AI-native runtime for organizational knowledge
Work from what your organization knows. Keep it current as you work.
Silkyware keeps your organization's knowledge, its structure and its relationships in one shared, persistent environment. People, AI agents, workflows and applications all work inside it.
Work starts from what is already known. What it produces comes back as proposed changes for your rules and reviewers to check, so keeping knowledge current becomes part of the work itself.
Conceptual view
- Applicationsfor people
- AgentsAI doing the work
- Workflowsyour processes
Governed bydomain modelrulespermissionsreview
Shared context
Knowledge that carries over from one piece of work to the next
Knowledge is often rebuilt for every task: copied into a new document, pasted into a prompt, explained again to the next person. Silkyware keeps it in one place, structured around your own domain model, with the relationships between concepts, decisions and requirements intact.
-
Bring in what exists
Import Word, PDF, Excel and Markdown files. Silkyware extracts the knowledge, structures it and links each item back to the passage it came from.
-
Create knowledge directly
New knowledge can start in Silkyware in structured form, without being drafted in a file first.
-
Use the same knowledge everywhere
People, agents, workflows and applications use the same current knowledge, with its relationships and history attached. Other systems can read it through the API.
Workflows and applications
Work that runs on your knowledge and feeds its results back
You describe your domain in your own vocabulary: the kinds of things you work with, how they relate and which rules must hold. Workflows and agents are configured around that model and the outcomes you need, and can be changed without a software release.
-
1
Start from context
A workflow or agent begins with the relevant knowledge and rules already in place.
-
2
Do the work
AI drafts, structures, compares or checks within the rules of your domain. People step in where judgement is needed.
-
3
Contribute back
Results return to the shared knowledge as proposed changes, ready to be validated and reviewed.
Applications for the people doing the work
Applications give people direct ways into the same environment. They all work with the same shared knowledge, so what is created in one is available to the rest.
- Guided authoringCreate a specific kind of knowledge step by step. AI asks questions and drafts within your rules.
- QuestionsAsk about your organization's knowledge and get answers that cite their sources.
- ReviewSee what a proposed change would add, change or replace, then approve or reject it.
- DocumentsGenerate PDF documents from current knowledge when a process still needs one.
Knowledge lifecycle
Knowledge work is often treated as finished once something has been imported or published. The harder part comes after: changes, reuse, and knowing which version applies.
Creating knowledge and keeping it current are part of the same system.
- create
- structure
- validate
- approve
- version
- maintain
- use
When a source document changes, the knowledge derived from it can be flagged for re-review, and duplicates can be found and merged. Every change takes the same path as new knowledge.
Governance
You decide how knowledge changes and when it can be relied on
Controls sit in the same environment that agents and workflows use, so they apply to changes from people and AI alike.
-
Domain model and rules
Changes are checked against the structure and rules you define. Validation shows whether a change fits your model; whether it is right remains a decision for people.
-
Human oversight
You decide which changes need review and which apply directly, such as routine imports. Reviewers see what a change would add, modify or replace before it takes effect, and each decision is recorded against the person who made it.
-
Sources and history
Every change keeps its author, and reviewed changes keep their approver. Earlier versions are kept, and an approved change can be reverted.
-
Access
Each organization's knowledge is kept separate. Access is granted by role at organization, project or workspace level, for people and for the systems that use it.
In use today
-
Running in production
Silkyware supports day-to-day knowledge work for a public-sector customer.
-
Separate by organization
Each organization works in its own access-controlled environment, with sign-in through Microsoft Entra ID.
-
Estonian and English
The application, and the text the AI writes, work in Estonian or English.
Get in touch
Talk to us about your knowledge and the work around it
Bring a knowledge-heavy process, a register, a methodology or a document set your organization depends on. We will show how it would run in Silkyware, starting from your domain.
We work closely with a small number of organizations at a time.