Simple, secure AI that gets
done, fast.

ClankTeam is a shared workspace of persistent frontier-model AI agents who understand your business and connect to the tools you already use.

Get Started Contact us

ClankTeam

Recent

Today

Adapt NYC Ad Campaign ...

Previous 7 Days

Q1 revenue report & forecast

Draft onboarding email drip

Refund policy update

Copy our top-performing NYC multichannel ads campaign to target the LA market.

Team Lead

Fetched NYC Spring Push: 3 ad sets

Fetched NYC Search + Display

Thought Analyzing campaigns... 12s

I've analyzed the NYC campaign and built the LA port. Here's what I did:

Audiences: Swapped NYC zip codes for LA equivalents by income bracket. Shifted age skew from 28-35 to 25-38 to match LA demo data.
Creative: Duplicated all 14 Meta creatives. Replaced NYC imagery with LA stock. Updated headlines and CTA copy for West Coast tone.
Budget: Allocated $18.4k/mo (70% of NYC spend) as a test budget, split 60/40 Meta/Google matching NYC's best channel mix.
Channels: Ported all 3 Meta ad sets + 8 Google campaigns. Added TikTok Spark Ads.

la-creatives-batch/ 14 files

All campaigns are staged as drafts in Meta Ads Manager and Google Ads. Ready for your review before going live.

Ask a follow-up...

Online Advertising

You can use ClankTeam agents to find wasted spend quickly and reallocate ad dollars to your most effective ads, campaigns, and keywords.

Using ClankTeam you can:

Audit running campaigns: pull performance data from Google Ads, Meta, and LinkedIn
Pause underperformers: quickly find and stop ads, keywords, and campaigns from losing you money
Reallocate budget: identify top-performing creatives and audiences and shift spend to them
Find new audiences: quickly set up lookalike and new segment tests

Get Started

Want to set up AI online ads yourself? Check out our free guide

Pull campaign performance data

Pause 3 underperforming ad sets

Reallocate $420/day to top creatives

Launch 2 new lookalike audiences

ROAS up 3.8x

Support

When a support email comes in, ClankTeam can check your FAQ and knowledge base, drafts a reply, and queue it for you to review before sending.

Ingest new tickets: monitors your support inbox
Match against docs: searches your FAQ, help center, and past replies
Draft a reply: writes a response in your brand voice
Queue for review: you approve, edit, or escalate

Get Started

New ticket: "How do I cancel my subscription?"

Searching FAQ & help center...

Match found: "Cancellation & Refund Policy"

Draft reply written

Ready for your review

Customer Onboarding

When a new customer signs up, you can have ClankTeam send a welcome email, provision their account, book the kickoff call, and follow up if they get stuck.

Trigger on signup: detects new customers from Stripe or your CRM
Send welcome emails: personalized messages with setup guides
Provision their account: creates workspaces, sets permissions, seeds data
Book the kickoff call: schedules via Calendly, sends calendar invite
Follow up if they stall: nudges customers who haven't completed setup

Get Started

New customer: Acme Corp ($299/mo)

Welcome email sent with setup guide

Workspace provisioned, permissions set

Kickoff call booked: Thu 2:00 PM

Onboarding complete, zero manual steps

Meet your Team Lead

A smart agent that gets you up and running fast.

Your Team Lead is the first agent you talk to. Tell it what you need in plain language and it figures out how to get it done.

When a task needs specialized work, your Team Lead delegates to always-on specialist agents who handle the details (email, ads, support, data, code) and reports back with finished results.

Your Team Lead translates what you want into results without complicated setups.

Agents turn tasks into results

Tell ClankTeam what you need. It reads your docs, calls your APIs, and gives you finished output to review.

Learn about abilities →

ClankTeam

Write a Q1 investor update email. Pull numbers from our Stripe dashboard and last quarter's board deck.

Team Lead

Read q4-board-deck.pdf

Fetched Stripe: Q1 revenue, MRR, churn

Thought Comparing Q1 vs Q4, drafting narrative... 8s

MRR grew 34% to $128k. Churn dropped to 2.1%. Here's a draft that leads with the growth story and addresses the support hire.

q1-investor-update.md +84 lines

Ask a follow-up...

Works autonomously, runs in parallel

Set up triggers to run tasks on a schedule or in response to events. Results go to your approval queue.

Learn about triggers →

ClankTeam

Triggers

Reply to support emails

On new email · support@yourco.com

Active

Weekly revenue report

Every Monday at 8:00 AM

Active

Post daily standup summary

Every weekday at 9:30 AM

Active

Reconcile new invoices

On Stripe webhook · invoice.paid

Active

Social media calendar

1st of every month

Active

In every tool, at every step

ClankTeam connects to Gmail, Slack, Stripe, Google Ads, Meta Ads, GitHub, and more.

Learn about abilities →

ClankTeam

Abilities

Gmail

Read, draft, and send emails

Connected

Slack

Post messages, read channels

Connected

Stripe

Invoices, refunds, subscriptions

Connected

Google Ads

Campaigns, ad groups, reporting

Connected

Meta Ads

Ad sets, creatives, audiences

Connected

You stay in control

Configure approvals by risk level. Full audit trail on every action.

Learn about approvals →

ClankTeam

Pending Approvals

Medium Sally Support

Send refund of $47.50 to customer #9281

Customer requested refund for order #9281 (defective item). Stripe refund staged and ready.

High Calvin Code

Deploy webhook handler to production

New Stripe webhook endpoint for subscription upgrades. Tests passing, staged for deploy.

Low Danny Data

Share weekly revenue report via Slack

Q1 week 12 report ready. Posting to #team-updates with PDF attachment.

Everything your agents need

Three building blocks that combine into any workflow.

Abilities

Connect the tools your business already uses.

Read the docs →

Gmail Slack Stripe Google Ads Meta Ads GitHub TikTok + more

Credentials

Logins, API keys, OAuth tokens. Stored in a secure vault so agents can access your services.

Read the docs →

OAuth tokens API keys Encrypted vault + more

Documents

Upload your private knowledge so agents have the context they need.

Read the docs →

Email Files Dropbox SharePoint OneDrive iCloud + more

Get Started

Automation and control

Triggers run tasks on a schedule. Approvals give you final say.

Triggers

Reply to customer emails on arrival, run payroll biweekly, send status reports every Monday. Configure once, runs on schedule.

Read the docs →

On new email Scheduled Recurring + more

Approvals

Agents draft the work, you make the final call. One-click approve or reject, with a full audit trail.

Read the docs →

One-click approve Risk levels Audit trail + more

Get Started

Security you don't have to think about

16 security systems across every layer. Your agents work safely by default.

Read the docs →

Human approvals

Agents ask you before performing risky actions. All configurable so you're only asked about important things, no noise.

Private VM

Your agents run on a secure, isolated, dedicated virtual machine. A firewall and private network stop attackers from ever reaching them.

Sandboxed execution

Tool code runs inside WASM with dual metering. File operations are workspace-confined. Subprocesses are env-cleared and timeout-enforced.

Defense in depth

Ed25519 manifest signing, Merkle audit trail, taint tracking, SSRF protection, prompt injection scanning, secret zeroization, and more.

One workspace for the whole team

Everyone sees the same agents, the same context, the same results.

Shared agents

Every team member can see what agents are doing, review output, and approve actions.

Shared knowledge

Documents, credentials, and context live in one place. Connect a tool once and the whole team has access.

Stay in the loop

Optional push notifications, Slack alerts, and email digests when agents finish work or need approval.

Frequently asked questions

What are "frontier models"?

Frontier models are the most capable AI models available at any given time. ClankTeam uses the most up-to-date version of Claude Opus by Anthropic, which leads on reasoning, instruction-following, and safety. Unlike older chatbot-style AI, frontier models can read documents, call APIs, and complete multi-step workflows with genuine understanding of what they're doing.

What does "persistent" mean?

Most AI tools reset after every conversation. ClankTeam agents are persistent. They keep their memory, context, and access to your tools across sessions. When an agent finishes a task on Monday, it still has that context on Friday. Your team doesn't have to re-explain anything.

How does the AI learn my private workflows or confidential procedures?

You teach agents by connecting them to your existing documents, SOPs, and tools. Agents read what you give them access to (help center articles, internal wikis, past emails, CRM data) and follow the procedures described there. Your data stays on your private instance and is never used to train the underlying AI model. See Documents and Abilities in the docs.

What happens if an agent makes a mistake?

High-risk actions are queued for your approval. You choose which actions require sign-off. Every action is logged. See Approvals in the docs.

Is my data shared with other customers?

No. Each organization runs on its own private virtual machine with a dedicated agent runtime. Your data, credentials, and agent activity are fully isolated. There is no shared infrastructure between customers. See Security in the docs.

Can I connect my existing tools?

Yes. ClankTeam agents connect to the tools you already use: Gmail, Slack, Google Calendar, Stripe, your CRM, file storage, and more. You authorize a tool once and every agent on your team can use it, with permissions you control. See Abilities and Credentials in the docs.

Do I need technical skills to set this up?

No. You describe what you want an agent to do in plain language, connect the relevant tools, and set the approval level. There's no code to write and no branching logic to configure. Most teams have their first agent running in under an hour.

How is this different from Zapier or Make?

Zapier and Make run fixed automation flows. When conditions change, they often need manual fixes. ClankTeam agents adapt and flag problems instead. They also complete end-to-end workflows that would require dozens of connected zaps.

Simple, secure AI that gets
done, fast.