisher that joins makes the corporate product more valuable, and every corporate customer makes the Gateway more valuable to publishers.

To close, the way research is distributed has shifted twice from print to online and now from online to AI, each time a new layer formed between the content and the reader, and that layer captured the growth. We built the layer this year. It's live, it's metered, and it's in the base of what we sell. Fiscal 2027 is about scaling it.

Back to you, Roy.

Roy Olivier

Thanks, Josh. I think Josh and Dave did a great job framing our results and what we've done to lay the groundwork for the future. I want to close covering a few items. First, let's talk about how I think about FY '27. It will continue to be a year of change in our industry as we continue to see mass adoption of AI. We have deep vertical market expertise and research, where being 100% correct is required. I think of us as filling the gap between what an LLM can do and what a research-intensive organization expects.

What is changing is how researchers start the research journey and what tools they use daily. What is not changing is that the business is driven by rights management and the costs to do research.

What we have always done, we will continue to do, produce tools that researchers can use where they are working in a copyright compliant and cost-efficient way, managing entitlements, company IP, rights, billing and reporting, along with access to almost 160 million journal articles from 1,800 publishers and societies is what is required in enterprise research. Scite's access to behind the Paywall information, the unique Scite badge that shows the quality of the article a researcher is looking at and all the supporting business intelligence to help publishers sell AI rights to researchers based on usage data to improve the LLM experience exponentially for both the enterprise and the publisher. When you add that to the fact that we have multiyear customer relationships with over 1,000 enterprises around the world, we believe we're well positioned to be part of the AI growth story going forward.

So what does that mean for FY '27 results? While we do not give guidance, I expect to see strong corporate and academic B2B sales in FY '27 based on our larger and more experienced sales teams and new AI solutions. We have seen and we expect to continue to see improved renewal and upsells based on a larger and more focused team, better tools monitoring leading indicators earlier in the ownership cycle and AI tools that when installed directly translate to higher renewal rates.

As noted in our press release and above, we've seen a large increase in AI-related product sales and have a strong pipeline going into FY '27. I do believe it will be a strong year in that regard.

I expect B2C to continue to be a challenging environment as increased competition chases the same individual researcher. I do think our unique value works here as well, but we are typically serving academic researchers in this segment that are very cost sensitive.

What we are losing to "try something else," we are gaining in higher retention, higher monthly payments using MCP with their LLM of choice. Our ASP lifetime value and renewal rates in this segment are going up. The challenge would be to drive any material growth that is acceptable at and acceptable customer acquisition costs.

In short, I think this is a flat business for the year. I do expect to see continued headwinds on transactions or DocDel sales resulting in a low single-digit year-over-year decline as it's a 5-year -- versus its 5-year CAGR of about 1% growth. It's an important business and will continue to be 1 in the new AI world. Our tools help customers find and acquire what they need in a cost-efficient and copyright compliant way. That's not going to change.

While I don't have a top line growth number to tell you, I can say that I expect the output of all of this to be improved EBITDA and cash flows in FY '27. We have almost $13 million in cash, no debt and are generating cash.

While we continue to look at acquisitions that will help us accelerate growth and add unique or strategic capability, we do not have anything lined up in the short term. We do recognize that our stock price is far below where we think it should be and are evaluating all options to increase shareholder value, including stock buybacks or other ways to use the cash to directly impact that.

With now -- with that, I will now turn the call back over to the operator for Q&A. Operator?

Operator

[Operator Instructions] We will take our first question from Jacob Stephan with Lake Street Capital Markets.

질의응답

Jacob Stephan

I appreciate you taking the questions and appreciate all the color as well. The AI-related ARR number, that's the first time you guys have broken that out. How much of that is genuinely incremental customer spend versus kind of like repricing and rebundling of existing subscriptions at renewal? And maybe if as a part B, you can kind of comment on attach rate with FY '26 renewable?

Roy Olivier

That's a net ARR growth number of AI-related products that could be an upsell or could be a new sale. I don't think we did that math and certainly haven't disclosed it. We will disclose an AI-related revenue number going forward, but I don't think we'll split it up across what's upsell and what's new. I guess we could, but I'd have to give that some thought. And that's -- I think we said this, but that that's up -- that's compared to near 0 a year ago. So the point is the MCP, the AI products are starting to generate traction, because that $800,000 number is out of a net ARR growth for the year of, I think, $1.8 million or $1.9 million. Dave, correct me if I'm wrong.

Dave Kutil

Correct. Yes.

Jacob Stephan

Okay. Got it. Maybe if you could comment on pricing a little bit. It seems like a lot of the industry is shifting towards kind of a usage-based model. How are you guys pricing this? Is this more of a per seat basis? Is there some usage component to it? Any color there would be helpful.

Roy Olivier

We do run some pricing models around usage, but typically, the price that's put in front of the customer is -- if it's not seats, it's an enterprise, but it has caps for usage and above that, they have to buy additional usage.

Jacob Stephan

Okay. So very similar to kind of how Anthropic and OpenAI are currently working as well.

Roy Olivier

Yes.

Jacob Stephan

Maybe just last 1 for me then. On Scite MCP and Article Galaxy MCP. So if a researcher gets your information or data inside of 1 of those 2, I guess, what keeps them from paying for either Scite or Article Galaxy? Is the MCP kind of a pull-through channel that you guys are using and driving transaction orders? Or is there some kind of intermediary in there?

Roy Olivier

Today, if you want to run MCP, you have to have the corresponding products. So you have to have an AG license to run MCP for AG. You have to have a Scite license to run MCP for Scite. However, we are experimenting with some new models where that won't necessarily be required.

On the AG side, it's really hard to separate those 2. On the Scite side, it would be easier to separate those 2, and we're looking at that.

Operator

[Operator Instructions] Our next question will come from Derek Greenberg with Maxim Group.

Derek Greenberg

Just continuing off the last question. I was wondering for the MCP product, you had mentioned that, that's primarily been an upsell from current customers. I was wondering how usage is tracking on MCP versus the traditional products? If you were to break those out separately, do you see a drop-off in traditional versus the MCP or how you view that dynamic?

Roy Olivier

Do you want to take that one, Josh?

Josh Nicholson

Yes. I discussed this a little bit, and I'll talk maybe first about the Scite MCP. So if you look at usage in assistant or search compared to MCP, really the users are moving over to MCP. And so assistant usage is declining and MCP is actually really skyrocketing. And so we have over 18 million AI reads. And if you look at the growth of that, which we publish on the Scite, you can see it ticking up even this last week, 1 million reads over that. And so we're seeing a lot, and this goes to the thesis going to where the users are. And so we're seeing better retention and more usage through MCP.

For Article Galaxy, we've been a bit more conservative on how we roll that out. We are discussing different ways of deploying that. We do see usage across the customer base, and that usage is interesting, but I would say it's still pretty early. And what we're tracking there is really looking at the article -- purchase to article use. How does that compare platform versus MCP. And in both of those cases, everything is basically plain language, so instead of clicking a button here or there, you're telling the tool to do this or to search that. And so I think it really facilitates the use of both core differentiators of the product in where users are starting to live including a lot of our own researchers and developers on the team. More and more of us are daily users of Claude, not just in software development, but really across the org. And we ourselves use MCPs from some of our tools such as our sales CRM and analytics and things like that.

Derek Greenberg

Okay. Got it. And I was wondering if you could talk a little bit more about the Gateway products and maybe the economics there on both the supply and the demand side and just how the traction look on that?

Josh Nicholson

Yes. On that, that's pretty early, but I think there's very large demand. I think what publishers are trying to figure out is how does their content exist in this world of AI. And so again, as I mentioned, we're trying to help bring content into AI just as hosting platforms brought content online. I think where we'll start to see some of that is that we provide subscription-based access to their subscribers. And so we get value to our product, which we're charging for, they get value to their subscribers and they get usage data for that. I think this is going to evolve over time. And I think we ultimately serve a variety of different kind of mechanisms and paths to serve the AI licensing. But I think it's still early, and there hasn't been much traction across anyone kind of in the space. But there is demand, and I think we are seeing good interest from some publishers in exploring this with us.

Derek Greenberg

Got it. And then just last 1 for me. I was wondering for the pipeline you mentioned it was growing nicely. I was wondering, last quarter, you provided us over $1 million on the AI products. I was wondering if there's any more detail you could provide in terms of what that pipeline may look like today?

Roy Olivier

It's well over that today, but I don't have an exact number.

Operator

I'm showing no additional questions at this time. I'd like to now turn the call back to Roy Olivier for any additional closing remarks.

Roy Olivier

Thank you, and thanks, everyone, for joining us on our call today. As a reminder, we'll be participating at the Lake Street Conference tomorrow in New York City. We look forward to speaking with you in November to discuss the first quarter FY '27 results. Have a great day.

Josh Nicholson

Thank you.

Operator

Thank you. This brings us into today's meeting. We appreciate your time and participation. You may now disconnect.