News You Can Use

Edition 29 · 1st - 14th Nov 2025

News You Can Use

Deep Dives

Three stories worth sitting with

McKinsey - The State of AI in 2025: Agents, Innovation, and Transformation

McKinsey - The State of AI in 2025: Agents, Innovation, and Transformation

What
According to McKinsey's 2025 global survey, AI use is very widespread - 88% of respondents say their organisation uses AI in at least one business function. However, scaling AI beyond pilots remains a challenge: only a fraction of companies have integrated AI deeply across workflows or captured enterprise-level value. The study also highlights a rise in agentic AI and an uptick in new roles, governance frameworks, and workflow redesigns required to embed AI in practice.
So what
This is a useful report to point to and show our own move away from experimentation and more into deployment. We are well on our journey compared to others. It also shows the importance of looking at value and ROI as we may start to get questions as the market moves. Two thirds of organisations haven't rolled out AI enterprise wide, despite the number of pilots ongoing - we should be talking to clients about how we are doing this already and doing it successfully. There is a lot of noise but not much practical application and we can capitalise on this with our experience.

How Do AI Agents Do Human Work?

How Do AI Agents Do Human Work?

What
The paper introduces a new benchmark comparing how AI agents versus humans execute complex workflows across various occupations - comparing 48 human workers with 4 agent frameworks over 16 real tasks. The agents are ~88% faster and 90-96% cheaper, but with systematically worse quality and frequent fabrication. This reveals that while agentic AI is improving, there remain significant gaps in understanding human workflow structure, tool-use, context switching and judgement.
So what
The findings underline that if we build agents (or advanced workflows), it is crucial to test end-to-end execution, and not just task completion. We should identify where human oversight remains essential, measure how our internal tools handle workflow transitions, and design for real legal work patterns rather than idealised flows. For example, for Power Automate/Copilot Studio workflows, our pilots should monitor workflow fidelity, transitions, error-handling and include human-in-loop triggers.

Briefing-HSBC: Law firm strategy and investment survey 2025/2026

Briefing-HSBC: Law firm strategy and investment survey 2025/2026

What
Based on inputs from senior legal business leaders in major UK firms, the survey highlights current strategic priorities in the legal sector. Key themes include investment in legal technology, reshaping cost-structures, new operating-models and the importance of talent and data as drivers of service delivery change. It highlights a prevailing confidence among law firms to continue expanding and investing in technology despite macroeconomic and geopolitical uncertainties, with a significant focus on international growth, particularly in North America and the Middle East.
So what
This suggests that our strategic focus areas remain extremely relevant: integrating AI into service delivery, optimising matter and resource models, and leveraging data to differentiate our offering. 82% of law firms surveyed said their technology spend is expected to increase or significantly increase next year - and 70% of firm leaders say tech drive service delivery as their main disruptor. The report validates that many firms are ramping up their investment in tech and transformation, we should be prepared to show how AG is ahead of that curve, and how our tools and structural initiatives support firm ambitions - this needs to be beyond the basics and demonstratable to clients.

Worth Reading

Everything else worth a click

Harvey - 3 Principles That Helped us Scale Agent Development

Harvey engineers set out three principles for moving from prompt-based features to scalable agent frameworks, covering shared infra, evaluation and safety. Useful if you're thinking about how to industrialise "agents" inside a firm rather than just ship one-off copilots.

Do LLMs Truly Understand When a Precedent is Overruled? (PDF)

[Internal AG resource] New benchmark on 236 SCOTUS case pairs finds long-context LLMs still struggle to reliably spot overruling relationships, especially in older cases. Shows era bias, shallow heuristics and "temporal reasoning" failures that matter for serious legal research use cases.

Harvey - Introducing BigLaw Bench: Arena

Harvey's new internal "arena" pits models and systems head-to-head on real expert preferences, not just leaderboard benchmarks. Interesting blueprint for how firms might evaluate models and vendors at scale rather than trust marketing slides.

Richard Mabey - Are lawyers a special case for AI?

Mabey argues lawyers aren't magically exempt from the automation curve and should stop assuming "this time it's different" for legal. A sharp take on where legal work really is and isn't defensible against AI.

Legora Client Portals launch

Legora's new Portal capability extends its knowledge platform into secure client-facing workspaces, tying together DMS content, portals and intake systems into one view. Useful benchmark for what client portals could look like in an AI era.

Legal IT Insider - White & Case Launches AI Assistant Atlas

White & Case unveils Atlas, a multi-model, ChatGPT-style assistant built on a proprietary AI platform, plus a firm-wide change programme to drive adoption. The kind of coherent AI roadmap clients will soon expect from every global firm.

Ada Lovelace Institute - The Dilemmas of Delegation

100-page deep dive on the policy and governance headaches created by AI assistants and agents - from accountability and labour impacts to "how much should we actually delegate?" Ideal for anyone drafting firm or regulator-level AI policies.

OpenAI - GPT-5.1 Launch

OpenAI's latest flagship promises faster, more conversational performance than GPT-5 with better tool use and cheaper tokens, plus simpler ways to build customised assistants. Relevant if you're revisiting your model stack or planning 2026 platform migrations.

Rosie Dent-Spargo - Level Up Your Vibe-Coding

Practical on-ramp for ops, product and legal folks who want to move beyond no-code and into tools like Cursor, v0 and proper engineering workflows. A good confidence booster for "non-technical" innovators who secretly are.

Jordan Furlong - Private Equity at the Gate

Furlong warns that PE will treat law firms like any other asset - squeezing margins, pushing scale and expecting exits - with big implications for culture and ethics. Essential reading if PE is sniffing around your market.

Claude - Best practices for prompt engineering

Concrete patterns that work: be explicit, set roles and constraints, use worked examples, add structure (XML tags), and iterate with evals/guardrails. Useful to standardise team prompts and cut rework.

Anthropic - Code execution with MCP

Shifts agents from brittle tool-call prompts to code that imports MCP tools on demand, filters data before the model, and runs logic off-context. Read if you build enterprise agents - fewer tokens, tighter security/state, clearer execution.

Vigilante Lawyers Expose the Rising Tide of AI Slop in Court Filings (PDF)

[Internal AG resource] Lawyers are publicly cataloguing AI-generated fake citations and other "slop" in filings; sanctions are up, but deterrence is patchy. Use this to justify source-verification checks, AI-use logging, and targeted litigator training.

Google - Context Engineering: Sessions, Memory (PDF)

[Internal AG resource] Google's internal whitepaper on "context engineering" lays out how to design sessions, memory stores and retrieval for long-lived agents. Gold dust for anyone building serious agent frameworks rather than single-shot chats.