Platform · Agent Hub
Every agent you run, on one screen.
The Agent Hub is where your agents become visible: who owns each one, what it was told to do, the systems it can reach, what it remembers, and what it costs you day by day. A fleet of agents is only defensible if you can see it.
Book a ConversationThe org chart
An org, not a pile of bots.
Agents and the people accountable for them sit in one structure, grouped by function. When your CEO asks who owns the agent that emailed a client, the answer is on the screen.
A named human owner
Every agent reports to a person, not to a team inbox. Accountability for what an agent does sits with someone who can answer for it.
Version and status
Each agent carries the version it is running and whether it is live, so a change to how an agent behaves is a visible event.
Delegation in the open
Agents that hand work to other agents show those lines separately from the reporting line, so a coordinator's output traces back to the agent that produced it.
Inside an agent
Open any agent and read it.
Every agent opens into its own record. How it behaves is never hidden in a configuration file your team cannot see.
System prompt
The instructions the agent is running under, as published, not as documented six months ago.
Skills
The procedures attached to it, each one a readable document rather than a black box.
Tools and systems
Everything the agent can reach, built-in and connected, listed and counted.
Memory
What it has learned and retained between conversations, browsable entry by entry.
What it costs
The second question every CIO asks.
The first is whether it will work. The second is what happens to the bill once thirty agents are running. The hub tracks spend by agent, day by day, so the answer is a number you watch rather than a surprise at the end of the quarter.
Spend by agent
Every day is broken down by the agent that spent it, so one agent cannot quietly run away with the budget.
Scheduled against interactive
Work an agent does on a schedule is separated from work people asked for. Scheduled spend is the part that grows without anyone requesting it.
Cost against outcome
A per-agent figure sits next to what that agent delivered, so the question of whether it earns its place has a number attached.
Spend is priced from token usage as it happens, so you see the shape of the bill during the month rather than after it.
Control
Governance built into the surface,
not bolted on after.
Live state, not a stale config
The prompt, tools and version you read are pulled from the running agent, not copied into a document that drifts.
Reach you approve
Each agent's channels and connected systems are listed and scoped. It reaches what you agreed it reaches, and the list is on the page.
External contact is a decision
Agents able to send email or message outside the company sit in an explicit tier, so outbound contact is deliberate rather than a side effect.
Runs you can audit
Every run is recorded, including scheduled runs that fail before they start, which is the failure that otherwise leaves no trace at all.
AI-native
We run our own company on it.
Kowalah's agents across sales, delivery, resourcing, operations and finance run in this hub every working day. We built the cost view because we needed to answer the same question you are asking, about our own fleet, before asking you to trust it with yours.
Built under load
A working fleet runs in this hub every day, so what it shows is what an operator needed to see.
Drift is checked, not assumed
Published agents are compared against their source, so what you read is what is running.
The same surface you get
Your hub carries the same record for every agent, not a cut-down summary.
Your own hub
Your agents, in a hub of your own.
Your people reach their agents at your hub, signed in with the identity they already use. It carries the same record our own team works with: chat with the agent, read its prompt, see its skills and tools, review what it remembers, and collect what it produced. Access is closed by default and granted person by person.
Managed by Kowalah
We run your agents in a hub we operate. Your team signs in with the identity they already use for Kowalah and reaches their agents on day one, with no infrastructure to stand up.
Your account, your costs
Bring your own Anthropic account when you want to hold the contract directly. Usage is metered per agent and attributed to your organisation either way.
Your own deployment
As the programme matures, your tenant lifts out into a hub of your own: your domain, your branding, your channels, your data boundary.
Ready when you are
See the hub
on your own agents.
Thirty minutes is enough to walk the hub and show what it looks like with your agents in it.