FAQ: frequently asked questions about Kollegia

Kollegia is an operating system for competence. It connects products, processes, roles, competencies, people and evidence into one continuous data model — instead of six separate silos. Nine digital colleagues work on this structure, each monitoring their own domain and speaking up before a gap becomes a problem.

Kollegia is built for technical companies where competence and quality are business-critical. Regulated industries such as medical technology, calibration and testing, inspection and certification services (TIC) benefit especially, because documentation requirements apply there. But competence transparency, absence cover and structured onboarding are needed by every growing technical team — regardless of the standard.

Kollegia is the connecting layer that classic tools lack. An LMS manages courses, a LIMS samples and measurement data, QM software document versions, HR systems master data — none of these systems link role, competence level, process and evidence together. Kollegia doesn't replace these systems; it closes the gap between them.

No. Kollegia doesn't manage samples, measurement data or test orders. It manages the competence and evidence side: who is allowed to perform which activity, what proves it and how long it stays valid. Kollegia runs alongside a LIMS or without one.

A chain of six links: products (what you deliver) → processes (how it's made) → roles (who owns it) → competencies (what it takes) → people & agents (who can do it) → evidence (what's proven). Change one link and all the others know. That's the difference between a system and a collection of tools.

German, English and Spanish — for distributed teams and locations across borders.

No. Coverage, minimum staffing and evidence validity are operational questions independent of any standard. Those who additionally work to ISO/IEC 17025 will find the matching registers and evidence formats on the ISO 17025 page.

A chatbot answers plausibly, but doesn't know who at your company is allowed to do what, can do what, or has to prove what — and in the end just speeds up the mess. The digital colleagues work on your chain of products, processes, roles and evidence. That's why their answers are your truth, not internet prose. Incidentally, the same structure that makes AI useful also makes your company manageable entirely without AI.

Each digital colleague owns a domain and works on your own data: Marco ISO 17025 and QMS, Antoine competencies and onboarding, Lena knowledge and standards, Lars the processes. They speak up before a gap becomes a problem.

The same permission model applies in every plan: a colleague may never do more than its role allows — and never more than you are allowed yourself. When in doubt, the AI doesn't write, it asks. Every action is logged in the audit trail.

Requests to your digital colleagues consume AI credits; how many depends on the scope of the task.

Emma — Setup. Sets up your system in days, not months.
Antoine — Competencies & Onboarding. Flags stalled onboarding.
Marco — ISO 17025 / QMS. Finds the QMS gap before the auditor.
Lars — Processes. Keeps processes running instead of in folders.
Lena — Knowledge & Standards. Makes knowledge findable before it walks out with the next employee.
Mateo — Commercial. Contracts, NDAs and management accounting on schedule.
Sofia — Organisation. Roles and responsibility kept clean.
Rafael — Router. Routes every request to the right specialist.
Angel — Supervisor. Oversees the entire fleet.

Every digital colleague works within a defined role and permission model: it may never do more than its role allows — and never more than you are allowed yourself. When in doubt, the AI doesn't write, it asks. Every action is logged in the audit trail. This keeps AI use documented and traceable — a requirement the EU AI Act also places on the use of AI systems.

The digital colleague's rights intersected with your own rights determine what may happen. Change the role, and the same action has a different outcome. A colleague can therefore never serve as a back door to obtain permissions that the triggering person doesn't have themselves.

When in doubt, the AI doesn't write — it asks. Unclear cases end in a question to a human, not in a silent write operation.

Every action is logged with timestamp, trigger and executing instance, and is traceable for an assessor.

Both — but never write without approval. They continuously watch their domain and report changes on their own. For every write action, the permission model applies: allowed, limited, logged — and when in doubt, a question rather than a write operation.

Key-person risk exists when a critical process step may only be performed by a single person. If that person is unavailable — through illness, holiday or resignation — the step comes to a standstill. This only becomes visible when you measure coverage per process step instead of qualifications per person.

A single point of failure is a process step for which only one person is demonstrably qualified. If that person is unavailable, the process step comes to a standstill. Kollegia flags such steps automatically as soon as coverage drops to one.

By setting a minimum staffing for each critical process step and continuously comparing it against the actual state. Skill Audit does exactly that: for each step it shows how many people can perform it with valid evidence — and how many there should be.

Key-person risk drops as soon as at least two people are qualified for every critical process step. Kollegia derives concrete qualification actions from the coverage level and prioritises them by risk — critical process steps first.

As a rule of thumb, at least two with independent competence and a third in training. What's decisive, though, is criticality: a step that blocks the order flow needs more coverage than one that can be postponed.

A skills matrix shows what individuals are missing. Skill Audit shows whether the process step is covered. The difference comes from the requirement being derived from products, processes and roles rather than maintained by hand.

An Excel matrix shows people and skills side by side. The Skill Audit ties every skill to a specific process step, making visible which work is actually covered. Instead of organised guesswork, you get a seamless chain from product to evidence.

By building onboarding around the role rather than around the person. The role's target profile remains when the person leaves — and the successor's evidence builds up over the course of onboarding, not in a handover week at the end.

When an employee leaves, Kollegia immediately shows which process steps are affected and which colleagues can demonstrably take them on. SkillMatch suggests candidates with the smallest qualification gap, so knowledge transfer starts before the last working day.

Yes — that's exactly what FlowChart and the Knowledge app are built for. Employees describe their workflow in plain language; from it a documented process emerges with steps, owners, inputs, outputs and tools. Linked to roles, skills and evidence, it stays a living process instead of a PDF no one opens.

Competence is tracked in levels from 0 to 3, where level 3 means independent execution. It's evidenced by proof — training, test, instruction or proficiency test — each with a validity date. In SkillMatch, mentors confirm the actual state instead of employees rating themselves.

Target profile and actual state are continuously compared, including expiry dates. An expiring qualification stands out months in advance instead of only during the audit. Once a piece of evidence has expired, the authorisation attached to it lapses — an approval without valid evidence is not possible.

The standard doesn't name a matrix as a format. It requires that competence requirements per activity are documented, competence is monitored and authorisations are evidenced. A competence or qualification matrix is the form commonly used in practice to demonstrably meet this requirement — and the form assessors expect.

A qualification matrix under ISO/IEC 17025 has to link role, test method, competence level and approval authorisation. A spreadsheet can't do this: it keeps no history of changes and monitors no deadlines. Kollegia connects roles, skills, processes and evidence into a live system — the org chart becomes a living competence map with validities and approvals.

A training record proves attendance, a competence record proves ability. The standard targets the latter: it's not enough that someone attended training — the lab must assess and document that the person actually masters the activity.

For a snapshot yes, for maintaining evidence rarely. Excel has no expiry dates, no version history and no link between evidence and approval authorisation. In practice several versions exist in parallel, and before the assessment the scramble begins.

Chapters 4 to 8 are mapped as 32 evidence registers and assigned to the standard's chapters — structure, resources, processes and management system. The registers are in place before you start. You fill them during operations, not before the assessment.

A digital colleague acts exclusively within the rights of the person it works for, and when in doubt, it asks instead of writing. Every action is logged with timestamp, trigger and executing instance, and is traceable for an assessor.

Yes. The mechanism — requirement per activity, evidenced proof, validity, authorisation — is independent of any standard and works just as well under ISO 9001, ISO 13485 or internal requirements. ISO/IEC 17025 is the most demanding case, not the only one.

ISO/IEC 17025 is fully mapped as a register structure today. The mechanism is built to work with any standard, and more standards will follow. If you need a specific standard, tell us — it feeds directly into the roadmap.

Yes, export as PDF and Excel is possible.

Of three. Company OS answers how your company creates value — products, processes, knowledge. Organisation OS answers who owns the work — roles, teams, people, permissions, including AI agents as staff with rights. Competency OS answers who can do what, proven — skills, tests, training, onboarding, SkillMatch, Skill Audit.

Six apps are live: Skills (Team Antoine), Processes (Team Lars), Knowledge (Team Lena), Organisation (Team Sofia), Compliance (Team Marco) and Commercial (Team Mateo). Marketing and Sales are in progress — we only announce what's actually coming.

No. You start with one app or with all of them. Every app — today and in future — runs under the same permission model: permission intersection, fail-closed, audit trail. Your governance grows automatically with it.

Integrations geared to SharePoint, HR systems and DATEV. Your own agents can be connected via the Model Context Protocol (MCP).

Yes, Microsoft Entra SSO — sign in via your existing company identity, including provisioning patterns.

Yes. Kollegia is multi-tenant: multiple organisations stay cleanly separated, with a sandbox for safe experimentation alongside production.

Role-based permissions plus processing rights at record level, with permissions down to area level. Control without freezing the organisation.

The app mechanism is built to be extensible, and your own agents can be connected via MCP. Suggestions for new apps feed directly into the roadmap.

Emma guides you through the setup: base structure, products, processes, roles, skills. Existing roles, employee data and qualification matrices can be imported. As a cloud solution, there's no on-premise IT installation. Instead of months of configuration, you have a usable system state in days — and see your first gap analysis.

In three steps. First: register for free and open your own workspace, no credit card. Second: Emma guides the setup — products, processes, roles and skills get structured, registers and levels are preconfigured. Third: the first gap analysis shows which competencies are in place, where the gaps are and where the risks lie.

Existing roles, employee data and qualification matrices can be imported.

No. Kollegia is a cloud solution, no on-premise installation needed. For sign-in, your existing company identity can be used via Microsoft Entra SSO.

Emma does the setup, not your QM officer on the side. Registers, levels and roles are preconfigured — no blank sheet. Procedures and target profiles are defined in a guided conversation.

AI credits are the billing unit for all AI-assisted functions in Kollegia — such as document analyses, Skill Audits, reports or requests to your digital colleagues. Each action consumes a different number of credits depending on its scope.

At the start of each billing period, your full monthly allowance is available again. Unused monthly credits are not carried over into the following month.

In the Pro and Max plans, you top up additional credits flexibly — either as a credit package or usage-based via pay-as-you-go. Your monthly allowance is always consumed first; topped-up credits only after that. They remain available even beyond the current billing period.

All functions without an AI element remain fully usable in any case. In the Free plan, top-ups aren't possible; the allowance is available again the following month.

Free includes the basic models. Pro adds the standard models. Max additionally includes the top models for particularly demanding analyses. Higher-tier models deliver more precise results on complex documents and consume correspondingly more credits per action.

Yes, anytime. A change takes effect immediately — for upgrades as well as downgrades. The new plan is billed in full from the switch. On a downgrade, the new plan's monthly allowance is available from that point; the higher allowance of the previous plan lapses.

Credit packages already purchased carry over into the new plan. Only when switching to the Free plan does your purchased balance rest — it becomes usable again as soon as you book Pro or Max again. The validity of topped-up credits is governed by our Terms.

Enterprise suits organisations with special requirements — such as custom credit allowances, particular security or compliance requirements, or their own integrations. We develop a plan tailored precisely to your processes and get back to you within 24 hours.

No. The Free plan starts without a credit card and without a contract commitment.

With annual payment you save up to two months. The Free plan is also free of charge on an annual basis.

The website and platform are hosted on servers of Hetzner Online GmbH in Germany. All data processed in operations is stored on servers in Germany. A data processing agreement pursuant to Art. 28 GDPR is in place with Hetzner.

Your data is stored in a GDPR-compliant way in your own tenant, separated from other organisations. For AI features, EU-friendly provider options are available — under your own key governance. Microsoft Entra SSO is supported. Details on data processing and technical measures can be found in the privacy policy.

Hetzner Online GmbH for hosting and server infrastructure with data location in Germany, Microsoft Ireland Operations Limited for transactional email delivery, ALL ONE for CRM and platform functions, and OpenAI Ireland Ltd. for AI-assisted functions. A data processing agreement pursuant to Art. 28 GDPR is in place with all of them. Transfers to third countries take place on the basis of EU standard contractual clauses.

Data transfer takes place exclusively encrypted via HTTPS/TLS. Technical and organisational measures are implemented that protect data against loss, destruction, manipulation and unauthorised access. Employees and service providers are contractually obligated to comply with data protection regulations.

Every action by a digital colleague stays documented and traceable — a requirement the EU AI Act also places on the use of AI systems.

Kollegia is developed by Labskills GmbH, based in Munich, founded by Shahab Houshangi and Philip Terhorst.

Leading teams shouldn't depend on memory or guesswork. When staffing changes, knowledge gaps appear or workflows stall, team leaders need to know exactly where to look. Kollegia isn't another tool add-on, but a method for ensuring stability.

By email at info@kollegia.ai, by phone at +49 89 356479-38 or via the contact form. We get back to you within 24 hours.