Find the People Who Hold Your Firm’s Project Knowledge

AI Strategy 11 min read
Featured image for Find the People Who Hold Your Firm’s Project Knowledge
Illustration: Dan Cumberland Labs with Gemini.

You already know who this person is. If they gave notice tomorrow, there are three or four things (client relationships, project history, process exceptions) that would suddenly be impossible to find. You'd know what the gap was. But you wouldn't know how to fill it.

The problem isn't the departure. It's that you never mapped what this person was carrying.

Every professional services firm has people who function as its living knowledge catalog. The firm only notices when one of them leaves. This article gives that kind of person a name, shows you how to identify who yours are, and gives you one concrete first step you can take this week.

What Is a Walking Index?

A walking index is a person in your firm who functions as its living knowledge catalog. They know where everything is, who knows what, what the history is, and what the exceptions are— and they weren't hired to be this, but they became it.

A walking index is not on any org chart as a "knowledge holder." They're just the person everyone asks.

They're also not the same as your top biller or your most senior title. This is specifically about knowledge holders. The distinction matters because walking indexes are often mid-level— a project manager who's handled every difficult client for six years, an admin who's been answering the same questions since the firm had five people.

Walking indexes carry five types of knowledge that knowledge management practitioners call "tribal knowledge"1:

  • Process knowledge: How work gets done, not how the manual says it should. ("We always send the submittals to Carlos directly, not through the portal.")
  • Exception knowledge: What to do when the rules don't fit. ("If the GC is Meridian, skip the standard RFI process entirely.")
  • Historical context: Why decisions were made the way they were. ("We stopped using that structural detail after the Prescott project.")
  • Relationship maps: Who to call, who trusts who, what the backstory is with a given client or sub.
  • Interpretive knowledge: What a policy means in practice, not just what it says.

Every firm has walking indexes. Most firms have 2–4. Most firms don't know who they are until someone gives notice.

Three Questions That Find Them

To find your walking indexes, ask three questions about the people in your firm. Anyone who triggers a long pause on more than one is carrying knowledge you haven't mapped.

According to APQC research2, expertise location— knowing who holds what knowledge— is a distinct knowledge management function that most organizations skip entirely. The identification step has to come before the documentation step. But you don't need a system to start. You just need three questions.

Question 1: "If this person left tomorrow, what would we suddenly not know?"

Run through your team. For most people, you'll have an answer quickly. For a handful, you'll pause. That pause is the signal. The person who makes you think— "hm, who else would know about the Hartwell account history?"— is a walking index.

Question 2: "Who does everyone else go to for answers?"

There's usually someone who gets pulled into conversations they weren't invited to, whose opinion gets sought before decisions get made, whose desk (or Slack DM) is a consistent stop. Track the informal question-traffic in your firm. The person at the center of it is your walking index.

Question 3: "Who's the only one who would know?"

Think about your last difficult project or last client complaint. Think about the last time someone on your team said, "I'm not sure— let me ask [name]." That name keeps coming up. It's not a coincidence.

Why They Don't Know They're Walking Indexes

Walking indexes don't think of themselves as walking indexes. They think they're just doing their job.

Tacit knowledge— the kind walking indexes carry— is unconscious by definition. It doesn't feel like expertise to the person holding it. It feels like knowing how things work. New employees, by contrast, spend 40–60% of their onboarding time just trying to acquire this undocumented knowledge from the people around them3. The walking index didn't decide to be the keeper of it. They just answered enough questions, worked through enough exceptions, and stayed long enough to accumulate what no one else has.

There's also a second dynamic worth naming: some people DO know they're walking indexes— and they use it. The person who's the only one who knows where something is has a structural reason not to document it. That's not cynicism— it's human nature. Which is exactly why the mapping step has to be done by the founder, not delegated to the people being mapped.

The practical implication: you can't just ask a walking index to "write down what they know." That almost never works. They're inside their own knowledge. They can't read the label from inside the bottle. You have to sit down with them and ask the right questions. That's why identifying them first— knowing exactly who to have that conversation with— is the step that makes everything else possible.

What Happens When a Walking Index Leaves

When a walking index leaves without a transfer plan, the cost is real— and most of it is invisible.

  • US private sector voluntary turnover runs 22–25% per year4. For a 10-person professional services firm, that's 2–3 people per year on average. At least one of them is probably a walking index.
  • Deloitte research5 found that organizations fail to capture knowledge from departing employees in 92% of cases. Most firms don't have a plan— they have an exit interview and a hope.
  • Replacing a mid-level employee costs 6–9 months of their annual salary6. That's before counting the knowledge gap: the time new hires spend relearning what was never written down, the mistakes they make that the walking index would have prevented.

And it doesn't stay contained. As ArgusLabs puts it7: "The tribal knowledge problem does not create a one-time gap. It creates a compounding deficit." Every unplanned departure makes the next gap wider.

In AEC specifically, the institutional knowledge crisis in AEC is acute. An estimated 1.7 million infrastructure workers leave their positions each year, largely due to retirement8. This isn't a future concern. It's already happening.

The Walking Index Mapping Exercise

The mapping exercise takes 20 minutes. You need a piece of paper (or a blank doc) and a list of everyone in your firm. That's it.

Step 1— List everyone. Start with every person on your team. Include part-timers, long-tenured admins, anyone who's been with the firm more than five years. This takes two minutes.

Step 2— Run the three questions. For each person, ask: Would we suddenly not know things if they left? Do they get asked questions more than they ask? Are they the only one who would know X? Mark anyone who triggers a yes on two or more.

Step 3— Rank by departure risk + knowledge weight. Of the people you marked, consider who's closest to leaving— approaching retirement, job-shopping, burned out— and who carries the deepest knowledge. The intersection of high-knowledge and high-departure-risk is your priority list.

Step 4— Schedule conversations. The output of this exercise is a short list of names and a calendar invite. Not a knowledge base, not a software implementation— just a conversation. Documentation comes after.

You're looking for 2–4 people, not 20. If everyone comes up as a walking index, the firm has a deeper structural issue. Most firms land on a small number of people who truly carry irreplaceable knowledge. That's who you need to sit down with first.

In conversations with firm owners, the pattern is consistent— most can name their walking indexes in 10 minutes once someone asks the right questions.

One honest note: this exercise also reveals which roles have been filled by single points of failure. That's a hiring and documentation signal going forward. For firms ready to take that next step, building out a knowledge management system is the natural follow-on.

The goal of the mapping exercise isn't a complete knowledge audit. It's a list of names— the people you need to have a conversation with before they leave. That list is worth more than any knowledge management software you could buy before you've done this step.

What Comes After the Map

The mapping exercise gives you a list of names. The next step is structured knowledge extraction— and this is where AI tools have changed what's possible.

Once you know who your walking indexes are, the capture step becomes concrete. A category of AI tools now exists specifically for this: interview platforms that record conversations, transcribe them, and organize what emerges into searchable documentation. The capture that used to take weeks of manual documentation now takes hours of structured conversation— with AI organizing the output and your team validating what matters.

The sequence matters. Map first, capture second. Most firms try to implement knowledge management systems without knowing who to target— and it fails. The mapping exercise is what makes the capture step possible.

Deloitte frames it this way5: "Organizations treating knowledge management as a strategic priority— not just technical implementation— are positioned to convert retirement waves from risk into competitive advantage."

The goal isn't to capture everything your walking indexes know. It's to capture enough that the firm doesn't lose continuity when they leave— and to signal that their knowledge is valued now, not just when they're handing in their badge. Building a culture where knowledge doesn't stay siloed starts with that conversation.

The sequence that follows is straightforward: structured conversations with the people on your list, recorded and organized by AI tools that can compress weeks of manual documentation into hours. For founders navigating exactly these decisions, working with an AI implementation partner can help you move from names to a structured extraction plan.

But none of that happens until you have the list. The mapping step isn't the beginning of a knowledge management project. It's the beginning of knowing what the project is.

FAQ

What is a walking index in a business?

A walking index is an employee who functions as their organization's living knowledge catalog— someone who knows where information lives, who to ask, and what the exceptions are, without being formally assigned to that role. They weren't hired to hold this knowledge; they accumulated it over time. Most firms have 2–4 of them, and most firms don't know who they are until one gives notice.

How do I identify who the key knowledge holders are in my firm?

Ask three questions for each person on your team: Would we suddenly not know things if they left? Does everyone go to them for answers? Are they the only one who would know X? Anyone who triggers a yes on two or more of these is a walking index— a priority for structured knowledge transfer. The fastest version: ask yourself who everyone else goes to for answers. That person is your walking index.

What is the cost of losing institutional knowledge?

SHRM estimates6 that replacing a mid-level employee costs 6–9 months of their annual salary— before accounting for the knowledge that doesn't transfer. APQC research2 found that only 8% of organizations consistently capture knowledge from departing employees. That means in 92 out of 100 departures, the knowledge walks out and doesn't come back5.

What is tribal knowledge in the workplace?

Tribal knowledge is informal organizational expertise— process shortcuts, exception rules, historical context— held by individuals but never formally documented. It's the knowledge that works but has never been written down, and it typically lives in the people who've been with the firm the longest. Walking indexes are the people who hold the most of it.

Can AI help with knowledge transfer?

Yes— once you've identified your walking indexes, AI interview transcription and analysis tools can convert structured conversations into searchable documentation9. But the mapping step must come first. AI capture tools work best when you know exactly who to sit down with and what to ask. The technology is ready; the bottleneck is almost always the identification step.

References

  1. Atlan, "Tribal Knowledge: Definition, Five Types and AI-Era Risks" (2025) — https://atlan.com/know/data-for-ai/tribal-knowledge/
  2. APQC, "Knowledge Management Tools for Expertise Location" (2024) — https://www.apqc.org/blog/knowledge-management-tools-expertise-location
  3. Atlan, "Tribal Knowledge: Definition, Five Types and AI-Era Risks" (2025) — https://atlan.com/know/data-for-ai/tribal-knowledge/
  4. U.S. Bureau of Labor Statistics, "Job Openings and Labor Turnover Survey (JOLTS)" (2023) — https://www.bls.gov/jlt/
  5. Deloitte, "Capturing Institutional Knowledge" (2024) — https://www.deloitte.com/us/en/insights/topics/talent/knowledge-management-plan.html
  6. Society for Human Resource Management (SHRM), "Retaining Talent Toolkit" (2023) — https://www.shrm.org/topics-tools/tools/toolkits/retaining-talent
  7. ArgusLabs, "The Tribal Knowledge Problem: When Your Best People Leave" (2024) — https://arguslabs.ai/tribal-knowledge-problem-when-best-people-leave/
  8. Knowledge Architecture, "Best Practices: Knowledge Management for AEC Firms" (2024) — https://www.knowledge-architecture.com/aec-knowledge-management
  9. University of Vermont, "How to Capture Knowledge from Retiring Employees Using AI" (2025) — https://learn.uvm.edu/news/how-to-capture-knowledge-retiring-employees-ai/

Our blog

Latest blog posts

Tool and strategies modern teams need to help their companies grow.

View all posts
Featured image for Build An AI Champions Network From The Bottom Up