--:--:-----...REALREAL + DEMODEMO ONLY🧭 TEAM STATUSSIGN INLOCKED⏳0APPROVALSv2026.10.22⬢ AGERON
−
100%
+
⧉ ADD EXISTING AGENT
Link an agent that already exists in the Dashboard to this team. It stays a reference — rename / configure / delete it in the Dashboard; here you just organize (and it can live on multiple teams).
QUICK STATUS / LOCATION
Applies to the person in view. (Login-scoped "me / my reports" comes with auth.)
LOCATION
STATUS
AGERON CORE NEXUS
Per-organization workforce view. Configure your company, departments, humans, and their teams of bots. Everything saves to your browser locally — start by setting your company name, then add departments and people.
First time? "Load Demo Data" gives you a fully-populated example (My Trusted Insurance Agency, Personal Lines department with 6 humans + their bot teams) to explore the layout.
SETTINGS
Company
Saved to the shared fleet doc — agents inherit it (agent > team > company cascade) and it heads every goal selector.
Standing limits every department, team and agent inherits. Each row says what the limit does and asks for the right kind of value. Quiet hours, jobs per day, jobs per hour and spend per day are walls the relay itself enforces when an agent asks for work; every other limit is delivered to the agent to honour.
Pick these from the notify fields instead of retyping — the department column groups the pick-list. Saved to the shared fleet doc, so every machine running Nexus sees the same book.
Access & lockdown
Connect a relay to see who may do what.
Approvals — who is told, and when
Connect a relay to see the approval ladder.
Presence connector keys
Connect a relay to see the presence connector keys.
Ninety (EOS) — optional
Not everyone has Ninety: rocks and KPIs typed on a person's card are the normal way and always work. With a connection, people are matched by the e-mail on their card and their rocks, to-dos, issues and scorecard numbers are pulled in beside the typed ones, marked from Ninety. The key (a Personal Access Token from Ninety's Developer Settings) goes on the relay host in .env as RELAY_NINETY_TOKEN — never here.
Connect a relay to see the Ninety connection.
Optional — Live Polling
If you'd like Nexus to also pull live agent data from an AgeronCORE-Relay, set the URL and bearer token here. Leave blank to use the manual data you've configured.
v2026.5.16.1: when set, Nexus shows only agents in this company. Can also be passed as URL param: ?company=smart-services
v2026.10.1: with the company key this Nexus sees the whole company (sign in to act as yourself). Leave it blank and name the company below: people then sign in, and each one sees only what their role gives them — the relay cuts it, not this page.
Approving / rejecting held tasks calls the relay's /admin endpoints. The token is stored ONLY in this browser (localStorage), is never exported with the config, and is never shown back — leave blank to keep the saved one.
Connect the relay to see what this company is holding.
Backup / Restore
One file with everything: this Nexus's org and settings and — with Everything picked — all the relay keeps for your company: memories and knowledge, documents (the files too), guardrails at every level, goals, routes, front doors, job types, library places and the work journal. RESTORE / IMPORT reads it on any machine: the org comes in at once (it REPLACES the org here — back up first), and the relay part is previewed and then put back with your company's own relay token — no admin token needed (relay v2026.9.27 or newer).
No relay is set above, so this backup can only hold the Nexus part (org + settings).
Admin token: not needed to put your company's data back (relay v2026.9.27+ restores with the relay token); only an older relay needs it. Including it makes the file a root credential. Off by default. The relay token (checkbox above) is likewise password-grade — treat it like a password; it stays OUT of the file unless you tick it, every time (the relay URL always rides).
⏳ pulling relay data…
DOCUMENTS
+ NEW DEPARTMENT
for showing Nexus to a prospect: shown in REAL + DEMO and DEMO ONLY, never sent to the relay, and nothing in it touches the real company
Icon
…OR UPLOAD A CUSTOM IMAGE
SIGN IN
+ NEW HUMAN
Identity
Pick a different department to MOVE this person; you'll choose which agents follow.
signed in, may change routes when they are locked to approvers (department heads always may)
decides company truth (company-level knowledge, policy, pricing, security, brand) and is the last step every approval climbs to
An operator can give and change jobs, recurring tasks, queues, approvals and chat for the agents of its teams, and never adds or removes people or agents. A viewer only looks.
KPIs (Scorecard) manual entry — API integrations tune these in later
NAMETARGETCURRENTLINK (optional)
Rocks (custom)
NAMETARGETSTATUSLINK (optional)
Avatar Image
👤
Assigned Agents0
Agents pulled from the relay's unified fleet. Drag from the right pool, or click to add.
ASSIGNED
none yet
AVAILABLE POOL
no unassigned agents
+ NEW TEAM
A key set here replaces the department's and the company's value for this team's agents. daily_tasks_cap, rate_per_hour, spend_cap_usd and quiet_hours (e.g. 22-06, your timezone) are enforced by the relay when an agent asks for work; the rest are the agent's to honour.
Bots that aren't under a lead's control
+ NEW AGENT
Identity
A key set here replaces every inherited value of it for this agent. daily_tasks_cap, rate_per_hour, spend_cap_usd and quiet_hours (e.g. 22-06, your timezone) are enforced by the relay when the agent asks for work; the rest are the agent's to honour.
Lead bot is highlighted in the team box
Status & Activity
Networking (optional — for live polling)
This is the agent's own health-probe token, kept with the agent's record. It is not the relay token: chat and the fleet use ☰ → Settings → Relay Bearer Token (the company token).
Live relay chat (and the Telegram deep-link) in the comms drawer.
Work Budget — relay-enforced
HARD STOP at the cost cap
Avatar Image
🤖
CROP PROFILE IMAGE
ZOOM100%
DRAG TO REPOSITION · SCROLL OR SLIDER TO ZOOM
COMMS
Live relay chat for agents + Telegram deep-links. Microsoft Teams = later.
CONFIRM
NEW QUEUE
id: —
ADVANCED — set the id manually
Pick the team this queue works for — the WORK page groups queues department → team. UNASSIGNED queues collect in their own section.
Tasks enqueued without their own goal inherit this at enqueue time; claiming agents receive the full lineage ("COMPANY GOAL … > ROCK …").
Only agents with a relay agent id can work a queue — bots without one are listed but disabled (link them in the Dashboard first).
TELEGRAM
EMAIL
Notify onPer-task NOTIFY wins over these; relay-level defaults (Dashboard → Relay Config) sit below.
HARD STOP at the cost cap — the queue stops dispatching and shows EXHAUSTED. OFF = soft: alert only, work continues.
The task cap always hard-stops at midnight rollover; the cost cap is soft unless HARD STOP is on.
A job an agent has held longer than this is stopped and marked FAILED (time limit); the queue moves on. A task can carry its own limit, which wins over this one.
Assigned agents start working this queue automatically on their next check-in (≈1 min). Turn off to manage agent configs manually.
When ON, every new task HOLDS as AWAITING APPROVAL until an operator approves it (approvals inbox ships next slice).
CREATE TASK
For people only: the agent never sees the name. The instruction below is what the agent works from.
The relay keeps a copy with the task; the agent gets them on its machine in a folder for the job, and what it saves in that folder's out comes back here.
Each THEN step spawns when its parent completes; write {{result}} in a step's instruction to inject the parent's result text (first 4KB). Up to 3 steps here; the relay caps chains at depth 5.
A recurring task that needs approval is approved once: it asks when you make it, and its runs then go straight through. Changing what it does, where it goes or when it runs asks again before the next run. Until it is approved it does not run.
TELEGRAM
EMAIL
TEAMS — coming soon
Notify on
VERBOSE LOG — record the turn's internals to the progress log (troubleshooting; needs reporter v2026.9.2+)
Repeat on a schedule
RECURRING — this creates a repeating schedule on the relay (not a one-off task). Each occurrence spawns a fresh task; a still-open previous occurrence is skipped, never stacked. The Name at the top names the recurring task and every run it makes.
Every
Atoperator-local time
MONTUEWEDTHUFRISATSUN
Source is recorded automatically as nexus.
TASK
APPROVALS INBOX
RESTORE RELAY DATA
WEBHOOK — SHOWN ONCE
NEW WORKFLOW
id: —
Write {{result}} in a step to inject the previous step's result text; {{param:key}} asks the operator for a value at run time.