Frequently asked questions about Kaizan.

Kaizan is an AI Account Manager built for client-services teams. It listens to every client meeting, email and message, scores the health of each relationship, surfaces risks before they become churn, and drafts the follow-ups, recaps and QBR decks that account managers spend most of their week writing. Kaizan is used by 60+ client-services teams including Anything Is Possible, The Kite Factory, Adimo, Verkeer, Jellyfish and Scale.

Kaizan was founded by Glen Calvert and is headquartered in London, with a remote-first team of fifteen people across the UK and Europe. The company has raised funding from notable angels and early-stage funds focused on B2B SaaS and AI infrastructure.

Kaizan is taken from the Japanese word kaizen (改善), meaning continuous improvement. The product is designed around the same idea: client relationships compound, and small, consistent improvements in how an account manager listens, follows up and reports compound into materially better retention and expansion outcomes.

Kaizan is designed for client-services teams: media agencies, creative agencies, consultancies, SaaS customer success organisations, and professional services firms. The typical buyer is a Managing Director, Head of Client Services, Chief Customer Officer, or Head of AI. The typical daily user is an Account Director, Account Manager, or Customer Success Manager who owns a portfolio of 6 to 25 client relationships.

Kaizan is most useful for teams of 10 to 500 client-facing people. Smaller teams (under 10) tend to manage relationships informally and do not yet feel the reporting burden Kaizan removes. Teams over 500 typically run Kaizan in 2 to 3 business units in parallel rather than one global rollout.

No. Kaizan is not a CRM and does not aim to replace Salesforce, HubSpot, or Pipedrive. Kaizan sits alongside the CRM, listens to the actual conversations happening with clients, and writes structured outputs (health scores, risks, action items, recap emails, QBR decks) back into the CRM and the team’s document tools. Kaizan customers keep their CRM as the system of record and use Kaizan as the system of work.

Kaizan ingests three sources: meeting transcripts (from Zoom, Google Meet, Microsoft Teams, Gong and Chorus), email threads (from Gmail and Outlook / Microsoft 365), and chat (from Slack and Microsoft Teams chat). Audio is transcribed by a speech-to-text model with speaker diarisation. Text is parsed for participants, topics, commitments, risks, sentiment and questions.

Four things, every week, for every account: (1) a health score across four dimensions — Coverage, Activity, Relationship, Expansion (the CARE model); (2) a list of risks and opportunities with the underlying evidence cited from real conversations; (3) drafted follow-up emails, recap notes and meeting agendas in the account manager’s voice; (4) a QBR-ready deck assembled from the quarter’s conversations, decisions and outcomes.

CARE is Kaizan’s framework for account health, with four pillars. Coverage measures how many stakeholders on the client side your team is in regular contact with. Activity measures the cadence and quality of touchpoints. Relationship measures sentiment and trust signals from language used in real conversations. Expansion measures observed buying signals and growth intent. Each pillar is scored 0 to 100 weekly, with the underlying evidence linked.

Kaizan’s sentiment model is trained specifically on B2B client-services language, which behaves very differently from consumer reviews or support tickets. Internal benchmarks across 4.1 million scored conversations show 92% agreement with human annotators on a five-point scale (very negative, negative, neutral, positive, very positive). Sentiment is always shown alongside the source quote so account managers can verify the call.

Standard onboarding is two weeks. Week one: connect data sources (calendar, email, meeting recorder, CRM) and import the last 90 days of history so health scores are populated on day one. Week two: shadow rollout with one team, calibration of scoring, and team training. Most clients reach full team adoption within 30 days of kickoff.

Kaizan has native integrations with: Salesforce, HubSpot, Pipedrive (CRM); Gmail, Outlook / Microsoft 365 (email); Google Calendar, Microsoft Outlook Calendar (calendar); Zoom, Google Meet, Microsoft Teams, Gong, Chorus, Fireflies, Otter (meetings and recordings); Slack, Microsoft Teams (chat); Google Drive, Notion, Confluence (documents); Linear, Jira, Asana (work tracking). Custom integrations are available on enterprise plans via Kaizan’s ingestion API.

Yes. Kaizan offers a REST API on Growth and Enterprise plans for ingesting custom data sources, exporting health scores and risks into internal dashboards, and triggering workflows in other systems. Webhooks are available for real-time events (new risk detected, health score change, meeting transcript ready).

Kaizan is priced based on the number of clients you have, there are no limits to the number of seats or users. List pricing and a calculator are at kaizan.ai/pricing.

No, we do paid pilots so you can assess the ROI and value received.

Standard contracts are 12 months, billed annually, with quarterly business reviews. Multi-year contracts (24 and 36 months) carry a discount and are common for enterprise clients. Month-to-month is available on Starter for teams piloting Kaizan before formal procurement.

Across the active client base, Kaizan clients report a median 146% net dollar retention versus a sector benchmark of approximately 105% to 115%. Account managers report saving an average of 3.2 hours per week on reporting and follow-up work. Quarterly business review preparation drops from a typical 4 to 6 hours per account to 30 to 60 minutes. Client satisfaction scores (CSAT) typically move 12 to 20 points within the first two quarters of rollout.

Day 1–14: integrations connected, last 90 days of history backfilled, CARE scores live for every account. Day 15–45: account managers using drafted recaps and follow-ups daily; first risks caught and saved. Day 46–90: first QBR cycle run inside Kaizan; usage benchmarks and account-level outcomes reviewed with the Kaizan customer success team and a written 90-day report delivered.

AI note-takers transcribe meetings and write meeting notes. Kaizan does not stop there. It connects every meeting, email and message for an account, scores relationship health, predicts churn and expansion risk, and writes the account-level artefacts (QBR decks, weekly client recaps, save plans). Note-takers solve the meeting; Kaizan solves the account.

A general-purpose LLM does not have access to your client conversations, does not understand client-services-specific signals (CARE health, coverage gaps, expansion language), does not maintain stateful health scores over time, is not SOC 2 Type II certified for processing client data, and does not come with the integrations, redaction pipeline, and per-account memory Kaizan provides out of the box. Several Kaizan clients evaluated building internally and chose Kaizan to reach production faster and with stronger compliance.

Demos can be booked at kaizan.ai/demo. The standard demo is 30 minutes and covers a live walkthrough on a sample account, the CARE health model, the redaction pipeline, and pricing. For teams over 100 seats, a tailored demo using anonymised data from your own meeting recorder can be arranged.

Yes. Kaizan is hiring engineers, designers, GTM and customer success roles, primarily in London and remote across UK and EU time zones. Open roles are listed at kaizan.ai/careers. The team is fifteen people as of 2026 and intentionally hiring slowly to preserve quality of work and culture.

Press and research enquiries should go to press@kaizan.ai. The team responds within two working days and can provide product screenshots, data on the CARE benchmarks (anonymised), and access to client references on request.