Your AI Product Marketing Team
Loading agent statusβ¦
Run History
All scheduled and manual agent runs
| Agent | Started | Status | Hours Saved |
|---|---|---|---|
Showing last 25 runs. All agent output is in Mattermost.
Competitor Registry
Loadingβ¦
By Category
Manage Competitors
Add Competitor
Hours Saved
Time reclaimed from manual research and repetitive analysis
Hours by Agent (All Time)
Weekly Trend (Stacked)
Dynamic Hours Model
| Multiplier | Classification | Description |
|---|---|---|
| 0.25Γ | Light | Simple lookups, single-source checks, formatting tasks |
| 1.0Γ | Normal | Standard multi-source research and synthesis |
| 2.0Γ | Heavy | Cross-referenced analysis requiring expert judgment |
Dollar value uses a blended rate of $150/hr. Multiplier is set per run at log time.
AI Usage & Cost
Token counts come straight from each API response. Dollar figures are estimates from a maintained price table, not actual billing β check your provider's own billing page for exact spend.
Settings
One rate for the whole team, not per agent. Multiplied by all-time total hours saved across every agent combined to produce the "Est. Dollar Value" figure on this tab. Think of it as "what would it cost, at this blended rate, to have a human do everything the agents have done so far."
Danger Zone
Permanently clears all run history β this resets every number on this tab and on
Run History to zero, and clears each agent's "last run" status on the Roster tab.
A backup table is created in SQLite first (agent_runs_backup_<timestamp>),
so it's recoverable via direct DB access, but not from the dashboard itself.
Type RESET to confirm.
Status
Live health of every service this dashboard depends on
Health checks run on a 30-second cache. The dashboard does not modify any system state.
To restart the scheduler: systemctl restart pmm-scheduler
Settings
Configure agents, edit prompts, manage sources, and monitor system health
Company Profile
This feeds every agent's default prompt β company name, industry, and topics are woven into what they search for and how they describe themselves. Edit this instead of any Python file when the app changes companies.
AI Providers
The API key(s) agents use to call an AI model. Editing here takes effect on each agent's next run β no server restart needed.
Edit Agent
οΌ Add New Agent
βΌ ShowCommunity Pulse β Run From Your Computer
Community Pulse (Reddit + Hacker News monitoring) needs to run from a personal computer, not this server β Reddit and Hacker News both rate-limit cloud servers much harder than home internet connections. Download this self-contained package for anyone who needs to run it: no coding knowledge required, just double-click a setup file and follow the prompts.
β¬ Download Home Runner (.zip)They'll also need this dashboard's web address and a home-runner API key β ask whoever manages the server for both. Full setup instructions are included in the download (README.md).