Skip to content
wwwdot.dev
Try live agents
2 AGENT LIVE · 17 TOOLS

AI systems you can
actually try.

Explore working AI agents built to solve real business problems. Watch each one reason, retrieve, call tools and commit an outcome — live, in front of you.

Sales EngineerSESSION REPLAY
Awaiting request
Live activity
Understanding01
Planning02
Knowledge search03
Tool execution04
Decision05
Action06
Result07
Run this agent yourself
01The difference

Most AI agencies show you a deck.

We ship the system and let you use it. Everything on this site runs on one engine — the same conversation loop, tool layer, retrieval and event stream we deploy into client businesses. This demo is itself just another tenant on that platform.

The activity panel is not a decoration. It is the actual event log: every stage transition, every retrieval, every tool call and its duration, streamed from the server as the agent works. If a tool fails, you see it fail.

That is the whole argument. An agency that can show you the inside of its system is an agency that built one.

02Live agent studio

Six agent categories. One engine underneath.

Open the studio
Conversational agent builder

Agent Factory

LIVE

Tell it what you want built. It works out the agent type, designs it with you, and produces a specification you approve before anything is created.

Blueprints
6
Output
Specification
Deploy
Studio
Try live
Qualification & solution design

Sales Engineer

LIVE

Qualifies a business problem, grounds every claim in retrieval, and designs a concrete system on the call.

Tools
7
Knowledge
12 docs
Captures
Leads
Try live
Deflection & escalation

Customer Support

PREVIEW

Answers from your documentation only, resolves what it can, and escalates with full context when it can't.

Grounding
Docs only
Escalation
Context-preserving
Systems
Order lookup
In build — template ready
Planning & synthesis

Research Analyst

PLANNED

Plans a research approach, gathers sources, extracts what matters and produces a structured report.

Output
Cited report
Method
Plan-first
Sources
Tracked
Architecture ready
Internal knowledge & reporting

Internal Operations

PLANNED

A private assistant over company data — reporting, policy lookup and internal tooling for your team.

Access
Role-scoped
Data
Internal
Audit
Full trail
Architecture ready
Triggers & multi-step execution

Workflow Automation

PLANNED

Triggered, multi-step execution with conditional logic, API calls and approval gates on anything irreversible.

Trigger
Event / cron
Logic
Conditional
Safety
Approval gates
Architecture ready
Product, accounts & billing

Custom AI SaaS

PLANNED

A complete product: customer accounts, custom workflows, usage metering and billing architecture.

Tenancy
Multi-tenant
Metering
Usage-based
Ownership
Your codebase
Architecture ready
03The engine

Built once. Configured per agent.

An agent here is data, not code: instructions, a tool allow-list, a knowledge base and a pipeline shape. Adding the Support agent means adding a configuration object — the conversation loop, streaming, event protocol, persistence and visual system are already there.

The same holds one level up: every client is an organization with its own agents, knowledge, customers, conversations and analytics, isolated at the database by row-level security. Onboarding a client creates rows, not a codebase.

Reusable tools
17
Knowledge documents
14
Event types
17
Tenancy
Row-level isolated
Platform architectureagent-engine
Agent configurations
Sales
Support
Research
Business
Workflow
SaaS
Agent engineshared
Conversation engine
Streaming transport
Event protocol
Tool execution
Knowledge retrieval
Supabase persistence
Claude API
Supabase / pgvector
Next.js runtime
04Capabilities

What is actually running on this page.

Streaming reasoning

Responses stream token by token while the agent's reasoning summary and stage transitions render in parallel. Nothing waits for the turn to finish.

Tool execution

Retrieval, lead capture, scoring and recommendation are real server-side functions with validated inputs, timing, and structured results you can read.

Grounded retrieval

The agent searches a knowledge base before making factual claims, and is instructed to say so when retrieval comes back empty rather than inventing an answer.

Persistence

Conversations, messages, events, tool runs, leads and analytics are written to Supabase Postgres with row-level security and a service-role key confined to the server.

Demo safety

Per-IP rate limits, per-session message ceilings, input length caps, a hard turn timeout and a per-agent tool allow-list. A public visitor cannot reach a privileged tool.

Graceful failure

Upstream errors, timeouts, rate limits and safety declines each get a distinct, readable state with the conversation preserved and a retry available.

Describe your workflow to the agent.

It will tell you where automation pays off, design a system against your actual process, and say so plainly if AI is the wrong tool for it.