News You Can Use

Edition 2 · 14th - 30th Sep 2024

News You Can Use

Deep Dives

Three stories worth sitting with

CoPilot Wave 2

What
MS Copilot is expanding its features, including being more integrated across Office apps and some advancements around Copilot Studio. There are also links with OpenAI's custom GPTs. Addition of some interesting features (think deeper, voice briefings, improved search, personalised suggestions).
So what
A more robust Copilot Studio will enable us to build out knowledge and Q&A solutions on knowledge. Better links with Office apps may allow us to surface solutions using Copilot in Outlook and Teams easier. This may also drive more clients to purchase Copilot and push work towards our prompting workshops.

Spellbook Associate

What
An agent framework developed by Spellbook that will allow a virtual "associate" AI agent to look through documents and provide mark ups, essentially trying to automate drafting, negotiation and review processes through a central tool connected to the Spellbook solution in Word.
So what
This is a move towards more self-sufficient playbooks, being able to arm an AI Agent with the rules of a contract and allow it to independently review and mark up documents will be very useful for our MLS offerings as well as standard contract reviews internally. We should get access to the beta of this soon and will share findings.

Lexis Survey

What
Survey of 800+ lawyers. 82% of lawyers are currently using or have plans to use GenAI. A 4X within the last year! 71% of lawyers said that delivering work faster is the top benefit of AI. Clients expect law firms to be using AI, and therefore reducing their bills. Still only 17% of private practice think AI will end the billable hour model...
So what
Increasing adoption, engagement and actual growth in deployments show that GenAI is still on an upwards trend. It also shows that law firms and lawyers are still catching up and still unsure of how to practically start the wheels moving, even if they want to get things done. Keeping up with tech is seen as a major challenge for lawyers - and this is an opportunity to talk about our expertise and cross sell our workshops and trainings, as well as focus internally.

Harvey

What
Harvey have added SPA Dealpoints to their benchmarking data, as well as releasing some further info about features and tests.
So what
Dealpoints is a key use case around post completion work, and the approach taken by Harvey should be replicable within an AGPT based solution or something similar to Deal Clarity. It would be useful to assess our own tools in similar ways. There is already ongoing work on the previous Harvey benchmarks that we will share some results of.

FT Innovative Lawyers

What
Annual report setting out the most innovative law firms in Europe. AG were ranked 8th and marked Standout for our GenAI Strategy (with DLA Piper and A&OS).
So what
It was interesting to see what other firms are doing, with DLA Piper's acquisition of a data science team and use of CoCounsel driving their win in the GenAI Strategy award, alongside the work A&OS who have invested a lot of time and money with Harvey and developed their ContractMatrix tool. Good to be recognised as top ten in Europe as a firm, considering we are not at the same level as many other firms in that bracket (i.e. Magic Circle / Silver Circle).

Worth Reading

Everything else worth a click

New Zuva (ex-Kira) Product

Noah Waisberg's second act after Kira - Zuva Analyze combines GenAI and ML for contract review, with flexible project-based pricing aimed at in-house teams.

Legal Tech Trends Newsletter #32

Peter Duffy's fortnightly scan - covers the Lexis survey, Spellbook Associate, Workday buying Evisort, and the expectations gap between Fortune 1000 GCs and law firm leaders.

Scaling the state of play in AI

Ethan Mollick's state-of-play on scaling - explains the two scaling laws (training compute and inference compute) and why Gen3 models are about to arrive at billion-dollar training costs.

SPA specific Harvey benchmark

Harvey's task-specific benchmark on extracting SPA deal points - foundation models hit 66-72% accuracy, Harvey's specialised agent reaches 98.47%, making the case for purpose-built systems.

Lexis Survey

Lexis H2 2024 UK lawyer survey - 82% have adopted or are planning to adopt GenAI (nearly quadrupled since 2023), 41% are active users, 71% cite speed as the biggest benefit.