Grounded by default
Every AI call is prepended with your company profile, playbooks and knowledge base — so output is on-brand and accurate, not the open web’s average.
The operating model
How dolv works to execute and measure marketing
dolv is a closed loop, not a chatbot. It grounds in your company, executes real work across your stack through one-tap human approval, and measures every outcome in one funnel — so insight turns into action, and action into a measured result.
Connect → Execute → Measure · grounded · human-in-the-loop.
What makes the loop work
Six moves, one operating loop
dolv isn’t a smarter chatbot — it’s an operator. Each piece below grounds in your business, does real work, and feeds a single measured funnel.
25+ tools that execute
The command center doesn’t stop at advice. It calls real tools to draft content, build campaigns, create tasks and read live integration data.
Human-in-the-loop
Anything public or irreversible is prepared, never auto-shipped. A person approves; only then does the real provider call run and get recorded.
Measured in one funnel
On approval, the linked task auto-closes and the result lands in a unified TOFU/MOFU/BOFU funnel, scored against a rolling 30-day baseline.
Multi-agent campaigns
A Director orchestrates member agents in parallel or sequential handoff, then synthesizes one approval-ready deliverable toward your OKR.
Close the loop
A correlation engine and multi-touch attribution turn measured outcomes into the next move — keep what works, cut what doesn’t, compound every cycle.
25 + Callable tools that execute real work
20 Read + write integrations, not just reports
4 Approval states: prepared → approved → executing → done
The three phases
Connect, execute, measure
One closed loop takes you from company context to a measured outcome. These three phases frame the whole product — and each one feeds the next.
- 01 Connect
Ground dolv in your business
Link your read-and-write integrations and point dolv at your site, docs and data. From then on every reply is grounded by default in your company, playbooks and knowledge — not the open internet.
- 02 Execute
An AI command center that does the work
A grounded chat assistant calls 25+ real tools to draft, build, schedule and pull live data — instead of stopping at a suggestion. Internal moves run now; external moves are only prepared.
- 03 Measure
Every outcome lands in one funnel
Approve a prepared action and the linked task auto-closes, the provider call fires, and the result is scored in a unified funnel against a rolling 30-day baseline — so the next cycle is sharper.
Go deeper on the close-the-loop model in grounded AI, funnel intelligence and AI agents.
Execute, safely
Internal actions run now. External actions wait for you.
The rule is simple, and it’s the heart of human-in-the-loop AI marketing. Reversible, internal actions execute immediately. Public or irreversible actions are only prepared — queued in the approval inbox until a person signs off, runs the real provider call, and records the result. Nothing leaves the building without you.
That same approval hand-off is what fires the measure layer: approve a prepared action and its linked task auto-closes, so the work and its outcome are written back together.
Measure
One unified funnel, one health score
Live metrics from your integrations map onto three stages, score against a rolling 30-day baseline, and roll up into one weighted composite — so you read the whole picture and fix the single bottleneck next.
weight 0.25
TOFU
Awareness — reach, traffic and discovery at the top.
weight 0.40
MOFU
Consideration — engagement and intent in the middle.
weight 0.35
BOFU
Decision — pipeline and conversion at the bottom.
The three roll up into one composite health score weighted TOFU 0.25, MOFU 0.40 and BOFU 0.35. On top, a cross-metric correlation engine and multi-touch attribution (five models) tie measured outcomes back to the moves that caused them — the signal that compounds every cycle.
Who it’s for
For lean teams that own the outcome
If one or two people are responsible for pipeline, you can’t afford insight that stops at a slide. dolv gives a lean team the leverage of a stack: a grounded command center, configurable agents with a $250/mo cap and a full run-history log, a full CRM with intent-signal lead scoring, plus a North Star, OKRs and ICE experiments scored with a z-test.
You stay on the decisions that matter; dolv does the busywork and proves what worked.
FAQ
How the execute-and-measure model works
The loop, the approval gate, and exactly where you stay in control.
What does "execute and measure marketing" mean in dolv?
Execute and measure is the closed loop at the heart of dolv: a grounded AI command center does real work through 25+ callable tools, routes external actions through human approval, then records every outcome in one unified funnel. Insight becomes action, and action becomes a measured result — instead of dying in a doc.
Can the AI run marketing campaigns by itself?
dolv runs multi-agent campaigns where a Director orchestrates member agents in parallel or sequential handoff, but it is human-in-the-loop by design. The team synthesizes one approval-ready deliverable; a person approves anything that ships externally. You set the outcome, dolv does the busywork, and you keep control of the moves that matter.
Do AI actions need human approval before they go out?
Public or irreversible actions do. The four prepare_* tools — prepare_email, prepare_outlook_email, prepare_linkedin_post and prepare_wordpress_post — are queued in the approval inbox and move through prepared, approved, executing and done. A human clicks approve, the real provider call runs, and the result is recorded. Internal, reversible actions run immediately.
How does dolv measure marketing outcomes?
When an action is approved, the linked task auto-closes and the measure layer fires. Live metrics from your integrations map onto TOFU, MOFU and BOFU stages, score against a rolling 30-day baseline, and roll up into one composite health score weighted TOFU 0.25, MOFU 0.40 and BOFU 0.35 — surfacing the bottleneck stage and the next move.
What is the difference between internal actions and approved actions?
Internal actions are reversible and stay inside your workspace — create_task, create_campaign, schedule_calendar_event, create_google_doc — so they run immediately. Approved actions are public or irreversible, like sending an email or publishing a post, so they wait in the approval inbox until a human signs off. The line is simple: nothing leaves the building without you.
See the loop run on your stack.
Connect your tools, point dolv at the work, and watch it execute — grounded in your business, measured in one funnel, with you in the loop. Don’t just plan it. dolv it.