Walk into any modern sales org and count the agents. SDR autoresponders qualifying inbound. AE assistants summarizing calls. Ops bots reconciling pipeline against the warehouse. Every one of them logged in as a human rep, leaving an audit trail that says the rep did the work. The rep didn't. The agent did, on a credential the rep handed over because there was no other way.
That's the sales shape of a problem we've covered across the Dock blog: the work is real, but the substrate treats it as a session attached to a person. Sales feels it first because sales runs the most agents on the highest-stakes data.
The shape of sales work that fits Dock
Sales has two surfaces. The pipeline is a table: rows are opportunities, columns are stage, value, owner, next step, close date. The account brief is a doc: who the buyer is, what they care about, the calls summarized, the proposal in progress.
Today those surfaces live in different tools, and agents act on each through APIs wearing the rep's identity. Dock collapses that. One workspace holds the pipeline table and the account doc as two tabs of the same room. Reps are members. Agents are members too, with their own keys and their own attribution on every write. Same architecture as the agent collaboration primer: identity, authorization, shared surface.
One named account, one rep, two agents
Concretely. A workspace named "Acme · Q3 expansion." Priya, the rep, owns it. Two agents are members alongside her: an SDR that drafts outbound, and a call-summary agent that runs on her transcripts.
Monday: Priya edits the pipeline row. Stage moves from "discovery" to "proposal." Close date pulls forward two weeks. The log records Priya.
Tuesday: the SDR appends a section to the account doc, a three-paragraph draft for the new CFO on the buying committee, headed "Draft outreach · CFO," attributed to the agent. Wednesday morning Priya reads it in the same workspace, edits two sentences, sends. Her edits show as hers. The draft still shows as the agent's.
Wednesday afternoon: the call-summary agent runs after Priya's hour with Acme's VP Eng. It updates the pipeline row, adds line items to a Stakeholders column, bumps the confidence score, drops a 200-word summary into the doc. Three writes, all attributed to the agent.
Friday, the manager opens the workspace and sees a coherent record. No one had to ask who did what.
Why this matters for sales specifically
Rep credibility. Reps build their book on relationships and judgment. When an agent sends outreach under the rep's name and something lands wrong, the rep wears it. Distinct attribution lets the rep own what they wrote and the agent own what it drafted. The principal model is in agents are principals.
Manager visibility. Pipeline coaching depends on seeing the work, not just the outcome. If every action looks like the rep did it, the manager can't tell whether the rep is thinking or the agent is. Attributed edits turn a 1:1 into a real conversation about leverage.
Audit defensibility. Sales orgs ship regulated emails, store PII, attest to outreach controls. When the agent acts as the rep, the log is fiction. When the agent acts as itself, it's true. See agent audit and compliance.
Salesforce's State of Sales research finds reps spend less than 30% of their time actually selling. Gartner's 2024 seller survey adds that half of sellers feel overwhelmed by their tech, and overwhelmed sellers are 45% less likely to hit quota. Bolting more siloed agents onto that stack makes it worse. Putting the rep and the agents in the same room is the move.
Try it
Dock is in invite-only beta. If you already have agents in the loop, setup is fifteen minutes: one account, one pipeline table, one brief, your rep, one agent. See the attribution shape for yourself. Full pitch at Dock for Sales.