Skip to main content

Turn company knowledge into something your business can actually use.Secure. Structured. Built to scale.

Most companies already sit on a great deal of valuable knowledge. The trouble is that it is scattered across documents, inboxes, shared drives, ticket queues, wikis and the heads of a handful of long-serving colleagues. ThePromptMonk helps turn that knowledge into a secure, company-owned AI knowledge system.

The problem is rarely a lack of knowledge.

Consultancies, advisory firms, agencies, industrial businesses and technical organisations produce valuable knowledge every single day. Answers to client questions, process write-ups, internal standards, proposal logic, project lessons learnt, technical documentation and the small recurring decisions that keep the place running. The catch is that staff still spend a remarkable amount of time hunting that same knowledge down, rewriting it from scratch or piecing it back together from the memory of one or two key colleagues.

A private AI knowledge system shifts that baseline. It makes internal knowledge searchable, conversational and genuinely usable, without shipping sensitive company data into public AI environments where nobody quite knows where it ends up.

What is a private AI knowledge system?

A private AI knowledge system is a company-owned AI environment that makes internal knowledge from documents, processes, policies, projects and line-of-business systems genuinely usable.

Unlike public AI systems, a private AI knowledge system works with the organisation's own knowledge base, respects roles and permissions, and fits into existing governance, data protection and compliance structures rather than circling around them.

The point is not to replace people. The point is to make existing knowledge easier to find, more consistently applied and rather less dependent on individual memory.

Private AI Knowledge System

  • →Company knowledge
  • →Roles and permissions
  • →Traceable sources
  • →Governance and compliance
  • →Secure AI usage

Typical questions companies already want to answer today

The value of a private AI knowledge system does not come from the technology itself. It comes from the rather more useful ability to put existing knowledge in front of the right person at the right moment.

Tax advisory

“Which deadlines apply in this case and which comparable client matters have we already handled?”

Management consulting

“Which presentations, frameworks and project experiences already exist on this topic?”

Law firms

“Which briefs, contract templates and case law have we already produced for a comparable constellation?”

Corporate law practices

“Which internal standards, review schemes and matter experience exist on this regulatory question?”

Mechanical engineering

“Has this technical problem occurred before and how was it resolved at the time?”

Software companies

“Which documentation, tickets or product decisions exist for this feature?”

Accounting

“Which internal policies apply and how have similar cases been handled so far?”

Industry

“Which standards, processes and experiences exist across sites on this topic?”

Industries where a private AI knowledge system carries real weight

A private AI knowledge system tends to find its strongest footing in organisations where knowledge is the central source of value and where that knowledge has to remain firmly under the company's own roof. The following industries consistently see results in productivity, quality and data sovereignty.

Advisory firms and the professional services

Client knowledge, methods and internal standards stay reliably accessible without ever leaving the firm's own perimeter.

  • Structured search across client files and contractual archives
  • Reusable building blocks for opinions and formal statements
  • Swift onboarding of newly qualified professionals

Mechanical engineering and technical B2B companies

Technical knowledge from engineering, service and sales becomes a dependable foundation for everyday operations.

  • Service and spare parts answers drawn from internal documentation
  • Proposal processes informed by proven project references
  • Knowledge retention when experienced colleagues retire or move on

Software companies and technology providers

Tickets, repositories and internal sources are brought together into a single, shared knowledge layer for support, product and engineering.

  • Faster answers across first and second-line customer support
  • Consistent product documentation for sales and customer success
  • Accelerated onboarding for new engineers

Accounting, finance services and firm networks

Recurring matters are handled more swiftly, and internal rules are applied with the kind of consistency clients quietly expect.

  • Quick responses to recurring client questions in daily work
  • Consistent application of internal policies and review steps
  • Structured research across regulations and official guidance

Industrial groups and the upper mid-market

A consistent knowledge layer emerges across multiple sites, languages and business units.

  • Group-wide access to policies, standards and internal norms
  • Multilingual knowledge base for international locations
  • Clean separation of permissions along the org structure

Perspectives across the different areas of responsibility

A private AI knowledge system touches several functions inside a company. The following perspectives outline the questions that typically arise from each area of responsibility and how a private knowledge system addresses them.

Chief executives and owners

For the chief executive, the commercial leverage is what matters. A private AI knowledge system is an investment with an addressable return in productivity, reduced operational risk and meaningful independence from any single expert.

COOs, heads of operations and organisation

Operational leaders experience information loss, duplicated work and unclear ownership as constant friction in day-to-day business. A private AI knowledge system structures those friction points and makes improvements visible in numbers rather than anecdotes.

Heads of customer service, sales and knowledge management

In customer-facing roles the benefit becomes immediately tangible. Response times shorten, proposals become more consistent, and existing knowledge finally reaches the place where it actually creates value.

Heads of IT and information security

For IT and information security, architecture, data sovereignty, permissions, auditability and clean operations are central. A private AI knowledge system is built around precisely these requirements and complements existing security and compliance architectures.

What ThePromptMonk does

1

Orientation

We clarify which knowledge areas are truly valuable and where the business benefit actually sits.

2

Use case design

We prioritise use cases by impact, feasibility, data sensitivity and organisational readiness.

3

Knowledge structure

We help prepare documents, sources, roles, access levels and response logic.

4

Governance

We define rules for ownership, approvals, usage, quality assurance and continuous improvement.

5

Implementation with the right technology partners

When it comes to the technical build, we bring in the partners that genuinely fit the brief, depending on requirements, infrastructure and data protection needs.

Why not simply use ChatGPT in the browser?

Public AI tools have their place. For business-critical knowledge, that place is usually rather limited. Once internal documents, client data, contract details, process know-how or strategic material come into play, the conversation moves well beyond convenience. It becomes a question of control, accountability and being able to show, after the fact, who did what with which information.

A private AI knowledge system creates a properly governed environment for productive AI use inside the company. Staff get a familiar interface, while data sources, roles, permissions and usage scenarios sit under clearer organisational control.

Public AI and private AI knowledge systems compared

Public AIPrivate AI Knowledge Systems
General world knowledgeCompany-specific knowledge
No role-based controlRoles and permission concepts
Public data sourcesInternal documents and systems
Limited traceabilityTraceable sources and governance
Generic answersCompany-specific answers
Individual usageOrganisation-wide knowledge layer

Public AI and private AI pursue rather different ends. For many organisations the greatest benefit comes from combining both, in their proper places.

Example of one possible technical implementation

Example of one possible technical implementation: BaseGPT

ThePromptMonk remains technology-open. Depending on the requirements, the data protection profile, the existing infrastructure and the organisational starting point, different platforms and architectures may come into play.

BaseGPT is one example of a solution that can be particularly interesting for companies that care about data sovereignty, European infrastructure and the use of open-weight models. Built by coding9.de GmbH, the platform fine-tunes open-weight language models such as Llama, Mistral, Qwen, DeepSeek, Phi or Gemma on the company's own documents, terminology and processes. It runs on the company's own hardware, on bare-metal infrastructure or, if preferred, fully managed from a Frankfurt data centre, without quietly routing data through US clouds.

The remit covers domain fine-tuning with LoRA, QLoRA or full tuning, retrieval-augmented generation across internal documents and systems, and integration into the business systems people actually use, such as SAP, Microsoft 365, Confluence or Jira. A chat interface, admin console, single sign-on, role model and audit log come as standard. Compliance is not an afterthought. The platform is aligned with the GDPR, the German Federal Data Protection Act (BDSG) and the EU AI Act, and the usual data processing agreements, including technical and organisational measures, are part of the package.

BaseGPT also operates on properly costed fixed prices rather than a token meter. A setup on the company's own infrastructure starts at around €15,000 as a one-off fee. A more involved fine-tuning project with deeper system integration starts at around €25,000. Managed hosting inside Germany is available on request. From scoping to productive rollout, a typical engagement runs to about six weeks.

ThePromptMonk assesses up front, and independently, whether such a solution actually makes sense for a given organisation, both functionally and commercially. The technical platform is never the point in itself. What matters is which processes become easier, which knowledge gaps quietly close, and which decisions stop dragging on longer than they should.

Typical outcomes

Faster onboarding of new employees

Fewer repeated internal questions

Better use of existing documents

Less dependency on individual experts

Structured access to internal knowledge

A safer foundation for further AI applications

A sensible place to start

The most useful first step is rarely a large technology project. It is a proper diagnosis. Which knowledge areas are genuinely worth opening up? Which data may actually be used? Who owns the content? Which processes stand to benefit first? Where do the real risks sit?

ThePromptMonk therefore starts with a compact analysis and a prioritised implementation plan, well before any technology decisions are taken.

Discuss Private AI Readiness

Frequently asked questions

What is a private AI knowledge system?

A private AI knowledge system is a company-owned AI environment that makes internal knowledge from documents, processes, policies, projects and line-of-business systems genuinely usable. Unlike public AI systems, it works with the organisation's own knowledge base, respects roles and permissions, and slots into existing governance, data protection and compliance structures rather than sitting awkwardly beside them.

Which companies is this suitable for?

Private AI knowledge systems suit knowledge-intensive organisations where internal information is scattered, sensitive and quietly business-critical. Typical adopters include tax and audit firms, law firms, management consultancies, engineering firms, software companies, industrial groups and the upper mid-market with multiple sites or client structures.

How does a typical project unfold?

An engagement opens with structured scoping and a data audit, during which the relevant knowledge areas, data sources, use cases and compliance requirements are reviewed together. From there, the team handles data preparation with PII masking and taxonomy alignment, builds a golden evaluation set from real questions, runs the actual fine-tuning with LoRA and QLoRA, refines the retriever and rolls the system out within your own infrastructure. A typical project reaches a productive state within about six weeks, with a first meaningful proof of concept visible after roughly two weeks.

Does our documentation have to be in perfect shape first?

No. The preparation work itself tends to reveal where knowledge is already well documented and where the real gaps sit. The introduction becomes a structured clarification exercise whose value reaches well beyond the technical project alone.

How do public AI systems and private AI systems differ?

Public AI systems draw on general world knowledge but know nothing of your internal documents, your roles, your permissions or your governance. A private AI knowledge system works with company-specific knowledge, returns traceable sources, respects a defined role model and emerges as an organisation-wide knowledge layer rather than an individual tool.

Which data can be connected?

Both structured and unstructured content can be connected, including document repositories, knowledge bases, wikis, ticket systems, ERP, CRM and DMS platforms, client files, technical documentation, policies, contracts, meeting minutes and email archives. Which sources are actually worth connecting follows from the prioritised use cases, the protection requirements and the permission model.

Which compliance requirements need to be considered?

The architecture is aligned with the requirements of the GDPR, in particular Art. 5 and Art. 32, as well as with the relevant provisions of the German BDSG, in particular § 22 and § 26. The EU AI Act is reflected in the risk classification, so that a later assessment and documentation can be carried out on the basis of the existing structures. Data processing agreements in line with Art. 28 GDPR are part of the package, together with the technical and organisational measures, covering access concepts, encryption, audit logs and PII masking.

Is BaseGPT mandatory?

Not in the slightest. ThePromptMonk remains technology-open and evaluates every solution against the concrete requirements. BaseGPT can be a suitable technology partner when requirements, data protection, infrastructure and budget all line up. In other cases, different partners or architectures are brought in.

Which technologies are typically used?

The typical stack combines open-weight language models such as Llama, Mistral, Qwen, DeepSeek, Phi and Gemma in sizes ranging from roughly seven to one hundred and twenty billion parameters, with retrieval-augmented generation, vector databases, fine-tuning techniques such as LoRA and QLoRA and integrated role and audit mechanisms. Processing happens on-premises on your own infrastructure, on bare-metal hardware or fully managed in a German data centre in Frankfurt, without any detour through US clouds.

What investment should be expected?

When delivered with the technology partner BaseGPT, engagements run on clearly calculated fixed prices. A setup on your own infrastructure starts at around 15,000 EUR as a one-off fee, while a more comprehensive fine-tuning engagement with deeper integration into existing systems starts at around 25,000 EUR. Managed hosting in the German data centre is available on request. ThePromptMonk independently assesses upfront whether such an investment is genuinely viable for the situation at hand.

Ready to put internal knowledge properly to work?

Let us look together at whether a private AI knowledge system is the right move for your organisation.

Request an Introductory Call