HOMEOPENCLAW
> Personal Agents · Open by design · Built in Melbourne
Pulse.AI
OpenClaw™ framework

OPEN
CLAW.

The personal agent that works for you.
Open. Sovereign. Yours.

A custom AI agent built around your workflow. Reads your inbox, owns your calendar, researches in your voice, drafts on your behalf, and stops when you tell it to. Built on open weights and open-source runtimes — your data, your hardware, no vendor lock-in.

READS YOUR INBOX
Triage, summarise, draft
OWNS YOUR CALENDAR
Books, reschedules, defends
RESEARCHES + ACTS
Public web, your tools, your voice
2 build slots open this quarter|14-day delivery on Personal tier|Application-only on Enterprise
Gmail
Calendar
Notion
Slack
Sheets
CRM
Voice
Web
Files
Auth
Vault
Search
↑ tools we hand-code for your agent
openclaw — nikolaos@pulse
> openclaw/init
24/7
always-on
~$0.05
per task
100%
auditable
WHY NOW

The window is open.

Personal agents went from research demo to production tooling in 18 months. Four things shifted at once — and they don’t close again.

Open weights caught up
Llama 4, Qwen 3, DeepSeek and friends now match closed frontier models on agentic tasks. You no longer need to rent a black box.
Inference cost cratered
Cost per agent action dropped ~90% in 12 months. What was a $5/run novelty is now a $0.05 routine.
Tool-use is solved
Function-calling, MCP, the whole agent harness — these are commodity now. The hard part is what you wire it to.
Sovereignty is non-negotiable
Regulated sectors and serious operators want their data on their hardware. The market knows this. Vendors are scrambling.
WHAT WE ACTUALLY BUILD

We code the tools.
We train the agent.
You own the stack.

OpenClaw isn’t a SaaS rebrand. Every engagement is a real engineering build — hand-coded tools, trained behaviour, governed runtime, branded dashboard, source code in your repo. Here’s the literal deliverable.

CUSTOM TOOL CODE
We hand-code typed tool wrappers around YOUR APIs, databases, and SaaS accounts. Each tool is a Python/TypeScript function the agent can call — tested, documented, owned by you.
TRAINING & TUNING
We feed your past emails, docs, brand voice, and operating playbook into the system prompt + retrieval layer. Then we run hundreds of evals to tune behaviour before you ever go live.
ORCHESTRATION
We wire the agent runtime to your triggers — inbox webhooks, calendar events, cron, voice command, dashboard button. Plus the routing logic that picks which model handles which task.
GOVERNANCE LAYER
Approval queues, action policies, kill-switches, audit logs streaming to your SIEM or Sheet. Built-in by default — you never inherit a black box.
YOUR DASHBOARD
A web UI where you watch the agent work, approve actions, edit memory, and tune prompts. Branded to you. Hosted on your domain if you want.
SOURCE + RUNBOOK
You get the repo, deploy scripts, eval suite, and a runbook your engineer (or our retainer) can use to extend, redeploy, or fork the build.
// tools/draft_reply.ts
export const draftReply = openclaw.tool({
  name: 'draft_reply',
  desc: 'Draft a reply in Nikolaos\'s voice',
  input: z.object({
    threadId: z.string(),
    intent: z.enum(['accept','defer','reject']),
  }),
  run: async ({ threadId, intent }) => {
    const thread = await gmail.read(threadId);
    return await llm.draft({
      voice: VOICE_PROFILE,
      tone: TONE_RULES[intent],
      thread,
    });
  },
});
// policy/governance.yaml
actions:
  draft_reply:           { auto: true,  log: true }
  send_email:            { auto: false, approval: human }
  schedule_meeting:      { auto: false, approval: human }
  update_crm:            { auto: true,  log: true }
  invoice_send:          { auto: false, approval: human }

models:
  default: llama-4-70b-instruct
  hard_reasoning: claude-sonnet-4-7
  budget_per_task_aud: 0.15

memory:
  vector_store: pgvector@your-vps
  retention: 90d
  pii_redact: true
THE LOOP

How OpenClaw agents think.

Every action runs through the same four-step loop. Every step is logged, auditable, and reversible. No black-box reasoning.

01
OBSERVE

The agent watches your sources of truth — inbox, calendar, sheets, drive, CRM, your own files. It builds memory.

02
REASON

On a trigger (new email, scheduled tick, your voice command), it plans. Multi-step, with tool selection and self-critique.

03
ACT

It uses tools — write a draft, book a slot, fire a webhook, query a database, send a Slack — under your policy and approval rules.

04
REPORT

You see exactly what it did, why, and the evidence trail. Approve / reject / edit. The next loop learns from your call.

THE STACK

Six layers. All open.

No proprietary SDKs in the critical path. Every layer is swappable — today’s best model can be replaced by tomorrow’s without rewriting your stack.

LAYER 01
INTERFACE
How you talk to it
Web dashboard, voice (VAPI), Slack, email, or direct CLI. Pick one or all.
LAYER 02
ORCHESTRATION
The agent runtime
Open-source agent harness — task planner, tool router, memory store. No vendor SDK lock-in.
LAYER 03
REASONING
The model layer
Open-weight by default (Llama, Qwen, Mistral). Cloud frontier models optional, swappable per-task.
LAYER 04
TOOLS
What it can actually do
Your APIs, your databases, your inbox, your calendar, your CRM. We wire each one to a typed tool spec.
LAYER 05
MEMORY
What it remembers
Vector store (your hardware) + structured rolling-summary memory. Auto-redacted on user request.
LAYER 06
GOVERNANCE
Guardrails
Policy engine for which actions need human approval. Full audit trail. Kill switch.
THREE TIERS

From solo operator to enterprise fleet.

Same architecture. Different scope. Pick the one that matches where your business is right now — you can always upgrade the build later.

PERSONAL
AUD $4,500
one-off build · 14-day delivery · +$250/mo managed hosting

Your own agent. One operator. Built around your workflow.

Custom agent persona — trained on your writing samples
Inbox triage + draft replies (Gmail / Outlook)
Calendar defence + auto-booking
4 custom tools hand-coded for your stack (Sheets, Drive, Notion, Slack, your CRM)
Hosted by Pulse OR on your own VPS — your call
30 days of tuning + new-tool wiring after handover
Source code in your repo, runbook included
SCOPE A PERSONAL AGENT
MOST POPULAR
BUSINESS
From AUD $18,000
fixed-scope build · 4–6 weeks · +$650/mo managed

A team of agents. Multiple roles. Wired into your stack.

Multi-agent fleet — sales, ops, content, support roles
8–12 custom tools coded against your APIs / databases
Role-based access — staff get scoped agent permissions
Approval workflows for every outbound action
Audit log + replay for every agent action
Hosted on your cloud (AWS / GCP / Azure) or Pulse-managed
Custom dashboard branded to your business
90 days of tuning + new tool wiring included
SCOPE A BUSINESS BUILD
ENTERPRISE
Application only
engineering engagement · 8–14 weeks · retainer from $3,500/mo

On-prem. Sovereign. Built to your compliance posture.

Fully on-premise OR private VPC deployment
Open-weight models (no Anthropic / OpenAI dependency required)
SSO, RBAC, audit trail piped to your SIEM
Unlimited custom tools coded against your internal stack
Red-team review + safety harness before any prod action
Runbooks + handover docs for your engineering team
Quarterly tuning retainer + on-call support
IP assignment optional — repo can live in your GitHub org
APPLY FOR ENTERPRISE
WHO USES IT

Built for operators.

These are the shapes of business OpenClaw fits cleanly. If yours isn’t listed and you think it should be, that’s usually a sign you should book a call.

SOLO FOUNDERS

A chief-of-staff agent that drafts every reply, defends focus blocks, and keeps you out of the inbox.

REAL ESTATE AGENTS

After-hours buyer triage, open-home RSVPs, post-inspection follow-ups — all in your voice.

ACCOUNTING FIRMS

Client intake at EOFY, BAS reminders, document chasing. Tier-1 questions handled before staff touch them.

LEGAL PRACTICES

Conflict checks, intake summaries, deadline tracking. Read-only first, action-capable once trusted.

STUDIOS + AGENCIES

Brief intake, scope drafting, status updates to clients without a project manager.

OPERATORS WITH A PORTFOLIO

One agent per business unit. They talk to each other. You see the dashboard, not the chaos.

WHAT YOU GET

The deliverable. No fluff.

A working agent in your name

Persona, voice, context — all yours. Talks like you when it drafts.

A CLI you can talk to

Direct command access. Scriptable. Pipe into anything.

A web dashboard

See what the agent is doing, approve outbound actions, edit memory.

Source code and runbook

You own the repo. We hand it over with a runbook your engineer can use.

Deployment to your hosting

Pulse-hosted, your VPS, your cloud account, or your on-prem. You pick.

Tuning support after handover

New tools, new prompts, new triggers — wired in for the first 30/90 days.

▸ WHAT WE DON’T DO
Sell you a SaaS subscription disguised as "your" agent
Wrap a single closed model and call it sovereign
Pretend a Zapier workflow is an agent
Outsource the build to overseas contractors
Lock you into a proprietary agent framework
Ship without an audit log or kill switch
PROCESS

How we ship.

Five steps from kickoff to handover. No retainers, no scope creep, no surprise invoices.

01
Strategy call
30 minutes. We confirm fit, map the agent’s job, identify the top 3 actions worth automating, and lock pricing.
02
Scope + sign-off
Two-page brief. Roles, tools, triggers, governance rules, deployment target. You sign before any code is written.
03
Build (sprint 1)
Working agent in 7 days. Read-only at first — observes your inbox, calendar, sheets. You watch what it would do.
04
Trust + activate
Once you’re comfortable, we flip on outbound actions one at a time. Approval-gated by default.
05
Handover
Repo, runbook, dashboard access. Optional tuning retainer if you want us to keep iterating.
FIT CHECK

Read this before you book.

THIS IS FOR YOU IF
You’re drowning in inbox/calendar work and can’t hire fast enough
You’ve tried ChatGPT-as-tool and want something that actually does things
You care about owning your data, your prompts, your model choices
You want a build that scales — agent today, agent fleet next year
NOT FOR YOU IF
You want a free SaaS trial, not an engineering build
You want a single model vendor (we’ll point you at theirs)
You’re not ready to sign off on agent permissions and audit logs
You want fully autonomous AI with zero human in the loop — we won’t ship that
FAQ

Honest answers.

Wait — is "OpenClaw" a product I can buy off the shelf?
No. OpenClaw is Pulse.AI’s framework for shipping open, sovereign personal agents. We use open-weight models and open-source agent runtimes underneath, but the build is custom to your business. Think of it as "an agent built like infrastructure" — not a SaaS subscription.
Why not just use Claude / ChatGPT / Copilot agents?
You can, and they’re great for many use cases. Where they fall short: data sovereignty, custom tool surface, behavior tuning, vendor lock-in, and the $/action curve at scale. OpenClaw gives you the option to swap models, host wherever, and own the orchestration. We’ll happily tell you when a vendor agent is the better call — we’re not religious about it.
What models do you actually run?
Mix-and-match per task. Open-weights for routine work (drafting, summarising, classification), frontier closed models for hard reasoning when worth the cost. The orchestration layer routes the request to whichever model is cheapest-and-good-enough. You pick the budget envelope.
How safe is it to let an agent reply to my emails?
By default, every outbound action is human-approved. The agent drafts, you click send. Once you trust it on a category (e.g. EOFY client intake), you can flip that category to auto-send with a daily review log. You’re always in control of which actions are approval-gated and which are autonomous.
What about hallucinations / wrong answers?
Hallucinations come from missing context. The OpenClaw harness retrieves relevant emails, calendar events, and documents into the prompt before reasoning, and self-critiques the output before action. Wrong answers still happen — that’s why outbound actions are approval-gated until you trust them.
Can it handle voice?
Yes. Optional voice front-end via VAPI or Twilio. Useful for solo operators who want to direct the agent on the move ("draft a reply to Sarah saying we can do Tuesday at 3").
What happens if I want to change models / providers later?
Two-line config change. The orchestration layer was built for this. We’re aggressive about avoiding vendor SDK lock-in for exactly this reason.
Australian-hosted option?
Yes. We deploy to Sydney / Melbourne regions on AWS, GCP, or your on-prem hardware. Important for regulated sectors and Privacy Act 1988 compliance.
How does this compare to hiring a VA?
A VA scales linearly — one VA, eight hours a day, $25–60/hr. An OpenClaw agent runs 24/7, ~$0.05–$2 per task, and never forgets your context. They’re complementary: VA for high-touch human work, agent for everything else.
Will the agent learn my style?
Yes. We seed it with samples of your actual writing — emails, briefs, transcripts. Over time it adapts based on what you approve, edit, or reject. The longer you use it, the more it sounds like you.
READY?

Stop doing the work.
Build the agent.

30-minute strategy call. Free, no deck, no pitch. You walk away with a clear scope and a fixed price — or an honest “don’t build it yet”.

Australian-hosted option|You own the code|14-day delivery on Personal