AI notes for consultants in the US: client confidentiality, expert interviews, and the botless meeting

Management consulting runs on two things: the quality of the analysis and the speed with which it can be produced. Senior engagement managers at MBB firms — McKinsey, Bain, BCG — and at the Big Four strategy practices bill at $300–600 per hour. Junior consultants bill at $150–250. Every hour spent reconstructing meeting notes from memory rather than doing billable work is a direct cost to the engagement economics.

The documentation burden in consulting is specific: every client working session, every steering committee update, every expert interview, every team meeting produces content that needs to be captured, synthesized, and turned into the structured output that drives the engagement forward — slide decks, recommendation memos, workstream summaries, and action registers.

AI documentation reduces the capture-to-synthesis cycle. But consulting's specific constraints — client confidentiality, engagement security, the prohibition on bringing bot recorders into client meetings — make the architecture of the AI tool as important as its output quality.

The client confidentiality problem with cloud notetakers

Consulting engagements are among the most confidential professional relationships that exist. Clients share strategic plans, competitive intelligence, M&A intent, workforce restructuring plans, and board-level financial information with their advisors. The non-disclosure agreements that govern these engagements are comprehensive and seriously enforced.

A cloud-based AI notetaker that receives audio from a client working session is a third party receiving confidential client information. Even if the notetaker's terms of service and privacy policy are carefully worded, the audio has left the secure perimeter of the engagement and reached a commercial server.

Major consulting firms address this at the policy level: many prohibit the use of unauthorized third-party tools for client data, require IT security review of any AI tools used on engagements, and in some cases prohibit any recording of client meetings without explicit client consent and engagement-level approval.

On-device processing sidesteps the third-party problem. The audio stays on the consultant's iPhone. No client data reaches a commercial AI server. For a consultant operating under firm policy that prohibits client data leaving approved systems, on-device is the only AI documentation architecture that may be compliant.

The visible bot problem

Bot-based AI recorders — Otter's Notetaker, Fireflies' Fred, Fathom, tl;dv — join video calls as visible participants. In a client steering committee meeting with the CEO and CFO of a Fortune 500 company, a bot recording the conversation is not a neutral presence. Clients notice. Clients ask. The answer — "that's my AI notetaker" — is not always welcome in a meeting where strategic confidentiality is assumed.

The same dynamic applies in person. A consultant placing a recording device on the table in a client conference room signals something about the confidentiality of the conversation that may not be appropriate for the relationship.

Kuulo records from the iPhone in hand or pocket. The iPhone is a normal presence in any meeting. The recording is not announced by a bot joining a call — it is a device tool like any other. For consultants who need to capture client meeting content without signalling that they're recording, this is a material difference.

Expert interviews: the core consulting documentation challenge

Strategy consulting engagements rely heavily on expert interviews: conversations with industry participants, former executives, technical experts, and market analysts that build the primary research base for recommendations. An engagement may require 30–60 such interviews over a 6–8 week period.

Each interview is 45–60 minutes. The content is highly specific: market dynamics, competitive positioning, operational practices, pricing structures. The interview notes need to be structured enough to be synthesized across 40 other interviews, attributed well enough to be traceable, and available quickly enough to feed into the evolving analysis.

The traditional interview documentation workflow: the consultant takes rough notes during the call, writes the full interview summary from memory afterwards, and uploads it to the engagement's shared document environment. At 45 minutes per interview summary across 40 interviews, this is 30 hours of documentation work on top of 40 hours of interview time.

Kuulo records the expert interview on-device, generates an attributed summary, and produces a structured note covering: expert background, key claims made, supporting evidence cited, conflicting information noted, and strategic implications. The consultant reviews and confirms accuracy. Total post-interview documentation time: 8–12 minutes per interview. Across a 40-interview engagement, that is 22–24 hours recovered.

Team meetings and workstream documentation

Consulting teams work fast and change direction frequently. A workstream meeting on Monday produces hypotheses; Tuesday's client data changes them; Wednesday's team review realigns the approach. The decisions made in Monday's meeting, and who made them, matter when the engagement comes under scrutiny — in a client presentation, an internal review, or a dispute about recommendations.

Meeting notes in consulting are governance records as much as memory aids. Who said what, what was decided, what was flagged as a risk, what was escalated to the engagement manager — these are the content of the notes, and they matter.

Kuulo's diarization attributes contributions in team meetings — the project manager's synthesis, the associate's data concern, the partner's strategic reframe — producing attributed notes rather than an undifferentiated summary. For a team of 4–6 people, diarization reliably separates the voices that matter.

Offline operation at the client site

Many large client sites — manufacturing plants, hospital systems, government buildings, financial institutions — have restricted wifi environments. Connecting a personal device to a client's corporate network for the purpose of running an AI notetaker raises immediate IT security concerns. Relying on mobile data in a basement server room or a secure conference room is unreliable.

Kuulo processes on-device regardless of network availability. A factory floor interview, a secure government briefing room meeting, or a hospital department working session all produce the same quality of AI documentation as an office meeting with full connectivity. The infrastructure of the client site does not constrain the documentation capability.

The billable hours argument

At $250 per hour, 20 minutes of post-meeting note reconstruction costs $83 in billable equivalent time per meeting. At 6 meetings per day, 5 days per week, a junior consultant recovers nearly $2,500 per week in billable-equivalent time by reducing post-meeting documentation from 20 minutes to 5. The tool pays for itself in the first hour of use.

For engagement economics that are tracked at the workstream and engagement level, the time recovered from documentation is time available for analysis, synthesis, and client interaction — the work that actually determines whether the engagement recommendation is good.