A capability is a reusable building block. A solution pattern is a bounded composition selected from the current catalog according to problem shape, lifecycle position and operating scale. The goal is not to force every method into every task, but to make the smallest useful stack visible and inspectable.
Explore the combinationsThese patterns are intentionally broader than individual methods. They show how the same capability can contribute to different outcomes depending on context. The listed stacks are typical starting points, not fixed recipes.
Combine a problem, a stage of work and an operating scale to find relevant patterns. Each match shows a possible working output and the capabilities behind it.
Matches are curated starting points, not ranked recommendations. Start small; add capabilities only when the problem calls for them.
No listed pattern matches this combination. Broaden one filter or clear the filters to explore the full map. This is a limit of the map, not a claim that the problem cannot be addressed.
Read across a pattern to see its building blocks, or down a capability to see where it contributes. Filled dots mean included in this pattern, not a score. The rows follow your filters.
| Solution pattern | PISD | IPB | DIAL-4P | SRCB | ORBIT | SORR | CARD & POINTER | DIALECTIC | DIAL-4 | DIAL-4+ | Hermeneutic-didactic | RigVedan | Dialogue lifecycle | 5PP | AICS | Capability-Gap | CSR | GDSA | CBE | CBE-IX |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Governed AI delivery | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Included | Not included | Not included | Included | Included | Included | Not included | Not included | Not included | Not included | Not included |
| Context integrity | Not included | Not included | Not included | Included | Included | Included | Included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included |
| Evidence synthesis | Not included | Included | Not included | Included | Not included | Included | Not included | Included | Included | Included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included |
| Decision support | Included | Included | Included | Not included | Not included | Not included | Not included | Included | Included | Included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included |
| Reuse before create | Not included | Not included | Not included | Not included | Not included | Included | Included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Included | Included | Included | Not included | Not included |
| Portfolio orientation | Not included | Not included | Not included | Not included | Included | Included | Included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Included | Included | Included | Not included | Not included |
| Knowledge architecture | Not included | Not included | Not included | Included | Included | Not included | Included | Not included | Not included | Not included | Included | Included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included |
| Interpretation boundary | Not included | Not included | Not included | Not included | Not included | Not included | Included | Included | Not included | Included | Included | Included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included |
| Bounded activation | Not included | Not included | Not included | Not included | Not included | Not included | Included | Not included | Not included | Not included | Not included | Not included | Included | Included | Included | Not included | Included | Included | Not included | Not included |
| Comparative evaluation | Not included | Included | Not included | Not included | Not included | Included | Not included | Included | Included | Included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included | Not included |
| 11. Bounded capability expansion | Not included | Not included | Not included | Not included | Not included | Included | Not included | Not included | Not included | Not included | Not included | Not included | Included | Included | Not included | Not included | Included | Not included | Included | Included |
Problem: model output, human decision and executable action are being treated as one undifferentiated step.
Outcome: an explicit path from clarification through decision, bounded execution and property-specific verification.
Working output: A workflow with explicit decision gates, bounded instructions and acceptance checks.
Where can a suggestion currently turn into an action without a recorded decision?
Link to this patternProblem: relevant context is distributed across documents, repositories, histories or representations and the system cannot reliably tell what is the same subject.
Outcome: bounded context with explicit identity, relationships, provenance, authority and next routes.
Working output: An identity and source map recording relationships, conflicts and routes to authoritative evidence.
Which sources disagree, and are they actually describing the same subject?
Link to this patternProblem: many sources or interpretations must be reconciled without letting the emerging explanation silently strengthen weak claims.
Outcome: a bounded understanding that keeps source support, independent baselines, competing interpretations and claim strength visible.
Working output: A synthesis with source support, competing explanations and unresolved claims kept visible.
Which conclusion changes if the weakest source is removed?
Link to this patternProblem: a rich request or problem space contains several plausible intents, interpretations or paths and an early answer would erase useful structure.
Outcome: preserve the semantic field first, compare alternatives explicitly, then narrow only when the decision surface is strong enough.
Working output: A decision brief with alternative interpretations, assumptions and criteria for narrowing the options.
What would we lose by committing to the first interpretation?
Link to this patternProblem: a new tool, workflow or method is being proposed before the existing capability estate has been understood.
Outcome: identify existing realizations, reconcile overlaps and select the smallest sufficient intervention before adding another surface.
Working output: A reuse map linking existing capabilities to the need, with overlaps and remaining gaps.
Is the missing piece a capability, a route to it, or permission to use it?
If an existing target genuinely lacks an internal responsibility, continue to bounded capability expansion. CBE and CBE-IX operate after target identity and coverage have been established.
Link to this patternProblem: projects, repositories, copies, public projections and capability representations have become hard to navigate without false identity collapse.
Outcome: a typed map of what exists, how surfaces relate, where authority lives and which capability is actually discoverable.
Working output: A navigable map separating projects, repositories, copies and public representations.
Where does the current authoritative state of each subject live?
Link to this patternProblem: useful meaning must survive compression, routing and re-entry without turning labels or summaries into source truth.
Outcome: compact typed knowledge units, bounded orientation and explicit routes back to deeper evidence and authority.
Working output: Compact knowledge units with explicit relationships and routes back to the supporting sources.
Which meaning must survive compression and a later return to the work?
Link to this patternProblem: understanding, explanation, semantic classification and durable admission risk being collapsed into one representation.
Outcome: preserve the difference between iterative interpretation, didactic projection, semantic classification and claim support.
Working output: An explanation that separates interpretation, teaching choices and the evidence for each claim.
Can a reader distinguish the explanation from the source it interprets?
Link to this patternProblem: a useful capability exists, but discovery, relevance, authorization and actual use are being treated as the same state.
Outcome: move from durable existence to bounded activation while preserving instruction, decision and verification gates.
Working output: An activation path connecting discovery, relevance, authorization and verification.
What must be true before this capability is allowed to act?
Link to this patternProblem: a new method may appear effective because every attempt has already been shaped by the method being evaluated.
Outcome: preserve independent first-pass evidence, compare guided and unguided reasoning, and make differences in claim quality and interpretation inspectable.
Working output: A comparison record preserving independent first-pass evidence and differences in reasoning quality.
Was the baseline recorded before the method influenced the attempt?
Link to this patternProblem: a defined capability has internal gaps, but a proposed extension mixes its responsibilities with work owned by peers.
Outcome: a more complete target design with preserved identity, explicit handoffs and a justified stopping reason.
Working output: a responsibility invariant, classified candidates, boundary and handoff contracts, and an iteration ledger when repeated expansion is justified.
What is still missing for this target to perform its own responsibility? Use CBE for a bounded analysis; choose CBE-IX when accepted changes expose further internal gaps. They are alternatives within a method family, not two mandatory sequential steps. Execute here means constructing the analysis model, boundaries and handoffs; implementation requires separate authorization.
Example: an ingestion pipeline keeps format detection, delegates embeddings and splits determining retention policy from attaching the returned classification. Try the classification lab and iteration lab.
Link to this patternThe current catalog can be read as a set of complementary roles. A real problem may use one role, several roles or a cross-cutting subset. The sequence below is a composition aid, not a mandatory pipeline.
Keep ambiguity, independent baselines and plausible possibilities available long enough to avoid premature collapse.
Acquire bounded context, resolve identity and relationships, and keep representation separate from source authority.
Separate reasoning modes, grade claim strength, iterate interpretations and preserve typed semantic structure.
Separate lifecycle gates, execution discipline and instruction contracts so model output does not silently become completed action.
Find and reuse existing realizations, examine a selected capability for internal gaps, and activate it within explicit authority. Keep discovery, design decisions and authorization distinct.
After discovery, CBE and CBE-IX examine a defined target for internal gaps while preserving its boundaries. This design role does not confer activation authority.
Scale does not automatically require more methods. It changes which responsibilities become load-bearing. These examples show how a stack can expand only when the problem demands it.
CBE separates owned gaps from peer operations. CBE-IX adds a ledger when new accepted responsibilities reveal further gaps. Example: an ingestion agent attaches a returned retention classification while a policy service keeps ownership of the policy decision.
PISD, DIAL-4P and DIAL-4 may be enough when the central risk is premature narrowing inside one discussion.
Dialogue lifecycle, 5PP and AICS enter when decisions become instructions, actions and verifiable outcomes.
ORBIT, SORR, SRCB and CARD & POINTER become important when state and authority are distributed across surfaces.
Capability-Gap, CSR and GDSA enter when the question is no longer only what a system knows, but what capability already exists and can be surfaced responsibly.
The map is useful only if it prevents unnecessary ceremony as well as unnecessary reinvention.