Book an intro call
Solutions
Autonomous AI agentA digital co-worker that finishes the job Experience the agentThe experience in a two-minute scroll AI GovernanceEU AI Act-ready, with a living register AI Agent PlaybookBuild it yourself with copy-paste templates For installersQuotes, scheduling and client communication Interim AI leadProgramme steering, hands-on on site
Explore
How it worksThe interactive demo, step by step BlogInsights on AI, governance and the EU AI Act About FlowBaasWho builds this, and why Book an intro call
NL | EN
DIY Playbook · Templates · Quick-start kit

Build your own AI co-worker.

Not a chatbot that starts from scratch every session. A digital co-worker that knows your work, with identity, memory, tools and clear boundaries. Copy-paste templates you can put to use tomorrow morning.

A guide, not a guru. This playbook is built from months of working with an autonomous agent every day, including the mistakes and hard lessons you will not find in blog posts.

Reserve the playbook → Prefer it done for you? See the done-for-you agent →
Coming soon · €19.95 — early-bird price
01# Identity
02name: "Maya"  # your co-worker
03role: "Personal assistant"
04voice: "direct, no jargon"
05
06## Permissions
07- mail_triage: autonomous
08- quote_draft: with review
09- send_to_client: forbidden
→ ready in 30 min
Built withAnthropicOpenAIn8n· Enterprise expertise from an AI Portfolio Lead at a top-3 bank
DIY promise · What you get

A method, not a philosophy.

This is not an e-book full of meta statements about AI. It is the configurations, templates and boundaries we use ourselves every day, packaged so you can adopt them.

01 — What it is

Copy-paste templates, no theory

Complete IDENTITY.md, SOUL.md, MEMORY.md and TOOLS.md templates. Plus the configuration of an agent we use ourselves, which you translate to your own work in thirty minutes.

02 — Who it is for

Entrepreneurs who want to build it themselves

For founders, knowledge workers and operators who are technical enough to edit a config file, or willing to learn. Not for anyone looking for a no-code button.

03 — What makes the difference

Tested in practice, not from a whitepaper

Every template comes from an agent that picked up tasks daily for months. Including what went wrong: memory that was too complex on day 1, the cold start, the approval queues that overload themselves.

One method · six roles

The same building blocks, for every role you give it.

Identity, memory, tools, boundaries: those four blocks work for any function. You learn the method; which role you give it is up to you.

01

Personal assistant

Calendar, mail triage, summaries and follow-ups. Always up to date, always available.

02

Marketing strategist

Analyse campaigns, segment audiences, propose A/B tests and build out reports.

03

Content creator

Blog posts, newsletters and presentations, in your tone, with knowledge of your brand and audience.

04

Social media manager

Schedule posts, monitor trends, manage the community and track engagement, consistently across every channel.

05

Sales support

Qualify leads, automate follow-ups, update the CRM and prepare quotes. No lead slips through.

06

Operations & projects

Track progress, status reports, flag risks, update stakeholders. A grip on every project.

The playbook teaches you the method. Which role you choose is up to you, or build several side by side.

What's inside

Twelve modules from zero to a working co-worker.

Each chapter ends with a template or configuration you copy-paste straight away. No filler, no "why AI matters", only what works.

1

Chatbot vs. co-worker

Why most people get 10% out of AI. What makes a digital co-worker different from a prompt window.

2

Choosing a platform

OpenClaw, Claude Desktop, or a custom build? Pros and cons, honest, no affiliate talk.

3

Designing identity

SOUL.md and IDENTITY.md: voice, behaviour, boundaries and role. Includes a copy-paste base.

4

Memory architecture

Three-layer system: MEMORY.md, journals, knowledge graph. With nightly extraction and memory decay.

5

Tools & capabilities

What is essential (mail, files, web, shell) and what is optional (GitHub, browser automation, sub-agents).

6

Safety rules

Trust Ladder, approval queues, mail boundaries and defence against prompt injection, as configuration.

7

The working relationship

Communication, planning and autonomy: how you work day to day with a co-worker that knows your working style.

8

Automating tasks

From mail to reports. Parallel workflows, automatic restart when something stalls.

9

Working proactively

From reactive to proactive: the agent detects, triages and resolves, before you even see it.

10

Scaling up

From one co-worker to a team. Role division, cost optimisation, shared memories.

11

What went wrong

Honest lessons: memory too complex on day 1, cold start, approval overload, surprises about personality.

12

Quick-start kit

Seven steps, from a 30-minute install to a working agent in week 4, with checklists.

Quick-start · four phases

From zero to a working co-worker in four weeks.

Start this afternoon with thirty minutes of setup. By the end of week four you have an agent that knows your work.

1
Day 1 · 30 min

Install

Choose a platform, set up the repo, create your first IDENTITY.md. Start with one role, one task.

→ "Hi, I'm here"
2
Week 1

Identity + memory

Fill SOUL.md with your voice and boundaries. Activate three-layer memory, hot/warm/cold.

→ knows your context
3
Week 2–3

Tools + boundaries

Connect mail, files, web. Configure the Trust Ladder: what is autonomous, what needs review, what is forbidden.

→ does first tasks
4
Week 4

Scale up

Add a second task, first sub-agents, a ralph loop for long-running work. Only scale once layer 1 is stable.

→ a working co-worker
A taster · three core concepts

Three ideas that change your working relationship with AI.

These are the building blocks that make the difference between a chatbot and a co-worker. In the playbook you work through them step by step.

The Trust Ladder

Four levels of AI autonomy: from read-only to acting independently. You start restrictive and build trust on proof, not on hope.

Rung 1 · Read onlystart
Rung 2 · Draft & reviewweek 1
Rung 3 · Within the linesweek 3
Rung 4 · Full trustrare

The Ralph loop

No long sessions that stall. Short sprints, automatic restart on a stall, parallel agents in isolated worktrees. Work that carries on while you sleep.

agent #1 → stalled? → kill
agent #2 (fresh) → done
= 108 tasks in ~4 hours

Three-layer memory

Not everything matters equally. Recent work stays prominent, old work fades on its own. MEMORY.md for patterns, daily notes for the timeline, a knowledge graph for entities.

hot · <7 days — prominent
warm · 8–30 days — lower priority
cold · 30+ days — archived
Is it a fit for you?

Honestly: this is not for everyone.

The playbook is practical and technical, not a pretty e-book you cannot do anything with. Read below whether your situation fits.

This is for you if:

  • You are an entrepreneur, founder or knowledge worker who wants AI to genuinely know your work.
  • You are technical enough to edit config files (or willing to learn).
  • You want practical, copy-paste systems, not philosophical stories.
  • You are fed up with AI assistants starting from scratch every session with no context.

This is not for you if:

  • You are looking for a no-code, click-and-done solution (some setup is required).
  • You want AGI hype or science fiction, this is about working systems, not predictions.
  • You need enterprise-grade security guarantees, for that scale, done-for-you is the right route.
  • You expect one evening of tinkering with AI to solve the problem. This takes a few weeks of attention.
From practice

A guide, not a guru.

This playbook was not written from theory. It is the result of months of working with an autonomous agent every day, including the mistakes and surprises you read nothing about in neatly styled threads.

No demo magic. Working systems.

Every template, configuration and safety rule in this playbook runs in practice. You do not get a philosophical story about "what AI could become", but the exact files, prompts and checklists we work with every day.

The mistakes are in there too. Because it is not fair to sell only the highlights.

"No theory, the actual systems, configurations and lessons from building a working relationship between human and AI." — FlowBaas, from the playbook's introduction
Cover of the AI Agent Playbook
Templates
12

Complete, copy-paste, in markdown.

Lead time
~4 wks

From zero to an agent that knows your work.

Early-bird price
€19.95

One off, no subscription, no upsell.

Rather not build it yourself?

Two paths, choose what fits.

This playbook is secondary to our done-for-you track. For anyone with the time and the appetite for detail, building it yourself is fine. For anyone who wants it outsourced, there is the autonomous agent.

Done-for-you · Autonomous AI Agent

We set up an agent for you. Live in four working days.

One task, integrated into your existing tools, with logging and KPIs from the very first run. Four working days to live, no annual contract, scale up only once the numbers add up.

See the done-for-you agent →
Your time~4 weeks yourself vs. ~1 hour intake
Investment€19.95 playbook vs. pilot price tailored
Who buildsYou, with templates vs. We, for you
GovernanceBase set vs. top-3-bank level
Suited toFounder, small team vs. SME organisation
Reserve · or book a call

Ready to build your own AI co-worker?

Templates · 12 Quick-start · 4 weeks One-off price · €19.95

The playbook is out soon. Reserve at the early-bird price of €19.95 and get an email the moment it is available. No subscription, no upsell, no sales funnel.

Reserve the playbook → Book an intro call
Usually replies within 1 working day. No obligation, no sales funnel.