AI OS · Magic Pulpo
Every AI tool, running on your company’s memory.
Your team already uses AI in several disconnected tools, each starting from zero, none knowing your brand, your clients or your rules. An AI OS is the layer underneath, with one set of knowledge, guardrails and voice. Tools change; the layer compounds.
Our point of view
Scattered knowledge makes every model sound generic.
- Brand memory any tool can read
- Retrieval over your own material
- Guardrails set once, used everywhere
The system
The layer, in three parts.
01
Brand & knowledge memory
Your positioning, voice, offers, case history and client context, structured into a retrieval layer every tool and workflow draws from. Output stops sounding like a model and starts sounding like you.
02
Guardrails as infrastructure
What AI may see, say and do, encoded once and enforced everywhere: access boundaries, approval gates, tone rules, escalation paths. Governance lives in configuration instead of a PDF nobody opens.
03
An operating rhythm your team keeps
Tools and models will keep changing. The OS gives you the stable part: how knowledge gets in, how outputs get reviewed, how new use cases get adopted. Your team keeps running it after we leave.
The engagement
How it works
A structured engagement from first call to measurable results.
- 01
Knowledge inventory
Where your company’s actual knowledge lives: docs, CRM, inboxes, heads. What is current, what is contradictory, what is missing.
- 02
Memory architecture
Structure it into a retrieval layer: brand canon, client context, operational knowledge. Versioned, owned, queryable by any tool you adopt.
- 03
Guardrail design
Access, approval and tone rules agreed with you and encoded into the layer from the start, aligned with your governance obligations.
- 04
Wire the tools
Connect the assistants and workflows your team uses, so they draw on the shared memory instead of starting cold.
- 05
Adoption rhythm
Team onboarding, use-case backlog, and a monthly review of what the layer learned and what it should learn next.
What you get
Every engagement is scoped and priced upfront. No retainer surprises.
- 01Structured brand and knowledge memory, versioned, in your infrastructure
- 02Retrieval layer connected to the AI tools your team uses
- 03Encoded guardrails: access boundaries, approval gates, tone rules
- 04Integration with your existing stack: CRM, docs, storage, chat
- 05Team onboarding and a written operating rhythm
- 06Use-case backlog ranked with owners assigned
- 07Full ownership: the layer, the config and the docs are yours
- 08Quarterly architecture review for the first year
What this costs
Indicative: the final range is set on the call. Fees only; media spend is never marked up. Figures in EUR.
We run this ourselves
This site runs on one.
Every Insights post here is drafted by our editorial engine from a facts registry Mark owns, a map of the seven pillars and a set of slop rules shared with the gates, and nothing publishes until a person merges it. The same layer this page sells, pointed at ourselves.
Read what it writesThe proof
