Our most experienced person retires in two years — how do we keep what they know?
Undocumented knowledge costs large companies $47 million annually, and 55% of executives who replaced staff with AI already regret it. General AI lacks the organizational knowledge for nuanced risk assessment, exposing companies to significant risk if they fail to define a clear decision line.
The question
One or two people carry the judgment that makes our business work, and they will retire or leave within a couple of years. Almost nothing is written down. We can pair them with a successor, pay to have the knowledge documented, encode it into an AI system our teams can query, or accept the loss and rebuild. What actually preserves usable judgment rather than producing documents nobody reads, and where does AI genuinely help versus where is it a distraction from plain succession planning?
Counsel's position
Pair experts with successors, augmenting with structured documentation of critical decision frameworks, as the primary method to preserve usable judgment.
Verdict
The verdict: Pair experts with successors, augmenting with structured documentation of critical decision frameworks, as the primary method to preserve usable judgment.
How the criteria decide
2 of 3 criteria resolved on cited evidence. 1 had none either way.
| Criterion | Favours | Evidence |
|---|---|---|
| What survives the departure in usable form | Pair with a successor | Undocumented knowledge costs large companies $47 million annually in lost productivity It retrieves what's written down, and when what's written down is incomplete, it doesn't quietly compensate — it fills the gap with something fluent and wrong General AI lacks the organizational knowledge required for nuanced risk assessment Organizational knowledge is not something an LLM inherently understands. Yet in contracts-related work, it is often the factor that distinguishes a generic answer from a useful one. Static AI prompt libraries decay and produce inconsistent outputs over time In legal work, the failure isn’t that the AI can’t write. It’s that the AI can’t reliably write to your standards, under your constraints, and keep doing that as those standards shift. |
| Demand placed on the expert's remaining time | No evidence either way | |
| Cost and reversibility if it fails | Pair with a successor | 55% of executives who replaced staff with AI already regret it A separate survey found that 55% of the executives who replaced people with AI already regret it. |
Undocumented knowledge costs large companies $47 million annually in lost productivity
AI systems fail when fed incomplete documentation because they cannot silently compensate for missing institutional wisdom the way human successors do.
55% of executives who replaced staff with AI already regret it
AI successfully automates bounded, repeatable tasks but fails to replicate the unwritten context, judgment, and relationships that veteran employees carry.
General AI lacks the organizational knowledge required for nuanced risk assessment
Purpose-built AI tools that integrate your specific templates, precedents, and negotiation history outperform generic models on tasks requiring institutional context.
Static AI prompt libraries decay and produce inconsistent outputs over time
AI systems require an institutional "memory" layer that captures accepted edits, trusted sources, and routine decisions to provide compounding, context-aware improvements.
AI automates routine transactions but requires human judgment for accountability
Organizations must intentionally draw a "decision line" that delegates repeatable tasks to AI while reserving complex, context-heavy decisions for human experts.
Read another verdict
- Start with a small test, or take on the whole process at once?
- Our competitors advertise AI and we don't — match them, or hold the line?
- Our people already put client files into ChatGPT — ban it, frame it, or supply a tool?
- We can't hire the experienced people we need — automate, train up, or outsource?
- Slow our EU AI Act prep now the deadline's moved to 2027?
- Use AI to flatten middle management this year?
- Let an AI agent act on its own — or keep a human in the loop?
- Invest in pre-build costing or post-deployment ROI tracking?