Cloud AI Agent — Always On

Your AI agent.
Actually working.

Sidekick connects to your tools, remembers what you care about, and takes action — while you sleep, focus, or do literally anything else.

It reads your email. It monitors your repos. It messages your team. It never forgets what you told it last week. You define the rules — Sidekick does the running.

See how it works

sidekick / chat

You

Draft a reply to the client about the Q3 delay — keep it professional but honest.

Reading Gmail thread… found 3 emails from Marcus.

Draft ready. Want me to send it, or would you like to review first?

SK

You

Actually, cc Sarah too. Then send.

✓ Sent — Marcus + Sarah Also: you have a calendar conflict on Thursday. Move the 2pm or the 4pm?

SK

What Sidekick can do

Built for people who
hate babysitting software

Every feature earns its place. Here's what Sidekick actually does.

Memory That Persists

Three memory tiers — working, episodic, semantic — mean Sidekick remembers what you told it last month, not just last message. "Actually make it Manchester" just works.

Natural Language Commands

"Email Sarah about the Q3 report" — Sidekick resolves Sarah from your contacts, picks the right email account, drafts the message, and asks if you want to send. No forms. No dropdowns.

Proactive Alerts

Sidekick taps you when things need attention — unanswered emails, CI failures, calendar conflicts — before they become problems. Configure how often, through which channel.

You Stay In Control

Ask-first, act-then-report, or fully autonomous — per integration if you like. Destructive actions (deletions, financial transactions, new contacts) always need your say-so, no matter the mode.

Skill Marketplace

Browse 13,729+ community skills from ClawHub. Semantic search matches your intent — "help me schedule meetings" finds the right skills without exact keywords. One-click install, auto-mapped credentials.

Custom Skill Builder

Write your own skill in plain English. Specify which integrations it needs, test it in a sandbox, then publish it to ClawHub or keep it private. Your logic, running on your agent.

Decision Transparency Log

Every decision Sidekick makes is logged — what it did, why, and what triggered it. Full reasoning chains, real-time. You're never guessing what your agent got up to.

Integrations That Go Deep

Gmail, Google Calendar, GitHub, Linear, Slack, Notion, Telegram, WhatsApp, Discord, Outlook & Microsoft 365. Not just read — Sidekick acts across all of them, via a single unified pipeline.

Talk To It

Press-and-hold to give voice instructions. Sidekick transcribes and acts. Works on mobile, and supports voice notes forwarded from Telegram or WhatsApp. Type or talk — your call.

AES-256 Encrypted Credentials

Every OAuth token and API key is stored with AES-256-GCM encryption at rest. Auto key rotation with zero-downtime re-encryption. You can trigger manual rotation anytime.

Emergency Kill Switch

One button. Stops everything — cancels pending actions, halts the agent loop, disconnects all integrations. Available on mobile. Peace of mind that you're always the one in charge.

Agent Analytics

Actions per day, response times, most active integrations, memory utilization, skill usage — all in one dashboard. See exactly where your agent is spending its cycles.

Quiet Hours

Define time windows when the agent collects but doesn't act. Events queue up, actions fire when you're back on. Timezone-aware — Sidekick knows where you are.

Priority Rules Engine

Set rules like "always respond to messages from the CEO within 5 minutes" or "ignore marketing emails." Your priorities, enforced automatically across every connected channel.

OpenClaw Compatible

Already using OpenClaw? Export your skills, import them to Sidekick, connect your accounts — and you're running. Your existing work travels with you.

Works On Your Phone

Fully responsive dashboard with touch-optimized targets and swipe gestures. Activity feed, connection status, and the kill switch — all accessible from your pocket.

Rate Limits Per Integration

Set max emails per hour, max messages per day, spending caps for paid APIs. Sidekick stays inside your guardrails and alerts you before you hit them.

Streaming Responses

Watch Sidekick think in real time. Tool calls render as collapsible cards — see it check your calendar, read the email, query Linear — as it happens, not after.

Your Preferred Channel

Pick where Sidekick talks to you — browser, Telegram, Slack, WhatsApp, or email. Start a conversation in the browser, finish it on Telegram. Sidekick follows you.

GDPR & Data Export

Export everything in machine-readable format. Request full account deletion. Auto-purge policies remove old decisions and memory after configurable periods. Your data, your rules.

Works with the tools you already use

{["Gmail", "Google Calendar", "GitHub", "Linear", "Slack", "Notion", "Telegram", "WhatsApp", "Discord", "Outlook & M365"].map((tool) => ( {tool} ))}

Gmail Google Calendar GitHub Linear Slack Notion Telegram WhatsApp Discord Outlook & M365

Your first 10 minutes

From signup to agent
running in 4 steps

1

Connect your tools

Link the accounts that matter

Open the Connection Manager and add your first integration — Gmail, GitHub, Slack, whatever you use most. Each connection uses OAuth, so there's no copy-pasting of API keys. Sidekick's guided onboarding walks you through the first one in under two minutes.

2

Install a skill

Pick what your agent can do

Browse 13,729+ community skills from ClawHub or write your own in plain English. Search by what you need — "triage my GitHub issues" — and Sidekick finds the right skill. One click installs it and maps your credentials automatically.

3

Set your preferences

Tell it how you like to work

Choose your autonomy level — ask-first while you're getting comfortable, then switch to act-then-report as trust builds. Set quiet hours, priority rules ("always ping me if the CEO messages"), and pick the channel where Sidekick talks to you.

4

Watch it work

Your agent starts running

Sidekick starts monitoring your connected accounts. New emails, GitHub notifications, Linear updates, calendar conflicts — it surfaces what matters in the chat, proposes actions as cards you can approve in one click, and logs every decision it makes so you're never in the dark.

Questions

The honest answers

How autonomous is Sidekick, really?

As autonomous as you want it to be. There are three modes: ask-first (every action needs your approval), act-then-report (it acts, then tells you what it did), and fully autonomous (silent execution). You can override this per integration — so Gmail might be ask-first while your Linear account is fully autonomous. That said, some actions are always confirmed regardless of mode: deleting data, sending to new contacts, financial transactions, and revoking access. Those never run without a human green-light.

What does "memory" actually mean here?

Sidekick uses three memory tiers. Working memory holds recent conversation context in real time — so mid-conversation context switches ("actually make it Manchester") just work. Episodic memory stores daily summaries so it knows what happened last week. Semantic memory uses vector search to recall longer-term patterns — your preferences, your communication style, recurring topics — even across months of history. Previous conversations are searchable and resumable from within the chat interface.

How does Sidekick handle my credentials and API keys?

All OAuth tokens and API keys are stored with AES-256-GCM encryption at rest. Key rotation is automatic with zero-downtime re-encryption — your integrations don't go offline during rotation. You can also trigger manual rotation at any time and check the current encryption status from the settings page. Skill environment variables are encrypted with the same standard.

I'm already using OpenClaw. Do I have to start over?

No. Sidekick is OpenClaw compatible. There's a dedicated migration guide: export your skills from OpenClaw, import them to Sidekick using the ClawHub importer (supports SKILL.md format including YAML frontmatter and markdown instructions), then connect your accounts. Your skills validate for compatibility before installation, so you'll know upfront if anything needs adjusting.

What happens if I want the agent to stop immediately?

Hit the kill switch. One button — available on desktop and mobile — that immediately stops the agent loop, cancels all pending actions, and disconnects every integration. It's accessible from the main dashboard at all times. You can reconnect everything individually once you're ready to start again.

Can I see everything the agent has done?

Yes — in detail. The Agent Activity Feed shows a real-time log of every action: the trigger event, the reasoning chain, and the action taken. It's filterable by integration and action type. There's also a separate Decision Transparency Log that shows the full reasoning chain for each decision, not just the outcome. The Analytics Dashboard gives you a higher-level view: actions per day, most active integrations, skill usage breakdown, and response times.

What are the plan limits and how are they enforced?

Sidekick offers Free, Pro, and Enterprise tiers. Free includes 3 integrations and 50 actions per day; Pro gives unlimited integrations and 1,000 actions per day plus custom skills and priority support; Enterprise is unlimited with SSO, a dedicated support contact, and an SLA. Limits are enforced at the agent loop level — every action increments a daily counter, and your real-time usage vs. plan limit is visible in the dashboard header. Enterprise users on usage-based billing can see cost tracking per action in real time.

Security & Trust

An agent you can hand your keys to

We built the controls we'd want if it were our accounts on the line.

AES-256-GCM Encryption

Every credential — OAuth tokens, API keys, skill environment variables — encrypted at rest. Automatic key rotation, zero downtime.

Full Audit Trail

Every decision the agent makes is logged with its full reasoning chain. Real-time activity feed, filterable and auditable at any time.

Hardcoded Safety Gates

Destructive actions — deleting data, financial transactions, new contacts, access revocation — always require human approval. No setting bypasses this.

GDPR Compliant

Export all your data in machine-readable format. Request full account deletion. Auto-purge policies for agent decisions and memory after configurable retention periods.

Now in private beta

Your ideas.
Accelerated by AI.

Sidekick is the cloud AI agent that works the way you do — across your real tools, on your schedule, with you always in control.

Join the waitlist. Be among the first to run an agent that actually earns its keep.

Join the Waitlist