智能体式 AI 如何抬高律所门槛
法律行业正寻求将 AI 的应用扩展到聊天机器人之外,引入智能体来自动化律师的部分工作流程。尽管采用速度在迅速加快,但一些律所的推进速度慢于其他律所。The Information 总编辑 Laura Mandaro 与三位法律专家探讨了律所如何化解阻力、推动采用并衡量 AI 的回报。
智能体的优势
在许多律所,起草保密协议(NDA)是一项耗时的工作。AI 驱动的合同管理公司 Ironclad 的 Sunita Verma 表示,这一工作流程可以轻松外包给智能体。
“我们从客户那里反复听到的一个主题是,他们希望律师把更少的时间花在 NDA 上,把更多时间花在战略性交易上,”她说。对于已将 NDA 外包给 AI 的一些客户而言,节省下来的时间使他们能够把此前由外部法律服务供应商处理的工作收回内部完成。
但并非每家律所都拥抱了 AI。Isabel Parker 表示,尽管 White & Case 的律师们乐于使用 AI 处理文档摘要等基础工作,但大多数人仍对更广泛地部署智能体持抵触态度。
该律所确实使用智能体来改善客户体验,主要方式是通过针对特定客户的技能来捕捉偏好,从语气风格到哪些人需要被纳入合同和沟通之中。这些技能让机构知识更易于共享。
“过去,律师可能会说,‘我正准备把这个提交给 X 客户,他们希望看到的形式是什么样的来着?有人把这些写下来过吗?’然后邮件满天飞,每个人都急得火烧眉毛,”Parker 说。“而这让流程顺畅得多。”
让律师们接受
Jillian Burnett 承认,律师的怀疑态度可能成为采用 Legora 这类智能体工具的障碍——Legora 是面向法律工作的智能体操作系统。她表示,更大的障碍是 AI 疲劳,以及一种可以理解的不情愿:不愿牺牲时间去参加厂商培训、学习新工具。
“如果你刚发布了一项新功能,而他们才刚刚完成我们助手的培训,律师未必愿意再拿出更多计费时间,去学新东西,”她说。她强调,必须通过一对一的培训课程和小型焦点小组,让律师看到该平台及其智能体能带来什么,并指出,只要明显值得,律师是愿意付出时间的。
“等到 Legora 走进与客户的会议时,我们已经了解他们的工作流程,并能直接呈现与他们手头事务相关的用例,”她说。
另一个挑战是决定把哪些技能投入实际运营。
“当有很多人都在创建技能时,局面会变得非常嘈杂,”她说。减少这种噪音的一种方式,是组建由创新负责人、高级合伙人和其他决策者组成的知识团队或指导委员会,帮助制定标准,确定哪些技能应成为全所统一规范。随后,Legora 可以帮助用户通过标签(例如业务领域)找到技能,并让律所掌控这些技能的共享范围。
把成本因素考虑进去
随着厂商转向按用量计费,律所面临的成本变得远更难预测。
“对律所来说,核算技术支出完全是件新事。它完全不可预测,而我们还没有建立起相应的机制,”Parker 说。
Legora 利用客户数据,帮助客户预测用量并跟踪支出。
“我们日复一日地与客户协作,让他们清楚地了解自己的用量模式,并就如何优化额度消耗提出建议,”Burnett 说。
Burnett 表示,基准指标是效率、速度和风险降低。对一些客户来说,这意味着在人员编制不变的情况下承接更多工作;对另一些客户来说,则是更快地完成客户交付。
Parker 敦促律所采取更长远的视角,确保人才培养不会成为追逐效率的牺牲品。White & Case 保留了让初级律师学习的机会,而不是把每一项任务都自动化。Verma 表示,Ironclad 对开发者采取类似做法。
“我们做了一个非常明确的决定,让资深开发者与初级开发者结对,这样他们就能学到判断这门技艺,而这正是 AI 永远无法取代的,”她说。
译文由英文原文机器翻译,术语按行业惯例统一,可在文末展开原文核对。
查看英文原文(How agentic AI raises the bar for law firms)
The legal world is looking to scale AI beyond chatbots, bringing in agents to automate parts of lawyers’ workflows. While adoption is growing rapidly, some firms are moving slower than others. Laura Mandaro, managing editor at The Information, talked to three legal experts about how firms can overcome pushback, encourage adoption and measure AI’s returns.
- Isabel Parker, Chief Innovation Officer, White & Case
- Jillian Burnett, VP Engagement, Legora
- Sunita Verma, CTO, Ironclad
The agent advantage
At many law firms, drafting non-disclosure agreements (NDAs) are a time sink. Sunita Verma of Ironclad, an AI-driven contract management company, said this is a workflow that can easily be outsourced to an agent.
“One of the themes we hear consistently from our customers is that they want their lawyers spending less time on NDAs and more time on strategic deals,” she said. For some clients that have outsourced NDAs to AI, the time saved has made it possible to bring work previously handled by external legal providers in-house.
But not every law firm has embraced AI. Isabel Parker said that while lawyers at White & Case were comfortable using AI for basics like summarizing documents, most remained resistant to a wider rollout of agents.
The firm does use agents to improve client experiences, mostly through client-specific skills that capture preferences, from tone of voice to who needs to be included on contracts and communications. The skills make institutional knowledge easier to share.
“In the past, a lawyer might say, ‘I’m about to submit this to X client, how do they like to see it again? Has anyone written this down?’ And emails would be flying around and everyone’s hair on fire,” Parker said. “This makes it much more streamlined.”
Getting lawyers on board
Jillian Burnett acknowledges that skepticism from lawyers can be a barrier to adopting agentic tools like Legora, the agentic operating system for legal work. A bigger obstacle, she said, is AI fatigue, and an understandable unwillingness to sacrifice time for training with vendors to learn new tools.
“If you just released a new functionality and they just went through training on our assistant, lawyers aren’t necessarily going to want to give you more of their billable time to learn something new,” she said. She stressed the importance of showing lawyers what the platform and its agents can deliver through individual training sessions and small focus groups, noting that lawyers will give you their time when it’s clearly worth it.
“By the time Legora walks into a meeting with a customer, we already understand their workflows and can present use cases directly relevant to the matters on their desk,” she said.
Another challenge is deciding which skills to operationalize.
“When you have a lot of people creating skills, it can get super noisy,” she said. One way to reduce that noise is to set up knowledge teams or steering committees made up of innovation leads, senior partners and other decision-makers to help set the criteria for which skills should become firm-wide standards. Legora can then help users find skills through tags, such as practice areas, and give firms control over how broadly those skills are shared.
Factoring in the costs
As vendors move to consumption-based pricing, law firms face costs that are far harder to predict.
“Accounting for technology spend is entirely new for law firms. It’s utterly unpredictable and we’re just not set up for it yet,” Parker said.
Legora uses customer data to help clients forecast consumption and track spending.
“We’re working with our clients day in and day out to provide them with visibility into their consumption patterns, and to offer suggestions on how to optimize their credit consumption,” Burnett said.
Burnett said the baseline metrics are efficiency, speed and reduced risk. For some clients, that means keeping headcount flat while absorbing more work; for others, faster client turnarounds.
Parker urged firms to take a longer-term view, ensuring that developing talent doesn’t become a casualty of chasing efficiency. White & Case is preserving opportunities for junior lawyers to learn rather than automating every task. Verma said Ironclad takes a similar approach with developers.
“We made a very conscious decision to pair our senior developers with junior developers so they can learn that art of judgment, which is something that AI will never be able to replace,” she said.
正文保留英文原文,标题与摘要由机器翻译。
原文发布于 The Information AI(How agentic AI raises the bar for law firms),标题、摘要与正文由机器翻译,术语按行业惯例统一,未经人工逐句校订。