Fireside Chat with Crystal Brown, Circnova And Biolabs

与 CircNova 及 BioLabs 的 Crystal Brown 炉边对话 Crystal Brown 分享了她从汽车制造业跨界走向生物科技创业的历程，阐述了初次创业的失败如何为 CircNova 开发用于环状 RNA 发现的专有 Nova Engine 平台奠定基础。她剖析了在带领由 RNA 生物学家和纯 AI 工程师构成的认知多元化团队时所需的战略平衡，并强调坚持严谨的湿实验验证以避免浮夸炒作。此外，Brown 还详细介绍了 CircNova 如何通过挖掘外部学术数据集、抵制不切实际的新药研发时间线，并在科学探索与工程落地之间搭建客观的商业桥梁，从而拓展出三大支柱商业模式。 Crystal Brown 结合在汽车制造业积累的运营经验与此前初创企业的失败教训创立了 CircNova，致力于通过自主原生计算平台 Nova Engine 发现并设计新型非编码环状 RNA（circRNA）。 (0:04) 在团队搭建上，CircNova 特意将资深 RNA 生物学家与缺乏生物学背景的机器学习工程师进行配对，旨在消除传统学术假设与领域思维定式。 (3:06) Brown 坚决反对“AI 可在 9 个月内研制出药物”的行业炒作，指出牺牲验证流程将破坏安全性与商业价值，公司将合理的研发周期目标定为约 18 个月。 (7:26) 为提前实现商业化，CircNova 开拓了第三条业务线，专门挖掘与转化科研机构沉睡的临床及临床前数据集，同时借助休斯顿与剑桥的外部实验室网络推进规模化实验验证。 (9:12)

CircNova CEO Crystal Brown argues that translating circular RNA into viable therapeutics requires pairing proprietary computational discovery with strict wet-lab validation, warning against unrealistic tech-industry timelines while maintaining disciplined commercial execution. Hybrid Validation and Lean Wet-Lab Architecture (10:50) Informed by the failure of her prior venture, Great Lakes Biologics—which collapsed after prematurely expanding lab footprint and headcount before hitting critical milestones—Brown enforces strict operational discipline at CircNova. To prevent computational drug discovery from relying on unsubstantiated algorithms or "smoke and mirrors," the company verifies its proprietary Nova Engine predictions through both internal testing and partnered wet labs in Houston and Cambridge. Brown deliberately keeps internal wet-lab facilities lean, explicitly citing past industry scandals like Theranos as a warning against unverified claims. This hybrid structure allows the company to validate core biological mechanisms and computational predictions directly while utilizing external developer AI tools to maintain software velocity without excessive capital expenditure. Cognitive Diversity and the Operational Triad (3:06) CircNova pairs established RNA domain authorities, including co-founder Alex Wesselhoeft, with software engineers and machine learning practitioners lacking formal biological backgrounds. Brown considers this absence of biological training an asset, asserting that it enables engineers to tackle computational challenges without being constrained by legacy scientific assumptions, despite the operational friction cross-disciplinary teams often produce. To balance these dynamics, Brown applies a three-part organizational model: research scientists represent theoretical ambition ("dreamland"), software engineers evaluate technical feasibility, and business leadership provides non-emotional governance. From her non-technical vantage point, Brown uses this triad to allocate finite capital and keep scientific exploration aligned with viable commercial milestones. Tempered Discovery Timelines and AI Risk Management (7:26) Brown rejects assertions from non-biological tech executives who claim artificial intelligence can compress drug discovery down to nine months. While CircNova targets a condensed timeline of roughly 18 months, Brown insists biological therapeutics involve complex physiological variables that cannot be treated like rapid-cycle consumer software. In her view, accelerating timelines at the expense of safety, potency, and quality benchmarks destroys commercial value. Addressing potential computational errors and unintended clinical side effects, Brown views adverse events in AI-designed therapeutics as inevitable. Comparing the current state of computational biology to the Industrial Revolution, she contends that early failures and system calibrations are unavoidable prerequisites to establishing safe, automated drug design standards. Diversified Commercialization and Investor Alignment (9:12) After transitioning from an exploratory "throw things at the wall" phase across multiple undruggable diseases to focused target indications, CircNova broadened its revenue structure. In addition to developing proprietary drug assets and offering its discovery platform, the company established a third revenue stream by partnering with academic research institutions to clean, mine, and monetize dormant preclinical and clinical datasets through the Nova Engine. This commercial strategy requires navigating a sharp divide among venture capital backers. Brown observes that traditional life science investors demand rigorous clinical trial roadmaps and regulatory evidence over AI branding, whereas TechBio investors prioritize computational leverage and software scalability. Consequently, she stresses the necessity of selecting aligned investors who respect both algorithmic discovery and the empirical requirements of wet-lab validation. While CircNova seeks to establish an 18-month discovery paradigm for non-coding circular RNA, the company's long-term commercial success depends on whether computationally designed molecules can reliably demonstrate clinical safety and therapeutic parity against traditionally engineered drugs.

CircNova 首席执行官 Crystal Brown 阐述了如何将汽车制造业的运营纪律与原生人工智能技术相结合，开发针对非编码环状 RNA（circRNA）的药物发现平台 Nova Engine。她强调生物技术企业必须在追求算法加速的同时保持严格的湿实验验证，警惕脱离生物学规律的“九个月成药”炒作，并通过聚焦优势适应症与释放外部沉睡数据实现商业化落地。 跨界经验与复合团队构建 (0:04) Brown 介绍了其从克莱斯勒和通用汽车一级供应商长达 15 年的汽车制造与财务运营背景跨界进入生物技术的经历。她的首个创业项目 Great Lakes Biologics 曾致力于 mRNA 疫苗酶的生产，但因过早建立湿实验并过快扩张团队而在达成关键验证前倒闭。吸取该教训后，她与首席科学官 Joe D'Angelo 以及环状 RNA 机制权威专家 Alex Wesselhoeft 联合创立了 CircNova，依托自研平台 Nova Engine 专门从事非编码环状 RNA 的计算设计与靶点发现。 在团队搭建上，Brown 主张认知多样性，特意将顶尖 RNA 生物学家与不具备生物信息学背景的纯软件与机器学习工程师相匹配。她认为缺乏传统生物学训练的技术人员不会受到既有学术教条的束缚，能为计算建模带来全新视角。尽管跨学科磨合伴随摩擦和人员更迭，但她强调团队的核心筛选标准在于跨领域的敏捷协作能力，而非单一纸面技术资历。 原生计算底座与研发节奏审慎 (1:43) 针对人工智能制药领域的虚假宣传与概念包装，Brown 表示 CircNova 拒绝使用现成的开源算法或“烟雾弹”式的表面噱头，而是坚持构建完全自研的原生计算基础设施与专有算法黑箱。她指出，计算生物学当前面临的核心行业挑战在于跳出纯理论模型的范畴，证明由 AI 设计的分子在安全性、有效性和可制造性上能够达到甚至超过传统人工设计药物的水平。 针对部分科技领域人士关于“AI 能在九个月内交付药物”的乐观断言，Brown 明确表示反对。她指出治疗药物研发涉及复杂的生理学反应与人体机制，无法等同于消费品或软件的快速迭代周期；一味压缩周期而牺牲数据验证只会破坏产品的商业价值。CircNova 设定的研发目标是在保证严谨性的前提下，将新型疗法的发现周期压缩至约 18 个月。 适应症聚焦与三支柱商业模式 (5:05) 在公司成立的前两年半中，CircNova 调整了早期的研发布局。由于 AI 发现空间广阔，团队初期曾在未成药疾病领域采取了广泛尝试的探索路径；随着生物学验证和计算反馈数据的积累，公司转向聚焦特定目标适应症的精细化路线图，集中资源建立可重复的疗效验证，避免研发精力分散。 商业化方面，除专有发现平台和自研治疗管线外，CircNova 拓展出了第三项收入支柱：与主动寻求合作的学术机构及临床中心对接，对其闲置的临床前和临床数据集进行清洗与挖掘，以生成新的候选药物资产。面对融资环境，Brown 提示创始团队需厘清注重临床里程碑与监管审批的传统生命科学风投，与看重计算扩展性的 TechBio 投资机构之间的诉求差异，避免资方预期与业务实际脱节。 精益湿实验纪律与深科技治理平衡 (10:50) 鉴于过往生物技术行业的造假教训，Brown 坚持在公司内部保留核心湿实验验证能力，以第一手实验核实所有计算输出的生物学效果，确保科学基础真实可信。同时，为了避免重资产扩张陷阱，公司在内部维持小规模实验室，将大规模实验执行外包给休斯顿和剑桥等生物技术枢纽的合作实验室，并使用生态系统内的外部开发者工具以优化技术团队的开发效率。 Brown 将深科技企业的治理架构定义为科学、技术与商业的三角协同：科学提出理论极限的构想，工程技术负责评估实际技术落地的可行性，而商业管理者则以无偏见的客观立场整合两端，根据有限的资本储备作出资源调配，防止科研团队无节制地追逐所有理论假说。针对 AI 药物潜在的计算幻觉和临床副作用，Brown 类比工业革命的发展历程，认为新技术突破不可避免会伴随未知失误，关键在于通过实践试错与校准逐步建立安全标准。 Brown 将 CircNova 定位为环状 RNA 领域兼顾序列结构推导与从头发现的平台合作伙伴，致力于通过 Nova Engine 建立先发优势。然而，将研发周期压缩至 18 个月的行业目标能否持续规避临床副作用并经受住监管审批的检验，仍取决于后续实体管线在临床转化中的实质表现。

## Background 本内容由 Ai4 频道于 2025 年 8 月 28 日发布在 YouTube 平台上。视频记录了 BioLabs 区域总监 Adam Jenkins 与 CircNova 创始人兼首席执行官 Crystal Brown 的炉边对谈，页面简介指出，Crystal 分享了自己作为缺乏传统科研与技术背景的跨界创始人，如何通过清晰愿景、坚定信念与跨学科团队建设，成功打造基于 AI 的环状 RNA 前沿深科技企业 CircNova 并建立行业公信力。 ## TL;DR 在本次对谈中，Crystal Brown 回顾了自己从 15 年汽车制造管理跨界至生物科技，并在首个创业项目受挫后创立 CircNova 的历程。她深入阐述了 CircNova 如何通过专有的 Nova Engine 平台探索非编码环状 RNA，并警惕“AI 烟雾弹”，坚持将原生 AI 架构与底层湿实验验证紧密结合。此外，她还分享了跨学科复合团队的搭建策略、管线从广撒网到高度聚焦的演进、与学术机构合作开拓的数据变现新赛道，以及非技术创始人如何以客观理性的商业思维统筹科学与工程落地。 ## Atmosphere & style 整场对话务实、坦诚且充满商业洞察力。Crystal 以极其接地气、不加掩饰的叙述风格直面过去的创业失败与跨界偏见，既破除了外界对 AI 制药的浮躁神话，又对环状 RNA 的科学潜力展现出高度自信。主持人与嘉宾之间的交流流畅自然，现场氛围既有深科技前沿探索的严谨与理性，又洋溢着打破行业壁垒、敢于破局的创业韧性。 ## Key takeaways - [00:04](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=4s) Crystal Brown 将 15 年汽车制造供应链管理经验迁移至生物科技，吸取首次创业盲目扩员与过早自建湿实验的教训，创立了基于 Nova Engine 的环状 RNA 企业 CircNova。 - [01:43](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=103s) 坚决拒绝开源算法套壳与“烟雾弹”式的 AI 概念炒作，坚持构建原生底层系统，并主张用无可辩驳的真实试验数据跨越从计算预测到临床审批的鸿沟。 - [03:06](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=186s) 团队组建采取“顶尖生物学家 + 纯 AI/软件工程师”的跨界组合，利用非生物背景开发者的无偏见视角打破思维定式，重塑药物研发工作流。 - [04:58](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=298s) 管线策略从早期发散式的“广撒网”快速收敛为聚焦单一核心适应症，并在看重确定性资产的传统药研风投与看重平台扩展力的 TechBio 投资人之间寻求平衡。 - [07:20](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=440s) 新药研发受制于极其复杂的体内生物学规律，本质不同于快消品或互联网软件的极速迭代，绝不能因盲目追求速度而逾越科学边界与安全底线。 - [09:02](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=542s) 商业化思考应从科研早期即深度介入，CircNova 通过清洗与挖掘高校及科研机构沉淀的海量数据，拓展出平台和自研产品之外的第三条数据变现营收赛道。 - [10:50](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=650s) 坚持核心机理必须由自有湿实验室自主验证以规避虚假泡沫风险，常规扩展实验与“卖铲人”类辅助 AI 工具则积极外包协作以提高资金效率。 - [12:34](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=754s) 警惕在研发中让 AI 喧宾夺主，AI 只是放大效率的工具，底层的生物学与化学本质才是决定成败的根基。 - [14:22](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=862s) 构建跨学科决策三角：科学家负责探索假设，工程师负责工程化落地，非技术背景的商业操盘手则以无情感偏见的全局视角做出最具商业可行性的决策。 - [15:37](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=937s) 将 AI 变革比作工业革命，主张理性面对大模型幻觉与早期未知副作用，通过持续暴露问题与迭代修正推动行业前行。 - [17:54](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=1074s) 依托专有 NOVA 引擎打造连接序列与结构的 RNA 研发核心枢纽，力争在约 18 个月内实现切实的治疗性成果产出并确立行业先发优势。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=4s) 从汽车制造到生物科技：在初创失败中孕育出 CircNova Crystal Brown 将自己形容为生物医药行业中“最非传统、反常规的领导者之一”。在投身生物科技之前，她在汽车制造行业深耕了整整 15 年，先后在克莱斯勒（Chrysler）与通用汽车（GM）的 Tier 1 一级供应商处负责管理。一次偶然的契机，她通过财务与运营岗位切入生物科技领域，从此彻底扎根于此。她将这段经历视为跨行业通用技能（transferable skills）能够成功迁移的最佳证明。 谈及创办环状 RNA 公司 CircNova 的契机，Crystal 直言这家公司的诞生源于一次刻骨铭心的失败。她的第一个创业项目是 Great Lakes Biologics，主要业务是为 mRNA 疫苗研发生产酶制剂。作为初次创业的创始人，她交足了学费——过早建立了自己的湿实验室（wet lab），并在尚未验证阶段性里程碑（milestones）之前就过早扩充了团队编制，最终资金链断裂被迫关停公司。然而，正是不得不关闭首家公司的阵痛，为 CircNova 带来了转机。彼时，首席科学官 Joe D'Angelo 向她展示了环状 RNA（circular RNA）的突破性研究方向，并探讨如何通过人工智能推动该技术。两人一拍即合，决定创立 CircNova，借助公司专有的“Nova Engine”技术平台，专注于识别并自主设计非编码环状 RNA（non-coding circular RNAs）。 - **首段职业经历**：长达 15 年的传统汽车制造背景，服务于克莱斯勒与通用汽车的一级供应链。 - **首次创业挫折**：Great Lakes Biologics 因湿实验开局过早、团队扩张脱离里程碑节点而清盘。 - **技术突破口**：与 CSO Joe D'Angelo 联手，将核心锚定在基于自主 Nova Engine 平台的非编码环状 RNA 发现与设计上。 ```mermaid flowchart TD accTitle: CircNova 创立路径与核心定位 subgraph S1["首次创业经验教训"] A["Great Lakes Biologics<br/>(mRNA 疫苗酶制剂)"] --> B["失败反思:<br/>实验室过早落地 / 盲目扩员"] end subgraph S2["CircNova 诞生契机"] B --> C["CSO Joe D'Angelo 提出<br/>环状 RNA (circRNA) 方向"] C --> D["引入原生 AI 基础设施<br/>研发 Nova Engine"] end subgraph S3["核心业务产出"] D --> E["非编码环状 RNA<br/>识别与自研设计"] end ``` ### [01:43](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=103s) 拥抱原生 AI：摆脱“科技烟雾弹”与药物研发的严苛现实 当被问及 AI 是否从成立第一天起就写在商业计划书中时，Crystal 解释称，起初的核心确实完全聚焦于环状 RNA 的生物治疗本身，这是 Joe 负责的科研基石；但从商业与产业发展趋势来看，她坚信如果不结合 AI，现代药物发现就无法顺应时代的浪潮。然而她对概念炒作抱有高度警惕，明确指出引入 AI 绝不能是“障眼法与烟雾弹”（smoke and mirrors），也不能只是直接搬用开源算法套壳，必须构建 CircNova 自己的黑盒架构与原生底层系统。 在生物科技中涉足 AI，最严峻的考验在于抵御资本狂热褪去后的质疑。当前全行业都在审视：AI 设计的药物是否真能顺利推进临床管线、完成人体试验并最终获批上市。对作为二次创业者的 Crystal 来说，最大的挑战就是用无可辩驳的数据去证明，通过 AI 计算设计的治疗方案，能够达到甚至超越过往人类科学家传统设计的疗效与可信度。 - **AI 整合基调**：拒绝现成算法套壳与公关噱头，坚持构建公司自有的专有底层设施。 - **核心行业关卡**：AI 辅助生成的药物分子必须跑通真实研发管线并获取监管审批，跨越从虚拟预测到临床疗效的鸿沟。 ### [03:06](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=186s) 跨界团队的化学反应：顶级 RNA 专家与非生物 AI 人才的融合 在人员招募与团队搭建策略上，Crystal 沿用了自己在汽车工业巨头时期的方法论——全力打造极度多元化的复合团队。一方面，团队必须拥有该领域的执牛耳者，例如联合创始人 Alex Wesselhoeft，作为顶尖学者，他在环状 RNA 前沿新发现方面起到了关键的定海神针作用；另一方面，团队积极引入在纯粹 AI 领域表现优异、但对生物学与生物信息学（bioinformatics）毫无涉猎的人才。 Crystal 认为，这种跨学科的“空白”非但不是劣势，反而是一种宝贵的资产。没有传统生物学思维定式的纯 AI 开发者，能从全新的视角观察问题，避免固步自封。而让“懂 AI 不懂生物”与“精通生物”的两类头脑在同一套工作流中高频碰撞，正是催生突破性产品与工艺的源泉。在这个磨合过程中团队也交过学费，曾有员工因无法适应交叉学科的交汇而被淘汰。Crystal 强调，这并不代表他们不是优秀人才，而是在生物与科技深度融合的界面上，仅凭简历上的光鲜技能远远不够，员工必须具备理解对侧领域并推动两者产生化学反应的“创造性思维”（creative mindset）。 - **团队金字塔尖**：包括 Alex Wesselhoeft 在内的顶级环状 RNA 科学家把控科学底线。 - **跨界人才策略**：大胆吸纳缺乏生物背景的纯技术/AI 开发者，利用其无偏见的新视角重塑工作流。 - **人才筛选标准**：硬核技能只是敲门砖，是否具备跨学科协同的思维弹性决定了人员留存。 ### [04:58](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=298s) 从“漫天撒网”到精准聚焦，以及应对两极化的投资人群体 创业两年半以来，CircNova 走过了从早期探索到体系成熟的转折期。Crystal 回顾道，初创阶段 AI 带来的广阔可能性很容易让人迷失方向，面对海量不可成药疾病（undruggable diseases）构成的“蓝海”，团队最初的策略宛如“把所有东西往墙上扔，看哪个能粘住”（throw things at the wall）。而随着两年半的研发数据逐步回流，团队迅速摒弃了发散式的探索，转而采取高度聚焦的管线策略，集中资源攻克单一特定适应症领域，待彻底跑通后再以此为跳板横向拓宽。 在面对资本市场时，AI 标签往往带来两极化的反馈。传统的深科技与生命科学投资人（deep life science investors）风格极其保守，对 AI 概念毫无波澜，他们只在乎经典的药物研发路径：管线进度如何推进、临床审批怎么解决，倾向于把鸡蛋放在经过时间验证的同一个篮子里；相反，TechBio 领域的投资人则渴望看到平台扩展的天花板与爆发力。作为兼具底层技术平台与自研药物管线的公司，CircNova 必须在两端寻找平衡。Crystal 警示创业者切不可饥不择食：“你绝不能跟错误的投资人结婚（You don't want to marry the wrong investor）”，必须找到真正懂得生物与技术交汇逻辑的长期伙伴。 - **管线战略演进**：从成立初期广撒网式的蓝海试错，转变为基于早期回流数据锁定核心适应症的聚焦打法。 - **传统药研风投的审视点**：看重临床落地、确定性管线审批与单一资产的确定性。 - **TechBio 投资人的审视点**：看重平台化交付能力、技术演进速度与跨资产复制潜力。 ### [07:20](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=440s) 跨界傲慢与生物学规律：为什么做药不是做快消品 当复盘初次在生物科技领域推介 AI 的经历时，Crystal 感慨于传统生物学界对新技术的本能抵触与潜在敌意。部分从业者对算法加速实验进程抱有警戒心态。更严峻的认知偏差则来自外部行业高管跨界带来的傲慢。 Crystal 曾与一位来自知名巨头机构的高管深入交流，对方由于缺乏生命科学常识，自信地断言：“AI 怎么可能搞不定这个？你们完全应该在 9 个月内做出一款新药。”对此，Crystal 明确指出，生物科技与快消品（CPG）等快节奏迭代的互联网或硬件行业有着本质区别。新药研发受到错综复杂的体内生物学变量与自然规律约束，决定了其必须具备严谨科学的节奏感。AI 能够大幅提高研发效率，但绝不能盲目追求速度而越过药物研发应有的科学边界，更不能牺牲安全与疗效的底线。 ### [09:02](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=542s) 商业化考量与数据变现新赛道 当被问及何时开始考虑商业化时，Crystal Brown 直言自己每天“从睁开眼睛到闭上眼睛睡觉”，无时无刻不在思考如何将产品商业化、如何切入市场打破壁垒，并确立 CircNova 在 RNA 疗法开发领域的先行者地位。她认为，无论产品在早期有多好的科学设想，如果达不到标准导致无法商业化，那就毫无意义。商业化思考不应该仅仅从初创公司阶段才开始，更应当直接从大学科研阶段就深度介入。目前高校和科研机构中沉淀了大量极具价值的 RNA 临床及临床前研究数据，亟需更高效的路径将其与产业结合。 这种产业与学术的结合，也为 CircNova 开辟了意料之外的商业化通道。公司创立初期，业务规划非常简单，仅包含“平台”（Platform）和“产品”（Product）两条线。但随着各大科研机构主动带着海量临床前与临床数据前来寻求合作，公司开拓出了第三条全新的收入赛道：通过清洗并挖掘这些庞大的数据集，获取更深度的信息，从而识别并生成全新的候选药物。 ```mermaid flowchart LR accTitle: CircNova 商业化演进路径 subgraph G1["早期商业规划"] A["技术平台 (Platform)"] B["药物产品 (Product)"] end subgraph G2["新增数据赛道"] C["学术与科研机构数据"] --> D["数据清洗与挖掘"] D --> E["发现全新候选药物"] E --> F["新增营收渠道"] end ``` ### [10:50](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=650s) 研发自主性、外包策略与 AI 生态协作 在谈到自建（In-house）与外包（Outsourcing）的平衡时，Crystal 明确表示核心技术的验证必须牢牢抓在自己手中。她半开玩笑道“可不想成为某些人——比如某个 Holmes（暗示 Theranos 的 Elizabeth Holmes）”，因此必须拥有自己的湿实验（Wet lab）来亲自验证底层技术和生物学机制。但出于规模化扩张和可持续发展的考量，常规扩展实验则会外包给休斯顿或剑桥等地的合作伙伴。 关于 AI 领域的外部合作，Crystal 透露每天都会收到大量希望提供 AI 助力的合作请求。她对此持非常开放和积极的态度，并以近期接触的“卖铲人”（Picks and shovels）类 AI 工具公司为例，说明这类工具能够极大优化数据整合与技术表现，大幅缩短摸索效率的时间成本。这种开发敏捷性也直接赋能管理层在团队搭建、成本控制和里程碑交付上做出更精准的商业决策。 ### [12:34](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=754s) 节奏把控：AI 狂热与生物学本质的平衡 关于生物科技领域决策速度的探讨，Crystal 指出这是一个充满矛盾的双刃剑。一方面，传统 Biotech 行业的推进节奏往往过于缓慢，在很多环节其实并未充分释放 AI 的潜力；但另一方面，如果过度狂热地部署 AI、将其作为绝对核心，从而忽视了底层的生物学（Biology）或化学（Chemistry）本质，就会陷入盲目依赖低质早期数据的巨大陷阱。因此，必须在拥抱 AI 高效工具的同时，在关乎商业战略的决策上保持深思熟虑的节奏。 - 警惕“AI 喧宾夺主”：AI 只是手段，底层的生物与化学逻辑才是决定成败的根基。 - 高级软件开发者（如团队核心 Robin）需要时刻跟踪竞品引擎与前沿技术，将先进方法反哺给自研的 NOVA 引擎。 ### [14:22](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=862s) 跨界决策三角：科学家、技术人员与商业操盘手 ```mermaid flowchart TD accTitle: 跨学科决策协同模型 subgraph Roles["跨学科团队分工"] S["科学家 (Scientists)"] -- "构想前沿假设 / 容易产生情感偏爱" --> D["梦工厂 (Dreamland)"] E["软件工程师 (Tech/Engineers)"] -- "评估可行性 / 工具工程化落地" --> P["实践落地 (Practice)"] B["非技术创始人/商业操盘手 (Business)"] -- "综合全局 / 无情感偏见决策" --> M["商业化闭环 (Commercial)"] end ``` 在讨论 AI 决策应当由工程师主导还是自上而下驱动时，Crystal 强调必须采取全方位的协同机制。她形象地将科学家比作“容易溺爱自己孩子的人”，科学家往往会为了某个研究方向倾注所有资源，而自己作为非科学家背景的创始人，恰恰具备“没有情感偏见”的超脱优势；软件工程师则负责将科学家的构想评估为可落地的技术实践；最后由商业管理层统筹所有拼图，决定最具商业可行性的推进路径。 - **科学家的角色**：探索“梦工厂”（Dreamland），提出前沿构想。 - **工程师的角色**：负责现实落地（Practice），验证技术可行性与工程化。 - **创始人的角色**：立足全局与商业回报，理性平衡资源，做出最终抉择。 ### [15:37](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=937s) AI 赋能的本质与对潜在风险的清醒认知 谈到此前是否有 AI 从业背景，Crystal 澄清自己过去曾接触过预测分析（Predictive Analytics），但并非深度的 AI 研发人员。在她看来，AI 就像一根“魔杖”，将科学家们默默耕耘多年的研究资产串联起来，展现出科学真正的威力，完成了过去几十年未曾想象的突破。 针对大模型幻觉与 AI 生成药物的未知副作用等顾虑，Crystal 表现得非常理性。她将 AI 浪潮比作“工业革命”：任何颠覆性技术的发展都会经历未知的并发症、副作用和令人沮丧的错误结果，唯有通过不断暴露出问题并修复它们，行业才能真正前行。 ### [17:22](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=1042s) CircNova 的未来愿景与 NOVA 引擎 面向未来，Crystal 充满信心地表示，CircNova 正在成为所有 RNA 疗法开发者的核心枢纽。无论合作伙伴是带着现有的序列数据来寻求结构层面的对应解析，还是没有任何基础、希望从零开始自下而上（Bottom-up）进行研发，CircNova 都能依托其强大的 NOVA 引擎，精准识别并生成全新的候选药物。 ### [17:54](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=1074s) 快速推进疗效输出与确立环状 RNA 药物研发的先发优势 Crystal Brown 强调了 CircNova 在药物研发周期上的惊人速度。公司致力于在极短的时间窗口内实现切实的治疗性产出（therapeutic output），整个周期预计仅需 18 个月左右。她指出，能够在如此紧凑的时间表内达成这一里程碑，充分证明了 CircNova 作为行业先驱者的硬实力，并将真正奠定公司在环状 RNA（circular RNA）药物研发领域的领导者地位。 对话尾声，主持人 Adam Jenkins 对 Crystal Brown 的精彩分享表示由衷感谢。他称赞了 CircNova 的愿景与进展，并向现场观众说明 Crystal 在对谈结束后仍会在场停留，欢迎有兴趣深入交流或有疑问的参会者留下来与她面对面探讨。Crystal 亦对主办方的邀请与大家的倾听表达了谢意，整场炉边对谈在热烈的氛围中圆满落幕。 - **快速转化周期**：CircNova 计划在约 18 个月内实现明确的治疗性成果产出。 - **行业定位**：通过 AI 驱动与技术创新，确立其在 RNA 创新药物发现领域的先发优势与领导地位。 ## Notable quotes > [00:04](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=4s) 最非传统、反常规的领导者之一 — Crystal Brown > [01:43](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=103s) 障眼法与烟雾弹 — Crystal Brown > [04:58](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=298s) 把所有东西往墙上扔，看哪个能粘住 — Crystal Brown > [04:58](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=298s) You don't want to marry the wrong investor（你绝不能跟错误的投资人结婚） — Crystal Brown > [07:20](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=440s) AI 怎么可能搞不定这个？你们完全应该在 9 个月内做出一款新药。 — 跨界高管 > [09:02](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=542s) 从睁开眼睛到闭上眼睛睡觉，无时无刻不在思考如何将产品商业化 — Crystal Brown > [10:50](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=650s) 可不想成为某些人——比如某个 Holmes — Crystal Brown > [12:34](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=754s) AI 只是手段，底层的生物与化学逻辑才是决定成败的根基。 — Crystal Brown > [14:22](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=862s) 科学家往往是容易溺爱自己孩子的人，而自己作为非科学家背景的创始人，恰恰具备没有情感偏见的超脱优势。 — Crystal Brown > [15:37](https://www.youtube.com/watch?v=w1WlqzrLBSQ&t=937s) AI 就像一根魔杖，将科学家们默默耕耘多年的研究资产串联起来，展现出科学真正的威力。 — Crystal Brown ## Mentioned Crystal Brown, Adam Jenkins, CircNova, BioLabs, Great Lakes Biologics, 克莱斯勒 (Chrysler), 通用汽车 (GM), Joe D'Angelo, Alex Wesselhoeft, Nova Engine, Elizabeth Holmes, Theranos, Robin, 环状 RNA (Circular RNA), mRNA ## Video description Adam Jenkins, Regional Site Director, BioLabs Crystal Brown, Founder & CEO, CircNova Abstracts What does it take to build a deep tech company when you're not a technologist? In this fireside chat, I’ll share how I founded CircNova, an AI-driven circular RNA company, without a traditional science or tech background—and why that was actually an advantage. I’ll explain how I built credibility, navigated imposter syndrome, and turned bold vision into real traction with world-class partners. The talk will explore how non-technical founders can lead innovation by focusing on clarity, conviction, and culture. Attendees will walk away with practical strategies to build, lead, and scale technical teams—even if they don’t write code or pipette RNA. If you've ever felt like the outsider in the room but knew you belonged at the table, this conversation is for you. [Source](https://www.youtube.com/watch?v=w1WlqzrLBSQ)

[0:04] Crystal Brown: 最非传统、反常规的领导者之一 [1:43] Crystal Brown: 障眼法与烟雾弹 [4:58] Crystal Brown: 把所有东西往墙上扔，看哪个能粘住 [4:58] Crystal Brown: You don't want to marry the wrong investor（你绝不能跟错误的投资人结婚） [7:20] 跨界高管: AI 怎么可能搞不定这个？你们完全应该在 9 个月内做出一款新药。 [9:02] Crystal Brown: 从睁开眼睛到闭上眼睛睡觉，无时无刻不在思考如何将产品商业化 [10:50] Crystal Brown: 可不想成为某些人——比如某个 Holmes [12:34] Crystal Brown: AI 只是手段，底层的生物与化学逻辑才是决定成败的根基。 [14:22] Crystal Brown: 科学家往往是容易溺爱自己孩子的人，而自己作为非科学家背景的创始人，恰恰具备没有情感偏见的超脱优势。 [15:37] Crystal Brown: AI 就像一根魔杖，将科学家们默默耕耘多年的研究资产串联起来，展现出科学真正的威力。

[0:04] Crystal Brown: fell into biotech by accident through finance and operations and never looked back [6:09] Crystal Brown: You don't want to marry the wrong investor [7:26] Tech Executive (quoted by Crystal Brown): There's no way that AI can't do this. You could be producing a drug in nine months. [9:12] Crystal Brown: From the time I open my eyes to the time I go to sleep, I am constantly thinking about how we commercialize our product. [10:50] Crystal Brown: I don't want to be certain people, hint hint Holmes [12:33] Crystal Brown: they love their babies [15:37] Crystal Brown: almost like that magic wand that's allowing us to do things that we dreamed of but didn't know could happen in our lifetime [16:37] Crystal Brown: Someone has an AI-designed drug and it does something horrific that we don't want. However, has that happened already? Absolutely. So, you have to expect it.

[0:04] Great. Well, Crystal, thank you for [0:05] being here. I [0:06] >> thanks for having me. [0:07] >> Um, so you have a very diverse [0:09] background in terms of where you come [0:11] from. Um, so I think for everyone who's [0:13] here, um, let's give them kind of a a [0:16] quick background run through of kind of [0:19] where you've come from um, and how you [0:21] got to this position. [0:22] >> Sure. So, I am one of the most [0:24] non-traditional, unconventional leaders [0:26] that you'll find in the biotech space. I [0:28] spent the first 15 years of my career in [0:30] automotive manufacturing working for [0:32] Chrysler and GM tier 1 suppliers. I fell [0:35] into biotech uh by accident through [0:38] finance and operations and never looked [0:40] back. So I am a true testament of [0:43] transferable skills and industries. [0:45] >> Um so in terms of Cirnova kind of coming [0:48] together um what did that background [0:50] look like? Was it um as you said you're [0:52] the you're the nontechnical founder of [0:54] it. Um did it come through research the [0:57] university? Did it come from uh kind of [0:59] industry? How did how did it come about? [1:01] >> So, it came through failure. Uh my first [1:03] startup was Great Lakes Biologics and we [1:05] were focused on doing uh producing [1:08] enzymes for mRNA vaccine development. It [1:11] was the first time I'd raised. I had had [1:13] some first founder failures, opened my [1:15] wet lab a little too early, staffed up [1:17] before I could hit milestones. Having to [1:20] shut that company down opened the [1:22] opportunity for Circ Nova. I started [1:24] getting a lot of research from my chief [1:26] scientific officer Joe D'Angelo pointing [1:28] us to our circular RNA and how we can [1:30] advance it using AI and so together we [1:34] came up with circa and really leveraging [1:36] the proprietary nova engine technology [1:38] to be able to identify and then design [1:41] our own non-coding circular RNAs. [1:43] >> So then was AI part of the um I'll call [1:47] it the playbook and the game plan from [1:49] day one? It was it was well day one it [1:51] started with just circular RNA and if we [1:53] were going to do therapeutic development [1:54] that was Joe's side my side of it is [1:57] well if we're going to do drug discovery [1:58] then we have to look at the way the [2:00] world is moving and there's nothing [2:01] that's advancing it more than AI right [2:03] but it's got to be our own native it [2:05] can't be smoking mirrors it can't just [2:06] be algorithms it really has to be our [2:09] own black box and infrastructure so when [2:12] we really looked at drug discovery [2:13] market and how AI was helping to [2:15] progress it was like a natural match [2:17] between the two RNA has so many [2:19] advancements and so does drug discovery [2:21] using AI. [2:23] >> What's been the hardest part of working [2:25] in the AI space in biotech? Has it been [2:28] the people? Has it been the technology? [2:30] What is, you know, from not a first- [2:33] time founder anymore, second time [2:34] founder? Um, what has that looked like [2:36] for you? [2:37] >> Yeah, the AI piece is hard, right? [2:39] Because because there's been like the AI [2:41] tech boom, there's a lot of questions on [2:43] is this real AI or again is it more [2:45] smoke and mirrors? And then we are still [2:48] looking at how AI design drugs are [2:50] actually advancing and getting into [2:52] pipeline and actually getting approved. [2:54] So I think the hardest obstacle or the [2:56] biggest challenge is making sure that we [2:58] can actually produce and are able to use [3:00] AI design therapeutics in the same way [3:03] we've been able obviously to use human [3:04] design. [3:06] One of the things that we talked about [3:08] in the last panel before you were here [3:09] is uh the intersection of um business, [3:13] biology, AI, the people that are [3:16] involved, the upskilling, how that [3:18] works. What has your approach been um as [3:21] a small company? Has it been to find [3:23] those pure play players who know [3:26] something really really well and and uh [3:29] kind of leverage their skills or has it [3:30] been to find jack of all trades that can [3:32] work across you know different parts of [3:34] the company? [3:35] >> Yeah. Yeah. So my recruitment strategy [3:37] is actually really similar to how it was [3:38] when I was in corporate and that means [3:40] we're building the most diverse team. [3:41] The way that I go about doing that is [3:43] making sure I have some of the most [3:44] brilliant experts in the business in [3:46] RNA. So if you look at one of my [3:47] co-founders, Alex Wesselhoff, he's been [3:50] just instrumental in what we've been [3:52] able to find for new discoveries in [3:53] circular RNA. So he's at the top of the, [3:55] you know, totem pole. Then we have [3:58] others on the team who are just [4:00] phenomenal when it comes to AI, but they [4:01] may not have the bioatics or the [4:03] bioinformatics or anything like that. [4:06] There's a gap there and there's a a [4:07] learning curve that can actually be of [4:09] value because they're looking at it from [4:11] a new lens and they're not pigeonholed [4:12] into their ideas and not open. So [4:16] there's value in having almost entry- [4:17] level people or people who know AI [4:20] really really well but they don't know [4:21] the biology and getting those two minds [4:22] to work together to come up with a [4:24] phenomenal product and process. [4:26] >> Any failures in that space with them [4:27] working together? [4:29] >> You know I think failures are lessons [4:31] right and as corny as that sounds I've [4:34] had some missteps. So had team members [4:36] had to let them go not because they're [4:38] not great people but becomes when you [4:40] look at the intersection of biology and [4:42] technology you have to really be able to [4:45] understand both and how to make them [4:46] cohesive and then that way move the [4:48] needle together. So the biggest part is [4:50] you may have all the skills on paper but [4:52] do you have the creative mindset to have [4:54] those two things intersect interact and [4:56] then progress? [4:58] >> That's that's something I think that [5:00] most people in many industries uh fail [5:02] fail to realize from from the get-go. um [5:05] more about the company. Um what has that [5:07] process looked like? How has it changed [5:10] from um since you said that you've had [5:12] kind of AI in that pipeline from from [5:14] day one, has it changed at all? Have you [5:17] had to as the technology has changed at [5:20] lightning speed, have you also had to um [5:23] reformat, you know, what the company's [5:25] game plan is or have you been able to [5:26] stay um relatively to plan? So, so we're [5:30] actually in the midst right now of that [5:32] nice little part where you're not quite [5:34] so early stage day one. You've kind of [5:37] flushed out some things in a process and [5:38] now we have more of I would say a more [5:40] focused approach. When we first started [5:42] the company, it's easy to get lost in [5:43] all the things we can do. That's what's [5:45] so exciting about AI. So, it's a blue [5:47] ocean water of opportunities for all [5:50] these undruggable diseases. So, which [5:52] ones do you go after? When we started, [5:53] it was let's throw things at the wall. [5:55] Let's see what makes sense. Now that [5:57] we're 2 and 1/2 years in, we've got some [5:58] data coming back. Now we know let's have [6:01] a more focused repo uh approach and [6:04] really concentrate on a specific [6:05] indication area that we can master and [6:07] then we can build on that. [6:09] >> From an AI perspective, how have [6:12] investors [6:14] um viewed that? Have they viewed that [6:16] positively? Is that just table stakes [6:18] now to get into the game? Like what is [6:20] that uh inclusion of those two letters [6:23] into the pitch deck, the conversation? [6:26] like what has that impact been overall? [6:28] >> Yeah, it's tricky, right? Because you [6:30] have your really conservative investors [6:31] that life science and they're real deep [6:33] life science people and they are not [6:34] swayed by the AI. They want to know how [6:36] you're going to get this therapeutic in [6:37] the pipeline, how you're going to get it [6:38] approved, put all your eggs in one [6:40] basket and you go that way because [6:41] that's the tried and true way. Then you [6:43] have your more like tech biofocused [6:46] investors and they are wanting to see [6:48] just how far you can come. So it's a [6:50] constant balance every day for me [6:51] depending on who I'm talking to because [6:53] the truth is we fit into both buckets. [6:55] That's what's wonderful about Surf Nova. [6:57] We are not your traditional life science [6:59] or biotech company. We're really able to [7:02] optimize our platform, but at the same [7:04] time, we've got a product that we're [7:05] developing as well. So, it's about [7:06] making sure that we talk to the right [7:08] investors that get that true [7:09] intersection we were talking about, [7:10] which can definitely be a challenge. You [7:12] don't want to marry the wrong investor. [7:14] So, you have to make sure they really [7:15] get and understand who we are as a [7:17] company, who I am as a CEO, and where [7:19] we're going. Is there anything you wish [7:21] that you knew um when you started [7:23] pitching kind of in the AI space in [7:25] biotech? [7:26] >> Oh man, for sure. I wish I knew um in [7:31] advance kind of how AI was going to [7:34] shape biotech. It's has such a huge [7:36] impact. I think what I didn't know [7:38] before was the resistance to AI and [7:40] almost like a resentment for how it's [7:43] accelerating certain things. I think [7:45] what we're working and what we're [7:46] looking towards now as an industry is [7:48] how to be more inclusive to be able to [7:50] accelerate but not getting caught up in [7:51] the rapid fire of what products can we [7:54] just you know shoot off. We're not CPG [7:57] we're not other lanes. So we have to [7:58] treat AI very differently in life [8:00] science and biotech. Do do you think [8:03] based off kind of what you said in terms [8:04] of smoke and mirrors and everything that [8:07] you've played with is a lot of that [8:09] coming from the tried andrue methods [8:12] that kind of the biotech and industry [8:14] have been doing for 20 years 30 years is [8:17] it that you know it's coming into just a [8:19] different realm of how we think of [8:21] biological sciences and how we kind of [8:23] incorporate them into the AI space like [8:25] where do you think those difficulties [8:26] come from? Yeah, I think the [8:28] difficulties come from you have a lot of [8:30] people who again are more traditional, [8:31] more conservative, but they're also [8:32] looking to integrate AI into their [8:34] platforms or into their uh therapeutics. [8:36] But then I see a lot of almost outside [8:39] industry people coming in and saying, [8:40] "Oh, that could be solved really [8:41] quickly." Right? So I I spoke with a guy [8:43] really really high up at a incredibly [8:46] well-known organization who didn't have [8:48] the bio background. He was just like, [8:50] "There's no way that AI can't do this. [8:52] You could be producing a drug in nine [8:53] months. There are variables when it [8:56] comes to therapeutic development that [8:57] allow us to be more paced. So, we can [8:59] move rapidly, but we can't move so [9:01] rapidly that we're compromising the [9:02] quality of the product that we're going [9:04] to have because at the end of the day, [9:06] if you have a product that doesn't meet [9:07] certain standards, you have nothing that [9:09] you can commercialize. So, it's it's a [9:11] balancing act. [9:12] >> How much do you think about [9:13] commercialization? Um, so early on, [9:17] >> every time I open my eyes, my director [9:20] of office is right here laughing. From [9:21] the time I open my eyes to the time I go [9:23] to sleep, I am constantly thinking about [9:24] how we commercialize our product, how [9:26] we're able to enter the market, break [9:28] barriers, and really establish ourselves [9:30] as a early mover in RNA therapeutic [9:32] development. [9:33] >> Do you think that um more companies [9:35] should think from about [9:36] commercialization earlier or do you [9:38] think that it should be more let the [9:40] research speak? [9:42] >> 100%. I think more companies should be [9:44] thinking about commercialization earlier [9:46] and not just companies. I think it [9:47] should start at the university level. [9:49] There's so much valuable research there, [9:51] especially as we're looking at the [9:52] advancements that RNA is making. There's [9:54] too much opportunity to commercialize a [9:56] lot of this research. I think that there [9:58] should be a more streamlined way to help [10:00] us get a lot of that data. You you see [10:02] that happening now. I've had multiple [10:03] research institutions reach out to us at [10:06] Circ Nova because they want to know what [10:07] can you do with all of this clinical and [10:10] preclinical research data that we have [10:12] and marry that with where you're going. [10:16] Is that opening up lanes that you didn't [10:17] think before of how Circova is working [10:20] in kind of um from a BD perspective? Is [10:24] that opening up lanes that just weren't? [10:26] >> Oh my gosh, 100%. We just had a call [10:27] this morning for the same reason. [10:29] Initially, when I started the company, I [10:31] was thinking it's two lanes, platform [10:33] and product, and that's it. What's [10:35] opened up now is an entirely new lane of [10:37] revenue generation. And that's from [10:39] having these large data sets, mining [10:41] them, making sure you're cleaning it up, [10:43] and then having that much more [10:44] information to be able to identify and [10:46] then produce novel drug candidates. [10:50] >> Um, [10:52] you'll know this as a as a founder. How [10:54] much of what you do is built inhouse [10:57] versus like how much do you spend on [11:00] vendors, platforms, things that you just [11:03] don't want to deal with? [11:05] >> Yeah. No, that's a great question. Um, [11:06] so to me it's very important that we [11:08] keep certain things in house. So when it [11:09] comes to validating our technology, you [11:12] know, I don't want to be certain p [11:13] certain people hint hint homes, right? [11:15] So we need to make sure that we're [11:16] keeping certain things within our own [11:18] wet lab and validating the technology [11:20] and biology for ourselves, but only on a [11:22] minimal level. We want to make sure [11:23] we're outsourcing as we're growing and [11:25] scaling is not sustainable. So a lot of [11:28] that type of work we're doing with [11:29] people in like Houston or we're looking [11:31] to do in Cambridge where we're splitting [11:33] our time. um really making sure that [11:35] we're partnering to make the right types [11:37] of choices as we're advancing. [11:40] >> How many times are you approached per [11:42] month and asked about um how people can [11:45] help you out on the AI front? [11:47] >> Oh my gosh, every day. Every single day [11:50] I get multiple messages and I don't get [11:53] tired of them. I think it's actually [11:54] like inspiring and it's letting us know [11:57] like where we're going as an industry. I [11:59] was at a dinner last night with another [12:01] startup who is kind of a picks and [12:03] shovels company when it comes to really [12:05] thinking about how you're integrating [12:06] data, looking at your technical [12:08] performance and then deciding which AI [12:10] tools that can use to optimize that. All [12:13] of that is cutting things down from us [12:15] having to figure out efficiencies and [12:17] productivity. Being able to in the snap [12:19] of a finger say this is what my software [12:21] developer has been able to do in this [12:23] amount of time helps me make more [12:25] decisions on how we're continuing to [12:26] build out the team from a cost [12:28] efficiency standpoint to how we're able [12:30] to hit milestones. How much are you in [12:34] based off kind of what you said and [12:36] again it goes back to since AI has been [12:38] kind of a core part of the engine in [12:39] terms of building the company how has [12:41] that enabled the company's thought [12:43] process as you envision you know going [12:46] through the R&D stages and going into [12:48] the clinic in terms of adopting further [12:52] AI usage in later stage assets or later [12:56] portions of the company's growth rate [12:57] does that make you more amicable to [12:59] those or does does not really kind of [13:02] play into your thought process. [13:04] >> Definitely more amicable. So my senior [13:06] software developer Robin who's here with [13:08] me and just phenomenal is always keeping [13:10] his finger on the pulse of what's new, [13:11] who's working on this technology, [13:13] existing technologies that are [13:14] comparable and contrasting to ours, what [13:17] can we take from it and use that [13:19] approach to apply to the Nova engine [13:20] technology to make sure that we are [13:22] becoming leaders in the field. So, I [13:24] think you have to keep your pulse on [13:26] things and I have a wonderful staff who [13:28] make sure they're doing that so we know [13:30] how other engines are advancing in [13:31] comparison to our own. [13:34] >> Do you think that um people move too [13:36] fast and make decisions too quickly in [13:38] this space? [13:40] >> You know, that's a trick question. Um [13:42] because in biotech, we like to move [13:44] really slow sometimes. So, sometimes no, [13:47] I don't think we're moving fast enough. [13:48] I think when you see um too much of AI [13:52] being deployed and that being the focus [13:54] and you're losing the focus from the [13:55] biology or the chemistry depending on [13:57] you know what which area you're working [13:59] on that's when too you run into the [14:01] troubles that are you waiting on the [14:03] right data not just oh we've got some [14:05] early data sets is it right can it help [14:07] guide the next decision so on one hand I [14:10] totally think we don't move fast enough [14:11] there's way more ways that we could [14:13] leverage AI in this industry on the [14:15] other's hand I think when it comes to [14:16] making decisions that are really going [14:18] to guide your business, you have to take [14:19] a more thoughtful paced approach [14:22] >> in terms of take the thoughtful [14:24] approach. I think you know we all like [14:26] to think that we're thoughtful in terms [14:27] of doing things but of course then we [14:29] make rash decisions when it comes to [14:30] making a decision. Um do you think that [14:34] you know decision- making processes [14:35] around the AI space? Um should it come [14:38] from the engineers who are going to be [14:42] performing the work or do you think it [14:44] should it come from a a top- down [14:45] business um impact? [14:48] >> It 100% has to be a collaborative [14:50] approach. So I have my scientists, [14:53] right? And I love my scientists, but [14:55] they love their babies. So they will [14:57] pursue everything and put all the [14:59] resources into that. We can't do that. [15:00] And because I'm not a scientist, I'm not [15:02] emotionally attached, right? On the [15:04] other hand, I need your input because I [15:06] need to know can we actually achieve the [15:08] results from an AI standpoint, from a [15:11] technical standpoint, the software [15:13] engineers really have to drive whether [15:15] this is practical. So I look at science [15:17] is dreamland. I look as tech as we can [15:19] actually put this into practice and make [15:21] it happen. Those are those sides. From [15:23] the business standpoint, I have to make [15:25] the decision now that I've got [15:26] everything. Now that I have all the [15:28] pieces, what is the best way for us to [15:31] actually pursue this? And it takes every [15:33] single piece to work together. [15:37] >> What portion of the AI project that [15:39] you've worked with? Well, I'll say how [15:41] much did you work in AI before um before [15:44] Circ Nova? [15:45] >> Not at all. [15:46] >> Is there something that you didn't [15:49] foresee it causing either on the [15:52] positive or in the negative side that [15:54] you're is that you wish more people [15:56] knew? [15:57] >> Yeah, I mean I was always really [15:59] interested in AI in in particular. Well, [16:01] that's not true. I did work in [16:02] predictive analytics for some time. So, [16:03] as I'm thinking about it, um, but I [16:05] think what's been most interesting and [16:07] just exciting to watch is how all those [16:09] pieces work together to be able to [16:11] optimize a lot of the work that these [16:13] scientists have been doing for years and [16:15] years and years and we just haven't been [16:16] able to get to so many different [16:18] options. So, I think um yeah, I think I [16:22] think AI is allowing us to see the true [16:24] power of science. It's almost like that [16:25] magic wand that's allowing us to do [16:27] things that we dreamed of but didn't [16:29] know could happen in our lifetime. [16:31] anything that you're worried about with [16:32] it, hallucinations, anything that you [16:34] think that could throw out, you know, [16:36] erroneous results? [16:37] >> Yeah, I mean, you know, listen, I'm I'm [16:39] a pretty kind of leveled person, so I [16:42] don't get but so worried. But of course, [16:44] when we think about things like side [16:45] effects, we don't know the side effects [16:47] yet of AI design therapeutics or AI [16:50] design drugs or products. So, are there [16:52] a myriad of different things that could [16:53] happen, consequences and complications? [16:55] Absolutely. But I look at it the same [16:56] way I would look at the industrial [16:57] revolution. You have to go through [16:59] certain things. You have to have [17:02] unexpected and undesired results so that [17:04] you're able to fix those things, work [17:06] out the kinks, and really things are [17:07] able to move. That's what I see AI being [17:09] able to do. So, what am I most concerned [17:12] about? Someone has an AI designed drug [17:14] and it does something horrific that we [17:16] don't want. Um, however, has that [17:18] happened already? Absolutely. So, you [17:20] have to expect it. [17:22] >> Um, so we're coming to the end. The last [17:23] thing I kind of want to ask you is like [17:25] what's next for Circ Nova in terms of um [17:28] you know you said you're you're kind of [17:29] at that inflection point where you're [17:31] not you're not new but you're not old at [17:33] this point. So like what's next for [17:34] Cirnova and kind of what are you excited [17:35] about? [17:36] >> Yeah, what I'm really excited about is [17:37] anybody who's developing an RNA [17:39] therapeutic should be coming to Circ [17:40] Nova to do that. Whether you've got some [17:42] sequential data that you want to work [17:43] with and we're drawing the structural [17:45] parallel, whether you're looking from [17:48] completely bottom up and you don't have [17:49] anything, we are now able to use the [17:52] NOVA engine again to identify new [17:54] therapeutic output and do that in a [17:56] really really quick timeline. We're [17:57] talking something like 18 months. So [17:59] being able to do that shows that we're [18:02] an early mover and we're really going to [18:03] be a leader in establishing drug [18:04] discovery for RNA. [18:06] >> Fantastic. Cool. Chris Brown, thank you [18:08] for coming. Thank you for spending the [18:09] time and talking to us about your [18:10] company. Um she will be around um so if [18:13] anyone wants to stay and talk to her or [18:14] ask questions feel free. Um and thank [18:17] you for joining us in this conversation. [18:19] >> Thanks for having me.