> **来源：[研报客](https://pc.yanbaoke.cn)** January 2026 # State of AI: Bi-Annual Snapshot The Execution Era of AI # Introduction We believe that building and operationalizing AI products is no longer just the new frontier of competitive advantage but rather becoming table stakes in the software world. In Q2 2025, we published “The AI Builder’s Playbook” to elevate the voices of the architects, engineers, and product leaders driving this work and emphasize what it takes to conceive, deliver, and scale AI-powered offerings end-to-end. Six months later, the picture is clearer. Over the last six months, we believe the AI market has entered a new phase of maturity. What started as the race to experiment with large models and launch early AI features has increasingly evolved into a challenge of scaling AI into durable, economically sound products. Given the speed of evolution in this market, this report is designed as a bi-annual update on how teams are building, deploying, monetizing, and using AI as adoption across the market matures. This report revisits core dimensions of the builder's playbook, highlighting the most important changes and developments over the last six months. Grounded in our proprietary Q2 2025 and Q4 2025 surveys of executives at software companies building AI products, alongside perspectives from our ICONIQ Community, the 2026 State of AI report seeks to offer a longitudinal operator perspective on what it takes to turn AI from a capability into a durable competitive advantage. In our view, the findings point to a clear conclusion: AI leadership in 2026 will be defined by disciplined execution across product, cost, trust, and go-to-market. Explore Our AI Perspectives # Data Sources # & Methodology<sup>1</sup> This study summarizes data from a Q4 2025 survey of $\sim 300$ executives $^{2}$ at software companies building AI products, including CEOs, Heads of Engineering, Heads of AI, Heads of Product, Chief Revenue Officers, and Chief Financial Officers. Throughout this report, we compare insights to our prior State of AI report, published in Q2 2025<sup>3</sup>, "The AI Builder's Playbook", where applicable. Where necessary, longitudinal data has been normalized to account for differences in firmographics to ensure trends are representative of the data. We also weave in insights and what we believe to be best practices from AI leaders from the ICONIQ community. All industry perspectives shared in this report have been anonymized to protect company-level information. # Respondent Firmographics By ARR or Revenue Range % of Respondents By ARR Growth Rate % of Respondents By Headquarters % of Respondents 1 - This data was collected anonymously by an external survey. Survey responses include some but not all ICONIQ Venture and Growth portfolio companies as well as companies not part of ICONIQ Venture and Growth's portfolio. 2 - Certain questions in the survey were optional or routed based on persona. Accordingly, some N-Size numbers in this presentation are less than 300. 3 - The Q2 2025 report summarizes data from an April 2025 survey of 300 executives at software companies, including CEOs, Heads of Engineering, Heads of AI, and Heads of Product. # From Models to Products: Where We See AI Differentiation Being Built We believe that AI product development has entered a phase of standardization and maturity. As the base models continue to improve, builders are no longer focused on creating foundational models but instead on delivering differentiated products at the application layer. Nearly $70\%$ of companies are building vertical AI applications, reinforcing that durable value is being created through domain-specific workflows rather than generalized intelligence. Consistent with this shift, $49\%$ of teams now cite application-layer innovation as their primary source of differentiation, compared to a much smaller cohort relying on proprietary model development. As model quality continues to improve across providers, our survey shows builders are increasingly adopting multi-model strategies to balance reliability, cost, latency, and customization. On average, companies now leverage $\sim 3.1$ model providers, up from $\sim 2.8$ six months ago, reflecting a growing emphasis on orchestration rather than allegiance to a single platform. However, despite increased investment in data pipelines and evaluation, most companies still report that their data foundations are only "mostly" or "somewhat" ready, particularly at enterprise scale, underscoring that data readiness remains a key execution bottleneck as AI products move from launch to scale. Application layer products continue to be the most common types of products being developed by AI builders, with almost $\sim 70\%$ of builders focused on vertical AI applications Q2 2025¹ Q4 2025² % of Respondents, Select All That Apply What is the primary AI product you are building? Source: Perspectives from the ICONIQ GenAI Surveys (April 2025 & December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network; $1 - \mathrm{N} = 300$ ; $2 - \mathrm{N} = 298$ As base models evolve and improve in efficacy, it appears application layer innovation is the primary differentiator for AI builders, competing on product UX, workflows, and integrations rather than proprietary model development Where does your team's primary differentiation come from today? % of Respondents, Single-Select, $N = 202$ Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network Application-focused builders most heavily rely on third-party model APIs, while proprietary model developers tend to leverage fine-tuned or customized models # Model Providers by Primary Differentiator % of Respondents, Select All That Apply, $N = 202$ Application Layer Innovation Balanced - differentiation on model and product innovation Proprietary Model Development Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network Top model selection criteria have remained consistent over the last 6 months, pushing builders toward multi-model strategies to manage trade-offs between model accuracy, cost, and customization Q2 2025 Q4 2025² Top Considerations When Choosing a Foundational Model for Customer-Facing Use Cases % of Respondents that Ranked in Top 3 # ICONIQ Community Perspective # Builders are Focusing on Model Stack Efficiency At ICONIQ's recent forum for enterprise Chief Data and AI Officers, leaders discussed their increasing focus on shifting to a cost-efficient model stack. Leaders have emphasized that frontier models are often unnecessary for most automation tasks and that open-source and fine-tuned SLMs deliver sufficient accuracy at lower cost. Additionally, routing strategies are emerging: the majority of tasks are pushed to smaller models, with only high-complexity cases escalated to improve cost management. OpenAI remains the most widely used model provider among survey respondents; however, builders are using a wider variety of models over time. Notably, Gemini has increased to the second most popular provider since our Q2 2025 survey Q2 2025 Q4 2025² Top Model Providers % of Respondents, Select All That Apply Source: Perspectives from the ICONIQ GenAI Surveys (April 2025 & December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network; $1 - \mathrm{N} = 184$ . $2 - \mathrm{N} = 194$ To measure performance of AI models, builders are adopting multiple evaluation methods; however, evaluation remains largely user feedback-driven and manual today, with only $52\%$ of builders adopting automated eval frameworks How do you evaluate the performance and reliability of your AI models? % of Respondents, Select All That Apply, $N = 198$ Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network Earlier-stage products tend to rely on manual controls to reduce hallucination risk, while scaled products tend to adopt more advanced and automated approaches # Hallucination Risk Mitigation Strategies by Product Stage % of Respondents, Select All That Apply, Top 3 Responses Only, $N = 202$ Pre-Launch Beta GA Scaling Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network Additionally, as AI products scale, teams tend to adopt reinforcement learning techniques (e.g., RLHF, RLAIF) in model training to improve performance and reduce hallucinations Is your company using reinforcement learning techniques to improve model performance or reduce hallucination risk? % of Respondents, Single-Select, $N = 194$ Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network Most companies rely on in-house data engineering teams to process and prepare data for AI models, with usage increasing over the past six months Q2 2025¹ Q4 2025² How do you process and prepare data for AI models? % of Respondents, Select all that apply Source: Perspectives from the ICONIQ GenAI Surveys (April 2025 & December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network; $1 - \mathrm{N} = 292$ ; $2 - \mathrm{N} = 197$ Despite increased investment in data preparation, few companies, especially $\$ 500\mathrm{M}+$ companies, believe they have fully ready data foundations to support accurate AI workflows How would you rate your data foundation's readiness to support accurate AI workflows? $\%$ of Respondents, Single-Select, $N = 198$ Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network From Models to Products: Where We See AI Differentiation Being Built Companies across revenue buckets are largely exploring agentic AI workflows for customer-facing use cases, with $\$ 500\mathrm{M}+$ companies leading in actively deployed AI agents Is your company exploring customer-facing agentic AI workflows? % of Respondents, Single-Select, $N = 278$ Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network While initially counterintuitive, we hypothesize that $\$ 500\mathrm{M}+$ companies are leading in agentic AI deployments likely because they have the operational maturity, workflow scale, and customer demand required to deploy agents safely in production. Agentic systems can introduce real execution and trust risk, which we believe later-stage companies are better equipped to manage through mature infrastructure, governance, and standardized workflows. These organizations also operate at a scale where repetitive, high- volume workflows can make agentic ROI easier to prove, and where enterprise customers are actively pulling vendors toward greater autonomy and automation. Finally, larger companies often have the brand credibility and customer relationships to survive early agent failures without stalling adoption, while smaller companies often remain in pilot mode. The survey data also showed that companies targeting vertical solutions and GTM use cases also lead in deployed agentic workflows, likely because of their clear use cases for deployment, repeatable workflows and easy-to-measure success metrics. Infra / developer customer-facing AI agents tend to have more permissions than other product groups, likely because they operate in controlled environments with more technical users and stronger safeguards Permissions granted for agentic products in customer-facing use cases % of Respondents with Customer-Facing Agents, Single-Select, $N = 115$ Product Target Use Case Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network # AI Economics Are Coming Into Focus AI is absorbing a growing share of product investment. Companies are allocating a larger portion of their R&D budgets to AI development in 2026, with high-growth companies spending $\sim 57\%$ of R&D on AI, compared to $\sim 38\%$ on average. We believe this shift reflects AI's central role in product roadmaps but also heightens scrutiny on cost structure and margins. As products scale, AI gross margins are improving, reaching a projected average gross margin of $\sim 52\%$ in 2026 on aggregate. Cost composition is also evolving: talent costs decline as a percentage of total spend over time, while model inference becomes the dominant cost driver at scale. These dynamics reinforce our view that long-term margin leadership depends on model selection, routing strategies, and infrastructure efficiency - not simply pricing power. Companies are allocating larger parts of their R&D budgets to AI development, signaling a key shift in product innovation towards AI products on the roadmap 2025 Budget<sup>1</sup> 2026 Budget² % Averages, Select All That Apply High-growth companies<sup>1</sup> are spending larger portions of their R&D budget on AI development (57% for high-growth companies vs 38% on average) % of R&D Budget Allocated to AI Development Source: Perspectives from the ICONIQ GenAI Surveys (April 2025 & December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network; $1 - \mathrm{N} = 166$ ; $2 - \mathrm{N} = 247$ 1-High-growth companies defined as companies that have $100\% +$ YoY ARR Growth As companies develop at scale, gross margins on AI products are projected to improve, underscoring the importance of cost management; companies that view balanced differentiation as their primary differentiator report the highest margins Average, By Year, $N = 269$ Gross Margin on AI Products (Aggregated) Primary Differentiator Application Layer Innovation Proprietary Model Development Balanced Differentiation Gross Margin on AI Products Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network # As products scale, talent costs to develop AI products trend down while model inference costs tend to increase % Average, $N = 202$ Breakdown of AI Product Costs Product Stage Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network # AI Is Forcing a Rethink of GTM, Pricing, and Proof of Value Survey results show go-to-market strategies for AI products are becoming more complex and diversified as AI reshapes both how products are sold and how value is proven. While sales-led motions remain the most common, nearly $60\%$ of companies now employ hybrid or product-led elements, reflecting the need to combine enterprise selling with hands-on product experience. Channel and partnerships are emerging as a meaningful growth lever, particularly with consulting firms, hyperscalers, and PE-backed platforms, contributing directly to pipeline generation and post-sale implementation. Monetization appears to remain in flux. While $58\%$ of companies still rely on a subscription or platform fee, usage-based $(35\%)$ and outcome-based $(18\%)$ pricing models have grown meaningfully in the last six months. Notably, $37\%$ of companies plan to change their AI pricing model in the next year, driven by customer demand for value-aligned pricing, competitive pressure, and margin concerns. Across interviews, hybrid pricing models (combining platform access with usage-based components and pricing safeguards) are emerging as the most pragmatic approach as customers and vendors converge on sustainable AI economics. AI Is Forcing a Rethink of GTM, Pricing, and Proof of Value Go-to-market strategies for AI builders are diversified across different motions, with sales-led motions leading among survey respondents but hybrid approaches gaining traction What is your primary go-to-market motion for AI products? $\%$ of Respondents, Single-Select, $N = 298$ Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network We believe channel and partnerships are emerging as a meaningful growth lever, even where channel is not yet the primary GTM motion. We've seen AI builders increasingly formalizing partner ecosystems, most commonly with consulting and PE firms, and hypercalers, to support topline growth. Partners can both add credibility to AI builders and contribute across multiple touchpoints in the customer journey, including deal sourcing and post-sale implementation. Several AI companies report channel/partner-sourced revenue accounting for a meaningful share of topline outcomes such as increased NNACV and new bookings, indicating that channel impact is growing over time. Partnerships are an incredibly efficient strategic lever for scalable growth. The earlier companies lay the foundation (ideally well before $25M ARR) the more likely they are to see channel revenue become a meaningful contributor down the line. Rob Bernshteyn, former CEO, Coupa Most AI builders utilize a subscription / platform component to their pricing models; however, consumption- and outcome-based pricing has grown in usage over the last 6 months AI Pricing Strategies % of Respondents, Select All That Apply, $N = 297$ "Outcomes" Tied to Pricing % of Respondents That Use Outcome-Based Pricing Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network Companies that use outcome- and consumption-based pricing for AI products most commonly use annual commitments and overages at tiered rates as pricing safeguards What pricing safeguards do you use? $\%$ of Respondents, Select all that apply, Consumption- and outcome-based pricing users only, $N = 137$ Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network AI monetization is still evolving, with many companies exploring consumption- and outcome-based pricing models to better align to AI business value Plans to Change AI Pricing in Next Twelve Months $\%$ of Respondents, Single-Select, $N = 298$ # Changes to Explore $\%$ of Respondents, Single-Select, Top 6 Responses Only, $N = 86$ Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network AI Is Forcing a Rethink of GTM, Pricing, and Proof of Value These pricing changes are primarily influenced by customer demand for pricing model changes and competitive pressures in the market What are the primary drivers for AI pricing model changes? % of Respondents that Ranked in Top 3, $N = 109$ Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network We believe another reason driving pricing model changes is the rise of agentic AI, primarily because agents are meant to execute tasks autonomously and their ROI is better aligned to consumption or outcomes, rather than licenses. However, in our view, pricing should also remain tied to total cost of ownership, not just list-price, to consider the cost of data, tokens, and infrastructure. Start hybrid: light subscription for platform access + usage for volume while outcomes are uncertain. Once outcomes stabilize, shift toward heavier subscription as it gives predictability and aligns with ARR growth. For example, one customer, averaged 1.6M monthly calls; their average call time was cut from $\sim 15$ minutes to $\sim 4 - 5$ minutes and customer satisfaction went up 3x. At that scale, outcome-based would have been more expensive, so [the customer] renegotiated to subscription-heavy. Head of GTM, Late-Stage AI-Native Company AI builders generally use proof-of-concept phases to drive adoption of their products; however, it is unclear who should bear the cost of the trial (customer-funded vs company-funded) Does your company use free trials or proof-of-concept (POC) phases when selling AI products? $\%$ of Respondents, Select All That Apply, $N = 297$ Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network As AI sales become more complex and POC-driven, companies are adjusting compensation structures to better support AI products, notably through adding new commissions and changing quotas Has the rise of AI in your product offering changed your compensation structures? How have your compensation structures changed with the rise of AI product offerings? Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network Additionally, the complexity of AI deployments is driving greater reliance on forward-deployed engineers (FDEs) as a critical part of go-to-market motions, generally used to bridge the gap between product and delivery functions # Which best describes the primary purpose of FDEs at your company? $\%$ of Respondents, Single-Select, $N = 171$ # For what percentage of customers does your company use forward deployed engineers? Median, For companies that utilize FDEs Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network # AI as a Force Multiplier Across the Organization Our survey indicates that internal AI adoption has moved beyond experimentation and is now delivering measurable productivity gains across functions. R&D teams continue to lead adoption, with high-growth companies reporting that $\sim 36 \%$ of code is now written with AI assistance, up from $29 \%$ six months prior. Use cases such as coding assistance, testing, documentation, and content generation show the highest reported relative productivity improvements, often exceeding $30 - 40 \%$ time savings. As adoption matures, survey respondents are increasingly measuring ROI through productivity gains, cost savings, and revenue uplift. Importantly, AI has not yet driven significant reductions in headcount; instead, it seems to be reshaping workforce composition. Companies are prioritizing AI-fluent talent while de-emphasizing administrative and repetitive roles. The data suggests that internal AI is becoming a force multiplier for existing teams, rather than a near-term lever for workforce reduction. R&D teams continue to lead internal AI adoption, which we believe reflects the tangible value of developer-centric use cases like coding assistance, testing, and code review Internal AI Active Adoption by Function Average % of Employees, by Function, $N = 298$ Top Use Cases for Internal AI Tools Ranked by Year, Top 5 Use Cases, $N = 201$ Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network 1 High-growth companies defined as companies that have $100\%+$ YoY ARR Growth However, integration with existing workflows and accuracy of AI models remain top challenges when adopting AI for internal use, in our view, highlighting the importance of model selection and change management to accelerate adoption Top Challenges When Adopting AI For Internal Productivity % of Respondents that Ranked in Top 3, $N = 298$ Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network Internal AI is increasingly funded through R&D budgets, with fewer companies relying on net new budget allocation as tools move from experimentation to deeper integration in workflows Q2 2025 Q4 2025² $\%$ of Respondents, Select all that apply, $N = 296$ Where is the budget for internal productivity coming from? Source: Perspectives from the ICONIQ GenAI Surveys (April 2025 & December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network; $1 - \mathrm{N} = 104$ ; $2 - \mathrm{N} = 247$ Annual spend on internal AI (as a percentage of revenue) is expected to increase in the next year and companies are split between building and buying AI tools for internal use cases Annual Spend on Internal AI (as a % of Revenue) $\%$ of Respondents, Single-Select, $N = 203$ Are there any internal use cases you are adamant about building internally? $\%$ of Respondents, Top 5 Responses Only, $N = 274$ Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network As internal AI adoption matures, companies are increasingly measuring business impact and ROI across multiple dimensions, most commonly through productivity gains and cost savings # How are you measuring the impact of AI on internal productivity? % of Respondents, Select all that apply Q2 2025¹ Q4 2025² According to our survey, on average, it takes $\sim 5$ months to ramp on a new AI tool. As companies get fully ramped on AI tools, they are also able to measure impact/ROI more easily. # ICONIQ Community Perspective # AI's Impact Still Focused on Bottomline Efficiency While we are seeing many use cases for AI tooling improve efficiency, we've seen fewer companies that are measuring the impact of AI on topline growth. At a recent ICONIQ forum for Chief Data and AI officers, the CDAO of a F500 consumer company noted that, "enterprises overweight using AI for efficiency, and underweight use cases around topline revenue growth. The reason for this being that 'Efficiency is the easy thing, and it's harder to measure the topline growth." Other enterprise CDAOs agreed with this sentiment, noting that AI adoption is driven more by cutting vendor & consultant spend than driving revenue. # Content generation and documentation use cases showcase the highest relative productivity gains for AI adopters Average relative increase in productivity for use cases where AI support is being deployed $\%$ Average, By Use Case, $N = 247$ R&D GTM G&A Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network # Spotlight: Internal AI Adoption in R&D Select Anecdotes from ICONIQ Portfolio <table><tr><td></td><td>Coding Assistance</td><td>QA and Testing</td><td>Code Review</td><td>Product & Design</td></tr><tr><td>Example use case</td><td>AI pair-programming, planning, and pull requests</td><td>Unit, integration, and UI test generation</td><td>Automated code review and logic validation on pull requests</td><td>Prototyping from requirements</td></tr><tr><td>How teams are executing on this use case</td><td>Engineers prompt AI tools with work tasks and the agent reads the repository, proposes a plan, estimates costs, and can open pull requests. Some teams are prompting the AI tools to write the prompt itself for higher quality results.</td><td>Engineers point AI tools at existing test files to scaffold additional test cases, generate factories, and write end-to-end specs. Teams are using AI to expand coverage quickly and standardize tests for common flows rather than writing tests from scratch.</td><td>Engineers are using AI tools to automatically review diffs, flag potential logic issues, and leave structured comments inline on the merge request. Engineers then jump directly into their IDE, keeping review tightly integrated with existing developer workflows.</td><td>PMs use AI to help draft requirement and then use AI-powered prototyping tools to quickly generate clickable prototypes. These prototypes are used to validate user flows, interactions, and assumptions with stakeholders.</td></tr><tr><td>ROI gained</td><td>Team reports that remote agents are tackling UI bugs and opening PRs; multiple engineers reported AI tools resolving merge conflicts and generating dependency graphs that followed internal patterns.</td><td>Engineers report significant time savings, with AI generating the majority of test code in minutes, materially reducing manual QA effort and accelerating release cycles.</td><td>Engineering teams report the agent surfacing issues that may have been missed in manual review, improving code quality without adding reviewer overhead.</td><td>Teams report that this process cuts early discovery from days to hours by enabling earlier feedback cycles, reducing downstream changes, and improving handoffs into development.</td></tr></table> Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network # Spotlight: Internal AI Adoption in GTM Select Anecdotes from ICONIQ Portfolio <table><tr><td></td><td>Customer Service</td><td>Marketing Automation</td><td>Sales Coaching & Enablement</td><td>Sales Engagement</td></tr><tr><td>Example use case</td><td>AI chatbots for customer inquiries</td><td>Marketing content creation & campaign development</td><td>Post-call coaching & follow-ups</td><td>Prospect identification and research</td></tr><tr><td>How teams are executing on this use case</td><td>AI tools powers generative chat support across consumer products. They pull from help center content and internal documentation to resolve high-volume issues, such as cancellations, refunds, billing, and product questions, before they reach agents.</td><td>Teams are uploading comprehensive messaging for all programs into AI tools to help draft marketing emails, brochures, and other collateral. AI tools generate drafts that match the brand voice and campaign objectives, which the team then refines.</td><td>After every sales call, teams use AI tools to parse transcripts, list action items, draft follow-up emails, and receive coaching notes.</td><td>Teams use AI tools to identify buying signals, so sales teams know who to target. The tools automatically create email messaging that is personalized to the customer's priorities and buying signals. It guides outreach by identifying which contacts to target using references from board minutes, news articles, strategic plans, and other data.</td></tr><tr><td>ROI gained</td><td>Customer service teams are seeing cost savings through material ticket deflection in repetitive categories and fewer conversations routed to live agents.</td><td>Initial marketing drafts can now be completed in one hour (vs two to three days), allowing teams to focus on more strategic tasks. Using AI to maintain brand consistency also helps generate cross-functional alignment, reduce review cycles.</td><td>Sales teams are reporting that this saves ~30 minutes per deal cycle, resulting in faster and more consistent follow-through in the sales process.</td><td>Teams are reporting higher success in getting meetings and winning opportunities, supporting meaningful topline / revenue growth.</td></tr></table> Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network # Spotlight: Internal AI Adoption in G&A Select Anecdotes from ICONIQ Portfolio <table><tr><td></td><td>FP&A Automation</td><td>HR & Recruiting</td><td>Legal</td><td>Data Analytics & BI</td></tr><tr><td>Example use case</td><td>KPI dashboard automation</td><td>Automated candidate selection and feedback</td><td>AI powered in-house legal assistant</td><td>Customer insights</td></tr><tr><td>How teams are executing on this use case</td><td>AI automates dashboard creation and generate contextual commentary explaining KPI trends and recommended actions. AI tools also cross-check invoices against contract records and flags anomalies or potential fraud.</td><td>AI helps screen applicants by bulk-downloading resumes from the ATS and analyzing them against job responsibilities. During candidate review, recruiters run a list of AI recommended candidates. During interview processes, AI tools are used to summarize feedback from multiple interviewers.</td><td>Teams are using legal AI agents as internal legal team support. These agents help review contracts, spot legal and business risks, and suggest practical changes to deals, while keeping a special focus on privacy, data security, and compliance obligations.</td><td>Teams connect Slack channels to AI tools to summarize sentiment and trend analysis across micro-surveys, without any manual data wrangling.</td></tr><tr><td>ROI gained</td><td>Anomalies are easily identified, enabling quick vendor corrections. Executive dashboard updates that previously required 4+ hours of manual analysis now auto-generate with narrative insights, freeing analysts for strategic work.</td><td>Teams have improved screening efficiency at scale and provide consistent, data-driven evaluations across candidate pools. Hiring teams are enabled to make faster, more data-driven decisions.</td><td>Legal teams can now handle increased volume in a variety of areas with the same headcount while maintaining thorough risk assessment and compliance standards.</td><td>Teams are seeing productivity gains in survey data analysis and faster sharing of key takeaways.</td></tr></table> Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network AI as a Force Multiplier Across the Organization Although companies are seeing AI's impact on productivity, most companies have seen little to no impact to headcount plans due to AI adoption in 2026 $\%$ of Respondents, Single-Select, $N = 298$ No significant impact to headcount plans Yes, slight decrease in headcount plans (e.g., due to AI-driven efficiency gains) Yes, overall increase in headcount plans (e.g., due to hiring for internal AI-related roles) Yes, significant decrease in headcount plans Other Has internal AI adoption impacted your headcount plans for 2026? Source: Perspectives from the ICONIQ GenAI Surveys (December 2025) and perspectives from the ICONIQ team and network of AI leaders consisting of our community of CIO/CDOs overseeing AI initiatives in enterprises, CTOs, our Technical Advisory Board, and others in our network 1-High-growth companies defined as companies that have $100\% +$ YoY ARR Growth While company headcount impacts vary across companies, survey respondents that are increasing their headcount plans are prioritizing AI professionals: developers, data scientists, prompt engineers, and those that have embraced AI into their personal workflows. Conversely, administrative, operational, and back-office G&A roles have been deprioritized. Additionally, some companies are also deprioritizing sales team hiring as they unlock productivity gains through AI adoption. High-growth companies<sup>1</sup> are also seeing more changes with their headcount plans and more likely to increase their headcount due to AI. While the fundamental structure of teams hasn't fully changed yet, how people work has already dramatically changed. AI is amplifying the extremes - I've seen top performers be at least 10x more productive and more novice employees using AI to upskill and accelerate outputs. I believe domain and technical expertise will become requirements for anyone building and traditional project management roles will begin to disappear. Leaders cannot effectively lead without knowing what's possible with AI. - Kipp Bodnar, Hubspot CMO # ICONIQ Where visionaries define the future of their industries # Disclosures Unless otherwise indicated, the views expressed in this presentation are those of ICONIQ ("ICONIQ" or the "Firm"), are the result of proprietary research, may be subjective, and may not be relied upon in making an investment decision. Information used in this presentation was obtained from numerous sources. Certain of these companies are portfolio companies of ICONIQ. ICONIQ does not make any representations or warranties as to the accuracy of the information obtained from these sources. This presentation is for educational purposes only and does not constitute investment advice or an offer to sell or a solicitation of an offer to buy any securities in connection with any investment fund or investment product that ICONIQ sponsors. Any such offer or solicitation will only be made pursuant to definitive offering documents and subscription agreements. Any reproduction or distribution of this presentation in whole or in part, or the disclosure of any of its contents, without the prior consent of ICONIQ, is prohibited. This presentation may contain forward-looking statements based on current plans, estimates and projections. The recipient of this presentation ("you") is cautioned that a number of important factors could cause actual results or outcomes to differ materially from those expressed in, or implied by, the forward-looking statements. The numbers, figures and case studies contained in this presentation have been included for purposes of illustration only, and no assurance can be given that the actual results of any ICONIQ portfolio company will correspond with the information contained in this presentation. No information is included herein with respect to conflicts of interest, which may be significant. The portfolio companies and other parties mentioned herein may reflect a selective list of the prior investments made by ICONIQ. Certain of the economic and market information contained herein may have been obtained from published sources and/or prepared by other parties. While such sources are believed to be reliable, none of ICONIQ or any of its affiliates and partners, employees and representatives assume any responsibility for the accuracy of such information. All of the information herein is presented as of the date made available to you (except as otherwise specified), and is subject to change without notice, and may not be current or may have changed (possibly materially) between the date made available to you and the date actually received or reviewed by you. ICONIQ assumes no obligation to update or otherwise revise any information, projections, forecasts or estimates contained in this presentation, including any revisions to reflect changes in economic or market conditions or other circumstances arising after the date the items were made available to you or to reflect the occurrence of unanticipated events. Numbers or amounts herein may increase or decrease as a result of currency fluctuations. For avoidance of doubt, ICONIQ is not acting as an adviser or fiduciary in any respect in connection with providing this presentation and no relationship shall arise between you and ICONIQ as a result of this presentation being made available to you. ICONIQ is a trading name of ICONIQ Partners (UK) LLP. ICONIQ Partners (UK) LLP (Registration Number: 973080) is an appointed representative of Kroll Securities Ltd. (Registration Number: 466588) which is authorised and regulated by the Financial Conduct Authority. ICONIQ Partners (UK) LLP is a limited liability partnership whose members are ICONIQ Capital (UK) Ltd, Seth Pierrepont and Lou Thorne, and it is registered in England and Wales and has its registered office at 27 Soho Square, London W1D 3QR. ICONIQ Partners (UK) LLP acts as an adviser to ICONIQ Capital LLC. These materials are provided for general information and discussion purposes only and may not be relied upon. This material may be distributed to, or directed at, only the following persons: (i) persons who have professional experience in matters relating to investments falling within article 19(5) of the Financial Services and Markets Act 2000 (Financial Promotion) Order 2005 (the "FP Order"), (ii) high-net-worth entities falling within Article 49(2) of the FP Order, and (iii) any other persons to whom it may otherwise lawfully be communicated (all such persons together being referred to as "FPO Relevant Persons"). Persons who are not FPO Relevant Persons must not act on or rely on this material or any of its contents. Any investment or investment activity to which this material relates is available only to FPO Relevant Persons and will be engaged in only with FPO Relevant Persons. Recipients must not distribute, publish, reproduce, or disclose this material, in whole or in part, to any other person. Copyright © 2026 ICONIQ Capital, LLC. All rights reserved. # A global portfolio of category-defining businesses <table><tr><td>IPassword</td><td>acuity</td><td>adyen</td><td>Age of Learning</td><td>airbnb</td><td>Airtable</td><td>aobaib</td><td>Alibaba</td><td>alteryx</td><td>Altruist</td><td>ANTHROP\C</td><td>APPRENTICE</td><td>APTTUS</td><td>articulate</td><td>AURORA</td></tr><tr><td>AUTOMATTIC</td><td>AXONIUS</td><td>bambooHR</td><td>Benchling</td><td>BetterUp</td><td>bill.com</td><td>BLACKLINE</td><td>braze</td><td>Calendly</td><td>Campaign Monitor</td><td>Canva</td><td>CaptivateIQ</td><td>causaly</td><td>chime</td><td>CLARA</td></tr><tr><td>coast</td><td>Collabra</td><td>conexiom</td><td>coupa</td><td>CROWDSTRIKE</td><td>cyberGRX</td><td>dbt</td><td>DX</td><td>databricks</td><td>DATADOG</td><td>dataiku</td><td>DeepL</td><td>DevotedHealth</td><td>dexcare</td><td>dialpad</td></tr><tr><td>DocuSign</td><td>DRATA</td><td>IIElevenLabs</td><td>enfusion</td><td>EPIC GAMES</td><td>EvolutionIQ</td><td>ez cater</td><td>FTX</td><td>fastly</td><td>fetch</td><td>Figma</td><td>Fireblocks</td><td>Fivetran</td><td>Flipkart</td><td>#floQast</td></tr><tr><td>FREEWILL</td><td>Gem</td><td>GitLab</td><td>glean</td><td>gofundme</td><td>GoodRx</td><td>GreenSky</td><td>Groww</td><td>Guild</td><td>HashiCorp</td><td>headspin</td><td>HEPTAGON*</td><td>highradius</td><td>HIGHSPOT</td><td>hightouch</td></tr><tr><td>Hippo</td><td>Honest</td><td>houzz</td><td>iex</td><td>invision</td><td>incidentIQ, INTERCOM</td><td>komodo'</td><td>Lead Bank</td><td>LEGORA</td><td>loom</td><td>Lucid</td><td>MARQETA</td><td>miro</td><td>MC MONTE CARLO</td><td></td></tr><tr><td>monzo</td><td>motorway</td><td>Moveworks</td><td>Nayya</td><td>netskope</td><td>NEVIS</td><td>ninjaOne</td><td>notable</td><td>Notion</td><td>omni</td><td>ORCA security</td><td>OURA</td><td>panther</td><td>people.ai</td><td>Pepper</td></tr><tr><td>PIGMENT</td><td>Pinecone</td><td>PLURALSIGHT</td><td>Pontera</td><td>Primer</td><td>PROCORE</td><td>QGenda</td><td>Quince</td><td>ramp ↓</td><td>recharge</td><td>REDVENTURES</td><td>Reify/HEALTH</td><td>Relativity</td><td>Reprise</td><td>Restaurant365</td></tr><tr><td>Rillet</td><td>Robinhood</td><td>SANITY</td><td>sendbird</td><td>ServiceTitan</td><td>shopmonkey</td><td>side</td><td>SIERRA</td><td>skuid</td><td>SMARTLING</td><td>snowflake</td><td>SPOTNANIA</td><td>sprinklr</td><td>SQUIRE</td><td>STATSIG</td></tr><tr><td>Swap</td><td>TENNR</td><td>TinyFish</td><td>TRUCKSTOP</td><td>turbanomic</td><td>twinhealth</td><td>Twistlock</td><td>Uber</td><td>unifyapps</td><td>Unit21</td><td>UNITE US</td><td>VIC.AI</td><td>virtru</td><td>WARBY PARKER</td><td></td></tr><tr><td>wayfair</td><td>Wealthsimple</td><td>Wolt</td><td>WRITER</td><td>zinier</td><td>zoom</td><td colspan="8"></td><td></td></tr></table> These companies represent the full list of companies that ICONIQ Venture and Growth has invested in since inception through ICONIQ Strategic Partners funds as of the date these materials were published (except those subject to confidentiality obligations or companies for which the issuer has not provided permission for ICONIQ to disclose publicly). Further, the list of companies may not reflect the most recent ICONIQ Venture and Growth investments. Trademarks are the property of their respective owners. None of the companies illustrated have endorsed or recommended the services of ICONIQ. # The force behind every founder