The AI That Reads the Room, Then Reads the Whole Account

Ask an account manager what they actually did last week and the honest answer is a little grim. They wrote things down. Recaps of calls, follow-up emails, status notes, and - if the calendar was cruel - a quarterly business review deck that ate a full afternoon. The client relationship, the thing they were hired to nurture, mostly happened in the gaps between the admin. Kaizan, a London company founded in 2022, is built on the bet that most of that gap can be closed by software that listens as carefully as a good account director does.

The company describes itself as "client super intelligence for professional services firms." In plainer terms: Kaizan plugs into the places where client conversations already live - meeting transcripts, email, chat - and turns the scatter into a single, current picture of how each relationship is doing. It does not want to be another place you have to remember to update. It wants to read what is already there and tell you what it means.

What it doesListening, scoring, drafting

Three verbs describe the product. It listens: Kaizan ingests transcripts from Zoom, Google Meet, Teams, Gong and the rest, plus email from Gmail and Outlook and messages from Slack. It scores: every account gets a live reading of its health, backed by quotes pulled from the source conversations rather than a vibe. And it drafts: the follow-up email, the meeting agenda, the QBR narrative - the writing that fills an account manager's week - comes back pre-written, waiting to be edited rather than composed from a blank page.

The scoring is the part Kaizan has built a name around. It is called the CARE model, and it grades each relationship from zero to ten across four pillars.

The illustrative bars above stand in for what a live account view looks like. The point of the model is uncomfortable in a useful way: a relationship can feel fine in a friendly call and still be quietly slipping on Activity or Relationship depth. Kaizan's argument is that those signals are already sitting in the email thread and the meeting transcript - nobody has had the time to read all of it at once.

The problemThe unhappy client who never says so

Churn in professional services rarely arrives as a dramatic breakup. It arrives as a slow fade - fewer replies, a stakeholder who left and was never replaced in the relationship map, a renewal that goes cold. By the time it shows up in the revenue numbers, the useful window to fix it has closed. Kaizan's pitch is early warning: surface the risk while there is still a relationship to save, and flag the expansion opportunity while the client is still receptive.

The numbers Kaizan puts to this are worth reading as claims rather than guarantees, but they sketch the shape of the value. The company reports account managers saving several hours a week on reporting, QBR prep dropping from a half-day to under an hour, and revenue per client growing north of 20 percent where the tool is used well.

There is also a retention figure that stands out: a reported median net dollar retention of 146 percent for clients using the platform, against a sector benchmark the company cites at 105-115 percent. Net dollar retention above 100 percent means existing clients are spending more over time, not less - which is precisely the outcome Kaizan is built to engineer.

How it's differentSolving the account, not the meeting

The obvious neighbours are the AI note-takers - Otter, Fireflies, Gong, Chorus - and Kaizan is careful to define itself against them. A note-taker captures one meeting and hands you a transcript. Kaizan's claim is that it connects every meeting, email and message into a running, stateful understanding of the relationship, and keeps a score that moves over time. The distinction it draws with general chatbots is sharper still: ChatGPT and Claude are powerful, but they cannot see your client conversations, do not hold a health score across quarters, and do not write structured outputs back into your systems.

Capability	Note-takers	General LLM	Kaizan
Captures the meeting	yes	no	yes
Connects every channel	no	no	yes
Stateful health score	no	no	yes
Writes back to CRM	no	no	yes
Flags churn / expansion	no	no	yes

Notably, Kaizan does not try to replace the CRM. It sits alongside Salesforce, HubSpot or Pipedrive and writes structured outputs back into them - a "system of work" next to the system of record. That is a deliberate positioning choice. Ripping out a CRM is a war; supplementing one is a Tuesday.

The foundersAn ad-tech exit and a Facebook engineer

Kaizan was started by two people who have shipped before. Glen Calvert, the CEO, previously founded Affectv, which was acquired, and was part of the founding team at Struq - a decade-plus in ad tech, a field that lives and dies on reading signals in messy data. Pravin Paratey, the CTO, has co-founded and exited technology companies and held senior engineering roles at Facebook and OVO. The pairing reads like a founder's balance sheet: one who has sold the value of prediction to enterprises, one who has built the systems that make prediction reliable.

The name is the tell. Kaizan comes from kaizen (改善), the Japanese philosophy of continuous improvement made famous on the factory floor. The company's read on client service is the same as the Toyota Production System's read on manufacturing: outcomes are the sum of small, compounding improvements, applied relentlessly. Do that across every client you have, and the aggregate is a very different business.

Business & marketPriced by client, not by seat

Kaizan sells B2B SaaS, but with a pricing quirk that says a lot about who it is for: it charges by the number of clients you manage, not the number of users you have. There are no seat limits. For an agency where the whole team touches an account, per-seat pricing would tax collaboration; per-client pricing lines the cost up with the thing being grown. Contracts run twelve months, billed annually, with multi-year options, and Kaizan runs paid pilots rather than free trials - a filter that selects for buyers who mean it.

The target customer is specific: client-services teams in media, creative, consultancy and SaaS, usually organisations with somewhere between 10 and 500 client-facing people. The buyer is typically a Managing Director or Head of Client Service; the daily user is an account manager carrying six to twenty-five relationships. Kaizan says it is used by more than 60 such teams, a roster that includes Jellyfish, Tradedoubler, The Gap Partnership, Gravity Global, NP Digital, The Kite Factory and Searchlab.

The money followed the traction. After a $1.2M pre-seed in 2022, Kaizan raised a £2.5M seed round in May 2026 led by Pembroke VCT, with Velocity Capital and a group of industry angels, earmarked for expansion into the United States and further work on the CARE model.

Where it fitsThe most under-automated job in B2B

Sales got its software a decade ago. Marketing got its stack. Customer success got platforms like Gainsight. The account manager - the person who keeps existing revenue alive and growing - has largely been left with a CRM built for pipeline and a calendar full of admin. Kaizan is aiming at that gap: not winning new logos, but growing the ones already signed. In a market where acquiring a new client costs far more than keeping one, that is not a small corner to own.

Whether Kaizan becomes the default layer for client service or one option among several will depend on how well the CARE score holds up under the weight of real, weird, human relationships - the ones that do not fit a four-pillar grid. But the framing is sound, and the timing is good. The tools to read a million conversations accurately only recently got cheap enough to point at a problem this unglamorous.