How does an investor turn a PitchBook query into an analyst-attributed comp set?
The agent runs the PitchBook screen, drafts a comp set in a Dock row, and pastes a pointer back to each PitchBook company profile. A named analyst reviews the draft, adjusts weights, and signs the row. PitchBook and CB Insights stay the system of record for raw deal data. Dock is the system of record for the interpretation: which firms were included, who decided, and why. That separation is what makes a comp set defensible six months later when the IC asks who picked the peers. See Dock for investors for the full pattern.
Architecture
PitchBook and CB Insights stay the system of record for the raw data. Dock is the system of record for what the agent interprets. Each Dock row carries a pointer back to the platform record, agent identity, decision, reviewer, and timestamp. The agent re-fetches platform data via fresh API reads when it needs current state.
The Dock surface: comp-set worksheet
One sheet per target. Each row is a candidate peer the agent proposes, with the analyst signing off.
| Peer | PitchBook ID | Why included (agent) | Multiple pulled | Analyst | Status |
|---|---|---|---|---|---|
| Ramp | pb:cmp:104882 | Spend-mgmt, Series E, $13B post, same ICP cluster | EV/Rev 14.2x (2025 round) | s.chen | Approved |
| Brex | pb:cmp:091245 | Spend-mgmt + banking, later stage, mixed ICP | EV/Rev 9.8x (secondary) | s.chen | Excluded, off-thesis |
| Mercury | pb:cmp:118903 | Banking-first, adjacent not direct peer | n/a (private) | s.chen | Watchlist |
The PitchBook ID column is the pointer. Open the row and the agent re-fetches the latest profile through the PitchBook API before showing the analyst the multiples. No stale cache.
Worked workflow: drafting a fintech comp set
A junior analyst kicks off a screen: "spend management, US, Series C+, last round within 18 months." The agent runs the PitchBook query, returns 23 hits, and proposes a 7-name comp set in a new Dock sheet. For each row, it writes a one-sentence rationale, pulls the most recent multiple, and attaches the PitchBook profile pointer. The senior analyst opens the sheet, drops two names she considers off-thesis, adds one CB Insights-sourced private peer the screen missed, and signs each remaining row. The signed sheet becomes the artifact the IC reviews. Six weeks later when a partner asks why Mercury was excluded, the row shows the agent's draft, the analyst's edit, and the timestamp. The full audit trail pattern is in agent audit and compliance.
Why it matters
Comp sets are the soft underbelly of private-market research. The data is rigorous; the peer selection is judgment. When that judgment lives in a Google Doc with no attribution, the IC has to take it on faith. When it lives in a Dock row with a named analyst and a pointer back to PitchBook, the judgment is reviewable. CB Insights reports it covers 12M+ company profiles across 1,600+ markets (CB Insights), and PitchBook coverage is similarly deep (CB Insights company page). The bottleneck is no longer data access. It is who decided which subset matters, and Dock answers that. See also Dock for research and the founder mirror at Dock for founders.
Try it on your next screen
Pick one live comp set. Run the PitchBook screen, let the agent draft rows in Dock, have an analyst sign each one. Compare against your current process at week six when someone asks why a peer was in or out. Start at Dock for research.
FAQ
Does Dock replace PitchBook? No. PitchBook stays the source of deal data and multiples. Dock holds the agent's interpretation and the analyst's sign-off, pointing back to each PitchBook profile.
What about CB Insights for private-only coverage? Same pattern. CB Insights rows live alongside PitchBook rows in the same Dock sheet, each with its own platform pointer. The agent re-fetches both at row open.
Who is the analyst of record if the agent drafted the row? The human reviewer who signed it. Agent identity is captured separately as the drafter, never the approver. See agent audit and compliance.
Does the agent ever auto-include a peer? No. The agent drafts and proposes. A named analyst must sign. Approval is a deliberate gate, not a default.
