KIN · AUTONOMOUS AGENTS

AI teammates that work on
your business data

Give a kin a job in plain words. It queries your databases, reads the web and your files, checks every number, and asks before it changes anything.

Request an invitation Visit kin.thinkingdbx.com

Free for 3 kin · Pro $49/month · invitation only during early access

kin.thinkingdbx.com/app
Your kin
Revenue Watch
Churn Analyst
Competitor Scout
Workspace
Inbox 3 new
Approvals 1
Connections
you: every Monday, check last week's revenue by region and flag anything that dropped more than 10%  working...
querypostgres_prod · weekly revenue by region · 412 rows
compareweek 40 vs week 39 · EMEA down 14.2%
readsupport_tickets.csv · billing outage on Oct 1
verifyevery figure in the report matches the query output
waitingneeds your approval before sending
Email the report to sales-leads@acme.com?
Summary, chart, and the 3 accounts behind the EMEA drop.
ApproveEditReject
Works with your data PostgreSQL MySQL Snowflake BigQuery ClickHouse MongoDB Amazon S3 MCP servers
How it works

You describe the job. Your kin does the work.

No workflow builder and no code. Each kin is a named teammate with its own role, tools, and memory.

  • 1

    Create a kin

    Start from a template like Revenue Watch or Competitor Scout, or write the role yourself. Pick which connections and tools it may use.

  • 2

    Give it work, or a trigger

    Ask it something now, or run it on a schedule, a webhook, an incoming email, or when a table changes.

  • 3

    Get a checked answer

    It plans, runs real queries, reads what it needs, and checks its own figures before the report lands in your inbox.

  task log · churn analyst
# triggered: schedule, Mon 08:00
plan      find accounts at risk this month

describe  warehouse.subscriptions  # 14 columns
query     "SELECT plan, COUNT(*) ... GROUP BY plan"
run_tl    churn score on 2,318 accounts
memory    "exclude internal test accounts"

verify    7 of 7 figures match source  ok
report    sent to inbox · 41 accounts flagged
What a kin can do

Data first, but not data only

Kin is built on ThinkingLanguage, so working with tables is native. It also browses, reads, and writes like a general assistant.

01

Queries your databases

Reads schemas, writes SQL, and runs analysis on live data across Postgres, MySQL, warehouses, and files you upload.

02

Checks its own numbers

Before a report goes out, every figure is matched against the actual query results. Mismatches get fixed, not sent.

03

Asks before acting

Sending email, writing data, or calling an external service waits for your approval. Approve, edit, or reject in one click.

04

Works on its own schedule

Cron schedules, webhooks, inbound email, and data change triggers. Your kin wakes up when there is work to do.

05

Remembers and improves

Facts, preferences, and corrections are kept as memory. Successful runs become playbooks it reuses next time.

06

Browses and uses tools

Searches and reads the web, works with PDFs and spreadsheets, and connects to any MCP server you trust.

Built for real work

Kin people actually run

Finance

Revenue Watch

Checks revenue, orders, and margins every morning. Flags drops, explains the likely cause, and sends a short summary.

Customer success

Churn Analyst

Scores accounts by usage and billing signals, lists who is at risk, and drafts outreach for your team to approve.

Market

Competitor Scout

Watches competitor pricing pages, releases, and news. Reports what changed and why it matters to you.

Safe by design

An agent you can let near production

Autonomy is only useful if you can trust it. These are the guardrails every kin runs inside.

Can it change my data?
Only with your say-so. Writes, emails, and outbound calls pause for approval, and you can edit the action before it runs.
Can I see what it did?
Every task keeps a full step log: each query, page, file, and tool call, with live progress while it runs.
What can it reach?
Only the connections you grant it. Web access runs through an egress-controlled sandbox that blocks private and internal addresses.
How are credentials stored?
Connection secrets are encrypted at rest and never shown to the model. Backups are encrypted too.
Who can sign in?
Your team, with strong passwords checked against known breaches and optional two-factor authentication.
Is my data used for training?
No. Your data is processed only to run your tasks. See the privacy policy for the details.
Pricing

Start free. One flat price when you grow.

One plan covers Kin and the ThinkingLanguage platform. Use the default Kin model, or bring your own API key.

Free
For trying Kin on a real problem.
$0
No card required
  • ✓3 kin
  • ✓300 tasks a month
  • ✓1M model tokens a month included
  • ✓Schedules, webhooks, and email triggers
  • ✓Approvals, memory, and full task logs
Request an invitation
For teams
Pro
For teams that run kin every day.
$49/month
Flat price · includes the TL platform
  • ✓25 kin
  • ✓5,000 tasks a month
  • ✓20M model tokens a month included
  • ✓Everything in Free
  • ✓Priority support
Request an invitation
Custom
For larger teams with more work to hand over.
Let's talk
Hosted on our cloud · priced on usage
  • ✓More kin, tasks, and tokens than Pro
  • ✓Organization accounts with several workspaces
  • ✓Use your own model API keys
  • ✓Help setting up your first kin
  • ✓Everything in Pro
Talk to us

Every plan works with the default Kin model or your own API key. Prices in USD.

Questions

Frequently asked

Anything else? Write to contact@thinkingdbx.com and a person will reply.

Do I need my own LLM API key?

No, it is your choice. Every kin can use the default Kin model, with usage included in your plan's monthly token allowance. Or bring your own key for OpenAI, Anthropic, Groq, or any OpenAI-compatible endpoint, and that kin runs on your account instead.

How is Kin different from a chatbot?

A chatbot answers when you ask. A kin owns a job: it runs on a schedule or trigger, works through many steps on its own, checks its results, and reports back. You only step in to approve actions.

Why is it invitation only?

Kin is in early access. We are onboarding teams in small batches so we can support each one well. Email contact@thinkingdbx.com and tell us what you would like a kin to do.

Which data sources can it use?

PostgreSQL, MySQL, Snowflake, BigQuery, ClickHouse, MongoDB, S3, files you upload, the public web, and any MCP server you connect.

How does it relate to ThinkingLanguage?

Kin runs its data work on ThinkingLanguage, our open-source language for tables and streams. Your Pro plan covers both Kin and the hosted TL platform.

Get started

Your first kin, working in five minutes

Pick a template, connect a database, and give it a job. The first report lands in your inbox the same day.

Request an invitation Sign in
0:30

Create a kin

Choose a template or describe the role.

2:00

Connect your data

A database, a file, or an MCP server.

5:00

First task running

Live progress now, a checked report when it is done.