Content knowledge describes facts and reference information. Process knowledge describes checks, decisions and actions. Keeping their roles distinct helps AI applications use current information within an agreed operating procedure.
Connect what to know with how to proceed
A correct fact is not always enough to complete a business interaction. A team may also need to verify identity, check conditions or escalate an exception. If a procedure is retrieved only in fragments, its prerequisites can be missed. This distinction is a practical knowledge-design approach, not a claim that all RAG systems ignore procedures.
The approach
Technica AI structures procedures and supporting content separately, then links them where needed. Research on cognitive architectures for language agents also distinguishes forms of memory used for facts and action. The business design remains specific to each application.
- Structure the procedure: Define steps, checks, branches, exceptions and handoff conditions.
- Organize reference content: Keep FAQs, manuals and policies retrievable with source information.
- Link the two: Connect steps to the content they require, reducing duplicated facts within procedures.
- Capture experience: Use interviews and permitted records to identify undocumented practices for assessment.
Business value
- More consistent handling
Explicit checks and branches help teams evaluate whether required steps are covered.
- Maintain content separately
Update linked reference information without duplicating every change across procedures; rule changes may still require procedure updates.
- Support authorized action
Use the procedure to guide AI actions and handoffs within configured integrations and permissions.
Application considerations
- Agree the operating scope, exception handling and action permissions before deployment. Test the procedure against representative cases.
- For AIKNOW Answer, proposed knowledge updates undergo review and testing before release. Update controls vary across products and deployments; they should match the impact of the change.
Related products
- AIKNOW Answer
Combines process and content knowledge for configured customer-service automation.
- AISECI
A knowledge-creation framework with a target architecture for capturing and sharing experience; capability scope is agreed for each project.
References
- Sumers, T. R. et al. (2024). Cognitive Architectures for Language Agents. Transactions on Machine Learning Research. arXiv:2309.02427. https://arxiv.org/abs/2309.02427