Founder ops is three recurring documents: the board deck, the investor update, the roadmap memo. AI can draft each one. The problem is the rhythm. Last month's edits, last quarter's framing, the named recipient list, all of it has to survive into next cycle. Dock for founders is the workspace where that interpretive layer lives, so the agent drafting in May remembers what the founder changed in April.
Notion, Pitch, Google Workspace, Linear, and Carta stay the system of record for the raw founder data: cap table, slide files, docs, issue tracker, financial model. Dock is the system of record for what the agent interprets from that data. The board narrative, the draft update, the reviewer's sign-off, the audit trail. Each Dock row carries a pointer back to the platform record (notion_page_id, pitch_presentation_id), the agent identity, the founder's decision, the reviewer, and the timestamp. The agent re-fetches Carta or Linear data via fresh API reads when it needs current state. Dock holds the persistent interpretive layer that survives across sessions.
One Dock surface: the founder-comms table
| row_id | doc_type | source_pointer | agent_draft | founder_edit | recipient_list | status | reviewer |
|---|---|---|---|---|---|---|---|
| FC-204 | board_deck_q2 | pitch_presentation_id:9f2a | "ARR +38% QoQ, burn multiple 1.4" | "Reframe burn around hiring plan" | board@, observers | sent | govind |
| FC-205 | investor_update_may | notion_page_id:c7b1 | "May highlights, May risks, May asks" | "Drop the asks section, too early" | 14 named LPs | sent | govind |
| FC-206 | roadmap_memo_h2 | linear_cycle_id:cyc_88 | "H2 themes from open Linear epics" | "Move infra theme above growth" | exec team, lead investor | in_review | govind |
Three rows, three rhythms. Board deck quarterly, investor update monthly, roadmap memo per planning round. Same table, same pointer schema, same audit fields.
Worked example: the May investor update
The agent reads last month's update (row FC-198), pulls current ARR and burn from Carta and the financial model in Google Workspace, pulls shipped issues from Linear, and drafts a structured update into row FC-205 with a notion_page_id pointer. The founder edits the asks section out and approves. The consent gate fires: Dock confirms the 14 named recipients against last month's list, flags one new LP added since April, and waits for confirmation. The update goes out, the row closes with reviewer govind and a timestamped audit entry. Next month, the agent re-reads FC-205 first.
Why it matters
Board prep and investor comms are the work most often offloaded to AI and most often re-done from scratch. Without a persistent interpretive layer, every cycle starts cold. The deck reverts to a template. The update repeats last month's framing.
Founders carry institutional memory in their heads. That does not scale past Series A. The NVCA's Board Service Excellence program treats governance as a structured practice with cadence and continuity (NVCA Resources). Bessemer's State of the Cloud benchmarks show how tightly investors track quarterly CAC payback, NRR, and burn multiple (Bessemer State of the Cloud 2023). First Round Review documents the recurring patterns founders rely on to scale operating rhythms (First Round Review). The agent needs a place to remember.
Read the architectural thesis in Cloud 2.0 for product, or sign up to give your founder agent a memory.
FAQ
How is this different from a Notion AI block inside the doc? Notion AI lives inside one page. The Dock row lives across pages, across cycles, and across agents. The pattern recurs in Dock for sales and Dock for research: the platform stays the surface, Dock holds what the agent decided.
What stops the agent from sending to the wrong list? The consent gate. The recipient list is a field on the row, the agent proposes, the founder confirms. The agent collaboration primer covers the handoff.
Who is the agent, and how do I trust the draft? Every row carries the producing agent's identity, scoped credentials, and audit trail. See agent identity.
Does this satisfy board governance review? The reviewer field, timestamp, and source pointer give a structured record. Agent audit and compliance covers what a reviewer can reconstruct.
