When you hire a law firm, you are buying knowledge. Knowledge of the law, of course, but also knowledge of the market: what the other side conceded in the last hundred deals like yours, which clause actually gets litigated, which “standard” term is not quite standard. That accumulated experience, more than any individual document, is the product.
This raises an increasingly important question as the legal industry incorporates AI: where does the firm keep that knowledge?
At Most Firms, the Filing System Is a Person
Amazingly, most knowledge at traditional law firms lives solely within people’s minds. The partner who has seen two hundred financings carries the market in her head. The associate who negotiated last year’s version of your vendor agreement remembers which fallback positions worked. The paralegal who silently, heroically keeps the whole operation going. Precedent lives in inboxes, in desktop folders, and in documents named “FINAL_FINAL_REALLY_FINAL_v7.” The firm’s most valuable assets walk out the door every evening, and can walk away from the firm at any time.
That is a structural risk, and it is getting worse. Recent industry reports put associate attrition at around 20 percent a year, with some studies placing firm-wide turnover across all seniority levels closer to 27 percent, and the National Association for Law Placement estimates that replacing an associate costs one and a half to two times their salary. Pricing the lost organizational knowledge due to personnel churn is even more difficult. Thomson Reuters has written about firms scrambling to preserve institutional knowledge as senior lawyers retire, because the strategies that worked and the mistakes learned along the way transfer to the next generation only if someone systematically captures them.
This risk costs clients, even if you don’t always see it: under hourly billing, clients effectively pay for the rebuild. When a team turns over, the new lawyers re-learn your business, your documents, and your preferences, and the re-learning is billable. Thus, in a way, the traditional model incentivizes knowledge loss, since filling gaps puts money back in the firm’s pocket.
Writing It Down Is the Easy Part
These observations aren’t novel. Law firms have maintained precedent banks and knowledge-management departments for decades. In practice, however, these efforts run into several hurdles.
The first is the incentive structure created by hourly billing. When revenue is measured in hours, time spent making the next matter faster is time not billed. Capturing knowledge ends up more as a side project than a primary responsibility.
The second issue, affecting even in-house teams, is difficulty in getting approval for “strategic investment in systems that would deliver efficiency gains in the medium and long terms” as opposed to simply hiring more attorneys.
The last issue is staleness. A repository of forms and positioning on common issues that nobody updates quickly decays in fast-moving practice areas.
A knowledge base is valuable only if it is used on every matter and improved by every matter.
Where Knowledge Lives at General Legal
General Legal, asan AI-native law firm, is set up at its core so that the firm’s knowledge lives in both systems and people. General Legal’s proprietary AI platform is trained on an extensive body of commercial-law knowledge, backed by our attorneys’ years of expertise. Our platform then brings that guidance to bear on the next matter, for every client, from day one. We even maintain bespoke playbooks for individual clients, so the system knows your terms, your red lines, and your preferences before the next matter begins. The hundredth contract review is informed by the ninety-nine that came before it. And by incorporating knowledge management directly into our attorneys’ processes, we’re continuously expanding its value.
We run the rest of the firm the same way. Policies, processes, and project status live in structured workspaces rather than in individual attorneys’ inboxes, readable by attorneys and tools alike. Software engineers call this managing the “bus factor”: the number of people a project could lose before it grinds to a halt. We run the firm so that no single person’s absence puts a matter, or a client relationship, at risk.
The Part That Stays Human
A playbook, however good, is only a record of judgment. It still needs judges. Our model pairs AI that is continually training on the law with experienced attorneys who do the things a model cannot: validate the corner cases, make the judgment calls that depend on what a client actually wants, and, just as important, identify where the model needs to keep learning. New and emerging areas of law, from AI governance to evolving privacy regimes, don’t come with a hundred precedents. There, the attorneys lead and the system learns behind them. As articulated in an earlier post: the AI tells you what is wrong; the attorney decides what to do about it. The playbook is where those decisions go so they don’t have to be made twice.
What This Means for You
For clients, knowledge that lives in systems shows up in three ways.
- Consistency: your third matter is handled with everything the firm learned on your first two, regardless of who staffs it.
- Speed: less time re-deriving known answers means faster turnarounds.
- Predictability: work informed by an ever-deepening playbook is work we can scope, which is part of how we offer flat, published pricing in the first place. Your lawyer’s memory is valuable. It shouldn’t be the only place your legal history exists.
Working With Us
If you’re a founder or operator who would rather build on your firm’s accumulated knowledge than pay to rebuild it, General Legal offers flat-fee, fast-turnaround counsel backed by experienced attorneys. Ready to see the difference an AI-native approach makes? Sign up here to get started.
Sources
• “How to save law firm institutional knowledge,” ThomsonReuters
• “How law firms move from human to institutional knowledge,” Thomson Reuters
• “The Problem of Attrition,” Lawyers Mutual
• “Balancing Attorney and Machine at an AI-Native Law Firm,” General Legal
- At most traditional firms, institutional knowledge, like which terms are actually standard or what got conceded in past deals, lives in individual lawyers' heads and inboxes rather than in any shared system.
- Industry data cited puts associate attrition around 20% a year (27% firm-wide across seniority levels), with replacing an associate costing 1.5-2x their salary, not counting the harder-to-price loss of institutional knowledge.
- Under hourly billing, clients effectively pay to have a new team re-learn their business every time turnover happens, since that re-learning time is itself billable.
- Traditional knowledge-management efforts struggle because time spent capturing knowledge isn't billable, systems investment is hard to get approved over just hiring more attorneys, and unused repositories go stale quickly.
- General Legal says it stores knowledge in a proprietary AI platform and shared structured workspaces rather than individual inboxes, including client-specific playbooks, so a matter is informed by everything the firm has learned before it.
- The firm still keeps attorneys in the loop for judgment calls, novel legal questions, and deciding what happens to newly learned lessons; the AI surfaces issues, but people decide what to do about them.
| The core question | Buying legal services means buying accumulated market knowledge, and the real risk is where that knowledge is stored once the person who has it leaves. |
|---|---|
| The status quo | At most firms, institutional knowledge lives in individual partners' and associates' heads, inboxes, and desktop folders rather than any shared system. |
| The turnover problem | Associate attrition runs around 20% a year (27% firm-wide across seniority levels), and replacing an associate costs roughly 1.5-2x their salary, before counting the harder-to-price loss of institutional memory. |
| Why hourly billing makes it worse | Clients effectively pay for the rebuild when a team turns over, since the new lawyers' re-learning time is itself billable, giving firms little incentive to prevent that knowledge loss. |
| Why knowledge-management efforts often fail | Capturing knowledge isn't billable work, systems investment competes against just hiring more attorneys, and unused repositories in fast-moving practice areas go stale quickly. |
| General Legal's approach | Knowledge lives in a proprietary AI platform and shared structured workspaces, including client-specific playbooks, so every new matter benefits from everything learned on prior ones. |
| What stays human | Attorneys still validate corner cases, make judgment calls, and lead on new or emerging areas of law that don't yet have precedent behind them. |
| The client-facing payoff | Knowledge stored in systems rather than individual memory is said to produce more consistency, faster turnaround, and more predictable, flat pricing. |
What does "buying knowledge" from a law firm actually mean?
It means buying not just legal expertise but accumulated market knowledge, what terms are actually standard, what got negotiated away in similar past deals, which is the real value beyond any single document.
Why does it matter where a firm's institutional knowledge lives?
Because if it lives only in individual lawyers' heads, that knowledge, and the client's history with the firm, walks out the door whenever that person leaves, and industry turnover runs around 20-27% a year.
Does hourly billing help or hurt knowledge retention at law firms?
It hurts, according to the article, because re-learning a client's business after turnover is itself billable time, giving firms less financial incentive to prevent that knowledge loss in the first place.
Why do traditional knowledge-management efforts often fail?
Because capturing knowledge isn't billable work, systems investment loses out to simply hiring more attorneys, and unmaintained repositories go stale fast in fast-moving practice areas.
How does General Legal say it stores institutional knowledge differently?
In a proprietary AI platform and shared structured workspaces, including client-specific playbooks, rather than in individual attorneys' inboxes and memories.
Does this replace the need for human attorneys?
No. The article says attorneys still validate edge cases, make client-specific judgment calls, and lead on emerging legal areas that don't have established precedent yet.