A document, decision, review, case, or operating brief that teams repeatedly produce.
Use cases by domain
One operating method. Different expert work.
Kautilyan does not apply one generic agent to every function. Each domain starts with its own evidence, terminology, expert judgment, review standard, and consequences. Skillification captures that expertise; the Harness governs how it is used.
The common architecture
What stays consistent across domains.
The subject matter changes. The operating discipline does not.
Domain map
See the work, skills, controls, and outputs together.
Only implementation and delivery is presented as a current reference build. The remaining domains are designed applications that must pass evidence, ownership, and evaluation gates before deployment.
Reference build
Explore this domain →Implementation and delivery
Turn senior solution and delivery judgment into stronger first drafts and more consistent review across projects.
Recurring work
- Discovery and requirements
- BRDs, FRDs, solution designs
- Configuration, tests, SOPs, training
Example skills
- Evidence extraction and terminology
- Domain and solution architecture
- Authoring and independent review
Harness controls
- Source-to-claim traceability
- Requirement and decision gates
- Client-ready release approval
Reviewable outputs
- Requirements and decision packs
- Buildable functional designs
- Quality-scored delivery documents
Designed application
View commercial use cases →Revenue and commercial operations
Convert fragmented commercial signals into human-reviewed forecasts, risk briefs, and next actions.
Recurring work
- Sales forecasting and S&OP inputs
- Account and dealer risk reviews
- Leadership reporting and follow-up
Example skills
- Signal and variance interpretation
- Relationship-risk analysis
- Executive brief authoring
Harness controls
- Source validation across CRM and ERP
- Commercial approval thresholds
- Decision and override capture
Reviewable outputs
- Forecast and variance brief
- At-risk account context pack
- Approved next-action workflow
Designed application
See the diagnosis method →Operations and supply chain
Make exceptions, constraints, handoffs, and operating signals visible before they become service or cash failures.
Recurring work
- Exception and constraint reviews
- Planning, allocation, and fulfilment
- Operational control and RCA packs
Example skills
- Process and bottleneck diagnosis
- Exception classification
- Root-cause and control analysis
Harness controls
- System-of-record reconciliation
- Authority and escalation rules
- Evidence-backed exception routing
Reviewable outputs
- Operating reality brief
- Exception recommendation pack
- Human-approved resolution workflow
Designed application
Read the engineering view →Product and engineering
Keep customer outcome, architecture, verification, and adoption connected instead of reducing delivery to an agent graph.
Recurring work
- Problem and requirement definition
- Architecture and design review
- Test, release, and adoption readiness
Example skills
- Outcome and requirement discipline
- Architecture and threat review
- Acceptance and evaluation design
Harness controls
- Requirement-to-evidence traceability
- Independent review and test gates
- Human approval for consequential change
Reviewable outputs
- PRD and solution note
- Architecture decision record
- Release evidence and readiness pack
Designed application
Check fit criteria →Professional services and assurance
Codify review methods without hiding professional judgment or presenting unsupported conclusions as fact.
Recurring work
- Workpapers and reconciliation packs
- Policy and evidence reviews
- Research, proposals, and client reports
Example skills
- Evidence sufficiency assessment
- Policy and consistency checking
- Finding and report authoring
Harness controls
- Claim-level citations
- Reviewer independence
- Issue severity and release gates
Reviewable outputs
- Evidence-backed workpaper
- Exception and review notes
- Partner-ready client draft
Designed application
See the governance model →Regulated case operations
Assemble case evidence and recommendations while keeping policy interpretation and consequential decisions with authorized humans.
Recurring work
- Claims and denial evidence packs
- Case summaries and document checks
- Exception and escalation handling
Example skills
- Document sufficiency checking
- Policy-grounded case analysis
- Next-action recommendation
Harness controls
- Permitted-data boundaries
- Mandatory expert review
- Decision, reason, and audit capture
Reviewable outputs
- Cited case summary
- Missing-evidence request
- Human-approved recommendation
Qualification test
Not every process should become a skill system.
A candidate becomes stronger as its recurrence, review burden, evidence availability, and error cost rise. Highly novel work, unobservable quality, or missing process ownership should be mapped before it is automated.
- Does the work recur across people, projects, or customers?
- Can an expert explain what makes the output acceptable?
- Can the system access the permitted evidence?
- Can a human review or reverse consequential action?
- Is there a measurable cost of delay, rework, or inconsistency?
Bring the artifact your experts keep rewriting or reviewing.
We will separate what should become a reusable skill, what the Harness must govern, what must remain human, and how the first evaluation should be designed.