Skip to content
Seneca

The home of specialised AI agents.We build an expert agent for each job and bring them together in one shared workspace.Open sourceSelf-hostable

Try the playground
npx @hachej/boring-ui-cli

Open source · Self-host with Docker and Postgres · Or use our managed deployment

Sample workspace, read-only. Sign in to get your own.
What Seneca is
THREE WAYS TO BUILD WITH SENECA

Start with the job. Choose how you want to build.

SENECA PRODUCT

Build and run vertical agents.

Give each job its own agent, tools, knowledge and interface. Bring the whole network together in one workspace.

Explore the product ↓ SENECA FOR CREATORS

Turn an expert archive into a job-specific tool.

Your audience submits real work. The tool applies your framework, cites the lessons and returns a useful outcome.

Explore Seneca for creators → SENECA SERVICES

Have your specialised agents built for you.

We map the job, connect your knowledge and tools, build the agent and help your team put it to work.

Explore Seneca services ↗
01 · Vision

One assistant cannot do every job.

You do not reach the Moon by giving one person every tool. You build a team of specialists. AI should work the same way.

The future is not one assistant trying to do everything. It is a team of experts.
Give each agent one clear job, the right knowledge and only the tools it needs.

One general assistant versus a team of specialised agents One general assistant tries to handle every job. The specialist approach gives five expert agents a clear role and the right model for their work. ONE GENERAL ASSISTANT One model EVERY JOB A TEAM OF SPECIALISTS REVENUE LEGAL SUPPORT RESEARCH OPS The right model for each expert FRONTIER · OPEN SOURCE · YOUR ENDPOINT One general assistant versus a team of specialised agents One general assistant tries to handle every job. The specialist approach gives each expert agent a clear role and the right model for its work. ONE GENERAL ASSISTANT One modelEVERY JOB A TEAM OF SPECIALISTS REVENUE LEGAL SUPPORT RESEARCH OPS The right model for each expertFRONTIER · OPEN SOURCE · YOUR ENDPOINT
One general modelFive bounded agents
The problem

One assistant, every job

Give one assistant every tool and instruction, and it becomes expensive, difficult to control, and worse at finding what matters.

The response

An expert for each job

Each agent knows its role, uses only the tools it needs, and works with the rest of the team.

The outcome

Better work, clearer control

Choose the right model for each expert instead of paying one model to do everything.

02 · Vertical · Your agent

Built for your vertical.

A vertical agent is software designed around one real job, not a general assistant with a long prompt.

You do not pick one off a shelf.
Describe the job. Wire its sources. Set the rules it runs under. Then it is yours — prompt, tools, UI and all.

The six parts of a vertical agentOne agent assembled from instructions, skills, tools, a knowledge base, computer access and its own user interface. One agent 01Instructions 02Skills 03Tools 04Knowledge base 05Computer access 06Its own UI The six parts of a vertical agentOne agent assembled from instructions, skills, tools, a knowledge base, computer access and its own user interface. 01Instructions 02Skills 03Tools One agent 04Knowledge base 05Computer access 06Its own UI
One bounded jobSix configurable parts
Start your first agentOr get it built by our team ↗
03 · Vertical · The workspace

A workspace, not a chatbox.

Agents do not live in tabs. They share work, context, and a computer.

The workspace is where the network comes together.
One place to assign work, review output, and step in when an agent needs you.

Four specialist agents connected to one shared workspaceFour agents connect to one workspace containing shared tasks, automations, an inbox and a cloud computer. Revenue analystChurn specialistReport writerReview agent One workspace TasksAutomationsInboxComputer Four specialist agents connected to one shared workspaceFour agents connect to one workspace containing shared tasks, automations, an inbox and a cloud computer. Revenue analystChurn specialistReport writerReview agent One workspace TasksAutomationsInboxComputer TASK → AGENT ON COMPUTER → INBOX
Four specialist agentsOne shared operating plane
Tasks

Who is working on what

Hand work to an agent, track progress, and follow it to done.

Automations

Your agents, on a schedule

Run any agent on a fixed schedule or trigger.

Inbox

Everything that needs you

Questions, approvals, and finished runs from every agent.

Computer

A computer in the cloud

Each agent has a real machine. Anything your computer can do, it can do there.

04 · Control · Models and infrastructure

Choose how every agent runs.

Use our managed service or deploy the same stack on your own infrastructure.

The right model depends on the job.
Choose a frontier model for difficult work, a lower-cost open model for routine work, or your own endpoint.

Choose where Seneca and its models runYour browser or CLI connects to the Seneca workspace. The workspace sends each expert's work to the model endpoint selected for that job. YOUR DEPLOYMENT YouBROWSER OR CLI Seneca workspaceMANAGED OR SELF-HOSTED Your chosen modelSELECTED FOR EACH JOB Choose where Seneca and its models runYour browser or CLI connects to the Seneca workspace. The workspace sends each expert's work to the model endpoint selected for that job. CHOOSE HOW EVERY AGENT RUNS YouBROWSER OR CLI YOUR DEPLOYMENT Seneca workspaceMANAGED OR SELF-HOSTED Your chosen modelSELECTED FOR EACH JOB
Managed when you want itSelf-hosted when you need it
Managed

Ready when you are

Open the workspace and start building without managing servers or model endpoints.

Models

The right model for each job

Give every expert the level of reasoning, speed and cost its work requires.

Self-hosted

Your cloud, your rules

Docker, Postgres, and a volume. Deploy on your VPS, Kubernetes, or bare metal.

Changing a model is a configuration choice, not a migration. Your agents, tools and workspace stay the same.

05 · Control · Enterprise

Run Seneca inside your infra.

For teams that cannot send work to a shared SaaS — or simply do not want to.

Your identity provider. Your models. Your data.
We install Seneca with you, configure it around your policies, and hand it over.

  • Self-hosted install — your cloud or on-premise
  • Per-user model access and monthly budgets, enforced server-side
  • Shared company context with a policy you control
  • Custom agents for your workflows
Start free Book a call ↗ Read the source ↗