Kosa vs. Affinity: Same Problem, Different Architectures

Affinity is the most widely deployed CRM in venture capital, and the parts it does well are worth naming plainly:

• Relationship strength scoring, warm introduction paths, and network mapping across the firm's collective contacts
• Automatic capture of email and calendar metadata: who emailed whom, how often, and how recently
• Contact and company enrichment from dozens of data sources
• Pipeline management with kanban-style deal views
• A long deployment record at firms including Kleiner Perkins, Bain Capital, and Bessemer
• AI features of its own, including deal assistance and summaries, plus a public API and integration ecosystem

Published pricing discussions put Affinity in the range of $2,000 to $2,700 per user per year.

Affinity captures contact metadata, not substance. It knows an email happened; it does not record what the email said.

• When a founder emails portfolio updates, Affinity logs the message. The ARR, headcount, and runway figures inside it stay unstructured.
• When you leave a board meeting, Affinity logs the calendar event. What was discussed and decided is not captured.
• When a term sheet arrives, Affinity logs the attachment. The deal terms inside it still need a human to type them somewhere.

The result is a familiar split: relationship data stays current automatically, while operational data still depends on manual entry. At most firms that manual entry loses to real work, which is why so many venture records show a majority of fields empty six months after rollout.

Kosa reads the communication itself. It extracts people, companies, deal terms, and portfolio metrics from email threads; processes meeting transcripts into structured records; parses board decks, term sheets, and fund documents into fields; and keeps all of it current without anyone entering data. Humans review the small fraction of extractions that need judgment, and corrections teach the system your firm's conventions.

Affinity is an input system with intelligence on top: humans and metadata feed it, AI enriches and scores what is there. Kosa is a capture system: AI extracts data from existing communication, and humans verify. This is not a feature gap that a release closes; it is a design philosophy difference. An input system will always need someone to fill in the deal terms. A capture system never does, because extraction is the product. Affinity has added AI features on top of its input architecture; Kosa is built on extraction from the ground up, and it builds the relationship layer from the same communication, without the logging.

The shortest version: Affinity captures who you know. Kosa captures what you know.

• Breadth of third-party enrichment data built up over a decade
• A long deployment record and a mature integration ecosystem
• Your firm's workflow is built around it and the switching cost outweighs the data-entry tax for now

• Your firm's knowledge lives in email, meetings, and documents, not in filled-in fields
• Nobody has time for data entry and the record is always stale
• You want meeting prep that assembles itself
• You want institutional memory that compounds over time and survives departures
• You want structured records without behavior change
• You want the relationship graph built from real communication instead of logged contacts

You can, and firms mid-transition do. But they solve the same problem, keeping the firm's knowledge in one trusted place, and Kosa is built to be that place on its own: it captures the operational record and the relationship graph from the same communication. Run both while you switch. Keep both only if switching costs demand it.