01 / DOMAINS · WHERE MAINTAINED STATE CREATES VALUE
One state layer. Different applications.
Whether equipping customer support with instant verified context, powering multi-preference personalization, or feeding structured memory to autonomous AI agents, ClueState keeps entity state current without destroying provenance.
APPLICATION 01 / CUSTOMER SERVICE
Give support teams context before they ask again.
Bring a customer’s reported issues, account tier, communication preferences and previous resolutions into the support workflow — with verified source quotes available when a detail needs checking.
✔ Eliminates repetitive questions on customer history
✔ Preserves verified timestamps of when issues occurred
✔ Connects directly into ticket routing via REST API
SUPPORT CONTEXT SPECIMEN // CUST_402
account_tier:
Enterprise / Dedicated VPC
Verified quote · char_start: 12 · Contract #481
preferred_channel:
Slack Connect (#cluestate-acme)
Verified quote · char_start: 94 · Ticket #8841
prior_resolution:
Database connection timeout resolved via pool tune
Resolved 2026-10-31 · Incident #102
MULTI-CARDINALITY AFFINITY GRAPH
brand_preferences (policy: append):
["Nike", "Adidas", "Salomon"]
Coexisting active claims accumulated across 3 purchases
sizing_category (policy: replace):
US 10.5 Men
Single cardinality · Last verified 2026-11-01
APPLICATION 02 / PROFILES
Remember preferences without flattening them.
Maintain coexisting preferences, affinities, sizing or price sensitivities as they emerge across interactions, instead of treating every new observation as a destructive replacement.
✔ Topic schemas dictate whether attributes accumulate or replace
✔ Avoids erratic recommendation shifts based on one stray interaction
✔ Retains event-time confidence on each affinity node
APPLICATION 03 / AI AGENTS
Give agents structured context, not just more transcript.
Let an autonomous AI agent retrieve a deterministic entity summary through the API or native Model Context Protocol (MCP) tools, with claims and verified quotes available for inspection.
✔ cluestate_get_intel returns both JSON and pre-formatted prompt blocks
✔ cluestate_list_memories allows surgical semantic state exploration
✔ Eliminates LLM hallucination of user state and facts
MCP TOOL // cluestate_get_intel
// Call from Claude, Cursor, or AGY
call_tool(
"cluestate_get_intel"
, { external_id:
"cust_402"
})
// Formatted prompt block injected directly into system context:
FIRST-PARTY APPLICATION SPECIFICATION
PERSONAL CONTEXT & AI EXECUTION
Elevist: An AI assistant needs to understand the person, not just the latest prompt.
In Elevist, founder-built AI execution software, goals, habits, notes and weekly reflections create a changing picture of what matters to a person. ClueState acts as the structured state layer behind this experience — turning notes and check-ins into typed, source-linked user context that Elevist’s AI workflows can retrieve.
current_focus
single · replace
Maintains one active focus; newer valid focus replaces prior value.
active_goals
multiple · append
Preserves concurrent objectives across weekly planning check-ins.
active_blockers
multiple · review
Quarantines ambiguous friction points with source notes for review.
user_preferences
multiple · append
Durable working styles, communication habits, and pacing guidelines.
ARCHITECTURAL BOUNDARY: Elevist remains the system of action for operational goal, task, and metric records. ClueState maintains the extracted, durable context that informs AI assistance.