{ "$schema": "https://auxfirst.com/capabilities.schema.json", "specVersion": "0.1", "name": "auxfirst", "displayName": "auxfirst — Agentic Experience Design", "url": "https://auxfirst.com/", "contactUrl": "https://auxfirst.com/index.html#contact", "description": "auxfirst is an Agentic Experience Design (AUX) agency. We design AI systems that remember, adapt, and earn trust — for product, developer, and leadership teams building agents in 2026. This document is the machine-readable surface of what auxfirst can do, for AI agents, automation systems, and MCP-ready tooling clients.", "license": "Definitions: CC BY 4.0 · TrustKit tooling: MIT", "lastReviewed": "2026-06-01", "discovery": { "llmsTxt": "https://auxfirst.com/llms.txt", "agentSkills": "https://auxfirst.com/.well-known/agent-skills/index.json", "apiCatalog": "https://auxfirst.com/.well-known/api-catalog", "mcpServerCard": "https://auxfirst.com/.well-known/mcp/server-card.json", "openapi": "https://auxfirst.com/openapi.json", "auth": "https://auxfirst.com/auth.md", "markdownHomepage": "https://auxfirst.com/index.md", "health": "https://auxfirst.com/api/health.json" }, "authentication": { "supported": false, "note": "No OAuth/OIDC, agent registration, or protected API yet. All discovery resources are public and unauthenticated. See /auth.md.", "doc": "https://auxfirst.com/auth.md" }, "mcp": { "liveServer": false, "discoveryCard": "https://auxfirst.com/.well-known/mcp/server-card.json", "note": "Resources-only discovery card; no live tool-serving transport yet. Standing up an MCP server is itself one of auxfirst's service offerings." }, "publisher": { "name": "auxfirst agency", "url": "https://auxfirst.com/", "founder": "Emil Krzemiński", "github": "https://github.com/auxfirst", "linkedin": "https://www.linkedin.com/company/auxfirst/", "substack": "https://auxfirst.substack.com/" }, "audience": [ "product_teams", "engineering_teams", "platform_teams", "executive_leadership", "holdco_strategy", "ad_agencies", "media_buyers", "family_offices", "legacy_saas_teams" ], "domains": [ "agentic_user_experience", "agent_experience", "agent_first_design", "model_context_protocol", "tool_calling_architecture", "agent_memory_systems", "trust_architecture", "multi_agent_orchestration", "agent_enablement", "agentic_advertising", "family_office_governance", "legacy_saas_retrofit" ], "capabilities": [ { "id": "agentic_api_experience", "type": "practice", "name": "Agentic API Experience", "tagline": "AX — the new DX. The API layer designed for the agent first.", "description": "Design and build agent-first developer experiences — from tool-calling architecture to MCP integration and composable agent pipelines. Where the agent meets your software, designed for the agent's reasoning loop, not the developer's stack trace.", "url": "https://auxfirst.com/agentic-experience-center/agentic-api-experience.html", "documentUrl": "https://auxfirst.com/agentic-api-experience.md", "inputs": [ "existing_api_specification", "openapi_or_graphql_schema", "current_tool_catalogue", "agent_target_use_cases", "autonomy_targets_per_action" ], "outputs": [ "action_layer_specification", "consequence_schema_per_action", "tool_descriptor_set_mcp_ready", "error_recovery_taxonomy", "permission_topology", "implementation_brief" ], "status": "launching_next", "tags": ["mcp", "tool-calling", "agent-experience", "ax", "api-design"] }, { "id": "ai_product_experience", "type": "practice", "name": "AI Product Experience", "tagline": "AUX patterns instantiated as shipped product, not slideware.", "description": "Transform a SaaS from a static tool into an adaptive partner. We design the interaction paradigms, generative UI, and trust scaffolding that makes AI products defensible.", "url": "https://auxfirst.com/agentic-experience-center/ai-product-experience.html", "documentUrl": "https://auxfirst.com/ai-product-experience.md", "inputs": [ "product_url_or_demo", "current_user_journeys", "target_user_segments", "shipped_ai_feature_inventory", "trust_or_adoption_metrics" ], "outputs": [ "trust_canvas_for_product", "interaction_blueprint", "pattern_instantiation_spec", "confidence_cue_system", "memory_surface_design", "escape_hatch_inventory" ], "status": "launching_next", "tags": ["aux", "generative-ui", "trust-design", "intent-handshake", "confidence-cues"] }, { "id": "agent_process_design", "type": "practice", "name": "Agent Process Design", "tagline": "Turning a workflow into a defensible agentic service.", "description": "Two lenses run in parallel — the aggregate workforce view a CTO uses on Monday morning, and the individual-agent design a product team ships. Ten stages from workflow analysis to audit trail. Every deliverable is a clickable HTML mockup.", "url": "https://auxfirst.com/agentic-experience-center/agent-process-design.html", "documentUrl": "https://auxfirst.com/agent-process-design.md", "inputs": [ "workflow_description", "current_actors_and_systems", "business_outcomes_and_kpis", "regulatory_constraints", "blast_radius_per_action", "named_human_approvers" ], "outputs": [ "agent_home_mockup_html", "agent_map_node_graph", "agent_inventory_and_roadmap", "trust_canvas_yaml_policy", "storyboard_walkthrough", "runtime_state_diagram", "build_spec_and_tool_catalogue", "evaluation_harness", "audit_log_schema" ], "status": "live", "tags": ["process-design", "workforce-view", "trust-canvas", "runtime-spec", "audit-trail"] }, { "id": "agent_experience_audit", "type": "engagement", "name": "Agent Experience Audit", "tagline": "Identify where an AI product breaks trust — and how to fix it.", "description": "A structured diagnostic of a shipped product, assistant, or workflow against the AUX framework. We analyse behaviour across intent understanding, memory and context handling, transparency and explainability, control and fallback, and autonomy clarity.", "url": "https://auxfirst.com/index.html#engagements", "duration": "3 weeks", "inputs": [ "product_url_or_internal_access", "workflow_description", "target_user_segments", "known_failure_modes", "shipped_ai_feature_inventory" ], "outputs": [ "trust_gap_inventory", "trust_scorecard", "ux_risk_map", "pattern_by_pattern_recommendations", "prioritised_remediation_roadmap" ], "bestFor": ["teams_already_shipping_ai_features", "copilots_with_stalled_adoption"], "status": "live", "tags": ["audit", "diagnostic", "trust-gaps", "remediation"] }, { "id": "blueprint_sprint", "type": "engagement", "name": "Blueprint Sprint", "tagline": "Three weeks to a build-ready agent-first surface specification.", "description": "Pick one workflow. Walk out with the agent home mockup, the trust canvas, the storyboard, the runtime spec, and the build brief — everything engineering needs to start building Monday.", "url": "https://auxfirst.com/index.html#engagements", "duration": "3 weeks", "inputs": [ "single_workflow_scope", "business_outcome_target", "current_actors_and_systems", "autonomy_boundary_constraints" ], "outputs": [ "process_kpi_translation", "trust_canvas_yaml_policy", "storyboard_walkthrough_html", "runtime_state_diagram", "build_ready_engineering_brief" ], "bestFor": ["product_teams_ready_to_build", "first_agentic_surface"], "status": "live", "tags": ["sprint", "blueprint", "build-spec"] }, { "id": "advisory_retainer", "type": "engagement", "name": "Advisory Retainer", "tagline": "Ongoing support as the agentic surface evolves.", "description": "The Blueprint Sprint outcome extended across the agentic workforce. We sit alongside your AI council, design successive agents on the same template, evolve the aggregate view, and own the consistency of the deliverable as the surface grows.", "url": "https://auxfirst.com/index.html#engagements", "duration": "quarterly_or_annual", "inputs": [ "ai_council_membership", "agent_roadmap", "trust_or_adoption_metrics", "regulatory_or_audit_calendar" ], "outputs": [ "per_quarter_agent_designs", "aggregate_view_maintained", "audit_and_eval_cadence", "ai_council_presence" ], "bestFor": ["multi_quarter_programs", "holdco_or_enterprise_rollouts"], "status": "live", "tags": ["retainer", "ongoing", "ai-council"] }, { "id": "executive_seminar", "type": "engagement", "name": "Executive Seminar", "tagline": "Align leadership on what agentic systems actually mean.", "description": "A focused session for founders, executives, and product leaders to build a shared understanding of what 'agentic' really implies beyond hype, how trust, memory, and autonomy reshape product design, where the current product or strategy stands, and what decisions need to be made next.", "url": "https://auxfirst.com/index.html#engagements", "duration": "half_day_or_full_day", "inputs": ["leadership_attendees", "current_product_or_strategy_summary"], "outputs": [ "shared_aux_mental_model", "risks_opportunities_blind_spots", "prioritised_strategic_directions", "decision_framework" ], "bestFor": ["leadership_alignment", "early_stage_exploration", "strategic_resets"], "status": "live", "tags": ["seminar", "leadership", "executive"] }, { "id": "team_workshop", "type": "engagement", "name": "Team Workshop", "tagline": "Turn abstract AI concepts into concrete product decisions.", "description": "A hands-on working session with product, design, and engineering teams to translate agentic principles into your actual product context. Theory → application → decisions.", "url": "https://auxfirst.com/index.html#engagements", "duration": "1_to_3_days", "inputs": ["cross_functional_team_attendees", "product_context", "candidate_workflows"], "outputs": [ "mapped_agent_user_journeys", "defined_trust_patterns", "initial_memory_and_context_strategy", "aligned_design_principles" ], "bestFor": ["teams_preparing_to_design_or_redesign_ai_features"], "status": "live", "tags": ["workshop", "hands-on", "cross-functional"] }, { "id": "mcp_readiness_assessment", "type": "engagement", "name": "MCP Readiness Assessment", "tagline": "Is your API ready for agentic interaction through MCP?", "description": "Assess whether a product, API, or workflow is ready for agentic interaction through MCP, tool calling, or structured integration. Maps gaps, recommends tool schemas, and produces a phased implementation roadmap.", "url": "https://auxfirst.com/ad-industry/mcp-for-advertising.html", "inputs": [ "api_documentation", "current_tool_catalogue", "workflow_description", "integration_goals" ], "outputs": [ "mcp_readiness_score", "integration_gaps", "recommended_tool_schema", "implementation_roadmap" ], "status": "live", "tags": ["mcp", "readiness", "tool-schema", "integration"] }, { "id": "agent_trust_architecture", "type": "framework", "name": "Agent Trust Architecture", "tagline": "The four-stage progression model of agent trust.", "description": "Functional → Contextual → Judgment → Advocacy. A model for how the relationship between a human and an agent matures over time. The stages are sequential. Used as the diagnostic spine of every audit and Sprint.", "url": "https://auxfirst.com/heuristics.md", "documentUrl": "https://auxfirst.com/knowledge.md", "stages": [ { "id": "functional", "name": "Functional Trust", "question": "Can it complete basic tasks reliably?" }, { "id": "contextual", "name": "Contextual Trust", "question": "Does it understand nuance, preferences, history?" }, { "id": "judgment", "name": "Judgment Trust", "question": "Can it make good calls in ambiguous situations?" }, { "id": "advocacy", "name": "Advocacy Trust", "question": "Will it act in my best interest, even when incentives misalign?" } ], "license": "CC BY 4.0", "tags": ["framework", "trust", "maturity-model"] }, { "id": "aux_heuristics", "type": "framework", "name": "The 10 AUX Heuristics", "tagline": "The AUX answer to Nielsen Norman's usability heuristics.", "description": "Ten evaluation heuristics for agentic products — optimised for quality of relationship rather than clarity of interface.", "url": "https://auxfirst.com/heuristics.html", "documentUrl": "https://auxfirst.com/heuristics.md", "items": [ { "id": "H01", "name": "Visibility of Agent Intent & Action" }, { "id": "H02", "name": "Progressive Transparency" }, { "id": "H03", "name": "User Control Through Steering" }, { "id": "H04", "name": "Trust Is Dynamic, Not Static" }, { "id": "H05", "name": "Clear Boundaries of Autonomy" }, { "id": "H06", "name": "Graceful Handling of Uncertainty" }, { "id": "H07", "name": "Appropriate Assertiveness" }, { "id": "H08", "name": "Context Efficiency & Awareness" }, { "id": "H09", "name": "Multi-Actor & Multi-Agent Clarity" }, { "id": "H10", "name": "Consistency of Behavior, Not Interface" } ], "license": "CC BY 4.0", "tags": ["heuristics", "evaluation", "aux"] }, { "id": "aux_patterns", "type": "framework", "name": "The 6 Foundational AUX Patterns", "tagline": "The interaction primitives of agentic products.", "description": "Six repeatable patterns for shipping agentic behaviour. Each carries a shape, a contract, a failure mode, and a trigger. Extended catalogue: 24 patterns across four families.", "url": "https://auxfirst.com/patterns.md", "items": [ { "id": "P01", "name": "Intent Handshake" }, { "id": "P02", "name": "Confidence Cues" }, { "id": "P03", "name": "Adaptive Canvas" }, { "id": "P04", "name": "Escape Hatch" }, { "id": "P05", "name": "Memory in Motion" }, { "id": "P06", "name": "Generative Momentum" } ], "extendedCatalogueUrl": "https://auxfirst.com/agent-first-design-patterns.html", "license": "CC BY 4.0", "tags": ["patterns", "ux", "design-primitives"] }, { "id": "autonomy_spectrum", "type": "framework", "name": "The Autonomy Spectrum", "tagline": "Five modes per agent responsibility.", "description": "Per-responsibility autonomy levels with required patterns at each rung. The placement is rarely uniform — bidding can be autonomous, brand-adjacent creative cannot.", "url": "https://auxfirst.com/knowledge.md", "modes": [ { "id": "human_only", "name": "Human-only", "behaviour": "Suggests, never acts" }, { "id": "approve_each", "name": "Approve-each-action", "behaviour": "Drafts and queues; human approves every move" }, { "id": "review_before_act", "name": "Review-before-act", "behaviour": "Batches for review window" }, { "id": "act_and_notify", "name": "Act-and-notify", "behaviour": "Acts immediately, logs for audit" }, { "id": "autonomous", "name": "Autonomous", "behaviour": "Acts within policy" } ], "license": "CC BY 4.0", "tags": ["autonomy", "control-vs-autonomy", "trust-spectrum"] }, { "id": "trustkit", "type": "open_source", "name": "TrustKit", "tagline": "The open-source canon. YAML schemas for the agentic era.", "description": "Heuristics, trust architecture, trust-gap taxonomy, agent specs, memory policies — all as YAML schemas. Forkable, citable, no vendor lock-in.", "url": "https://github.com/auxfirst/trustkit", "license": "MIT (tooling) · CC BY 4.0 (definitions)", "languages": ["yaml"], "schemas": [ "aux-heuristics.yaml", "trust-architecture.yaml", "trust-gap-taxonomy.yaml", "trust-contract.yaml", "agent-spec.schema.yaml", "memory-policy.yaml" ], "status": "live", "tags": ["open-source", "github", "yaml", "schemas"] }, { "id": "trust_canvas", "type": "open_resource", "name": "The Trust Canvas", "tagline": "A one-page artefact for mapping an agent's trust topology.", "description": "Walks a team through autonomy boundaries, handoffs, pattern applicability, and trust-stage targeting. Lean Canvas, for AUX. Free on Figma Community.", "url": "https://www.figma.com/community/file/1635742721765886887", "license": "Free · attribution requested", "useFor": ["greenfield_design", "audit_overlay", "cross_team_alignment"], "status": "live", "tags": ["figma", "canvas", "free-resource"] }, { "id": "agent_enablement_framework", "type": "framework", "name": "Agent Enablement", "tagline": "The operational layer between a capable model and a dependable product.", "description": "The discipline of building the context, tools, playbooks, and feedback loops required to perform reliably against real business goals. Four pillars: Grounding & Commissioning, Playbooks & Operating Patterns, The Enablement Stack, Feedback Loops & Run Forensics.", "url": "https://auxfirst.com/news/agent-enablement.html", "pillars": [ { "id": "01", "name": "Grounding & Commissioning" }, { "id": "02", "name": "Playbooks & Operating Patterns" }, { "id": "03", "name": "The Enablement Stack" }, { "id": "04", "name": "Feedback Loops & Run Forensics" } ], "maturityLadder": ["improvised", "provisioned", "observed", "evaluated", "compounding"], "license": "CC BY 4.0", "tags": ["enablement", "operational-layer", "evaluation", "run-forensics"] }, { "id": "agency_brains_design", "type": "framework", "name": "Agency Brains Design", "tagline": "Brains, plural, before agents. The cognition layer for ad agencies.", "description": "The working framework for defining, naming, and governing the AI brains an ad agency runs. Each brain has a declared reasoning type (deterministic, probabilistic, generative, retrieval-augmented), a manifest, an owner, and an audit cadence.", "url": "https://auxfirst.com/ad-industry/agency-brains-design.html", "implementationAtlas": "https://auxfirst.com/ad-industry/agency-brains-design-02.html", "license": "CC BY 4.0", "tags": ["agency", "advertising", "cognition", "brains", "manifest"] }, { "id": "ad_industry_advisory", "type": "industry_package", "name": "Advertising Industry Advisory", "tagline": "End-to-end agentic transformation for ad holdings, media agencies, brands, publishers, and platforms.", "description": "The Trustworthy Agents ebook, the 120 Agent Index, the 50 Specialist Stack for agent-to-agent advertising, MCP for Advertising, and Agency Brains Design. Read in sequence or commission an engagement.", "url": "https://auxfirst.com/ad-industry/index.html", "resources": [ { "name": "Trustworthy Agents ebook", "url": "https://auxfirst.com/trustworthy-agents-ebook.html" }, { "name": "The Agent Index — 120", "url": "https://auxfirst.com/ad-agent-index.html" }, { "name": "The Specialist Stack — 50", "url": "https://auxfirst.com/ad-industry/specialist-stack.html" }, { "name": "MCP for Advertising", "url": "https://auxfirst.com/ad-industry/mcp-for-advertising.html" }, { "name": "Agency Brains Design", "url": "https://auxfirst.com/ad-industry/agency-brains-design.html" } ], "audience": ["holding_companies", "media_agencies", "creative_networks", "publishers", "ad_platforms"], "status": "live", "tags": ["advertising", "media", "holdco", "agency", "brand"] }, { "id": "family_office_agentic_governance", "type": "industry_package", "name": "Family Office Agentic Governance", "tagline": "Bounded agents for wealth operations.", "description": "Seven agents worth building first for a family office — Governance & Mandate, Entity & Obligations, Advisor Orchestration, Liquidity & Capital Operations, Investment Decision Support, Risk Privacy & Security, Stewardship. Three-engagement programme.", "url": "https://auxfirst.com/agentic-family-office.html", "engagements": [ { "name": "Agentic Readiness Audit", "duration": "2 weeks" }, { "name": "Design Sprint", "duration": "6 weeks" }, { "name": "Managed Retainer", "duration": "12 months" } ], "license": "CC BY 4.0", "tags": ["family-office", "wealth", "governance", "fiduciary"] }, { "id": "legacy_saas_retrofit", "type": "industry_package", "name": "Legacy SaaS Retrofit", "tagline": "Convert a copilot-era SaaS into an agent-first product without a rewrite.", "description": "The retrofit discipline — layering an agent loop on top of an existing platform so the agent's surface becomes the primary loop and the legacy CRUD UI becomes a fallback. Seven-phase process from bolt-on to primary surface.", "url": "https://auxfirst.com/agent-first-design-legacy-saas-retrofits.html", "phases": ["bolt_on", "queue", "routine_acts", "expanded_scope", "primary_surface"], "license": "CC BY 4.0", "tags": ["retrofit", "saas", "migration", "agent-first"] } ], "knowledgeBase": { "url": "https://auxfirst.com/knowledge.md", "concepts": [ "agentic_user_experience_aux", "agent_experience_ax", "agent_first_design", "model_context_protocol", "tool_calling_architecture", "agent_memory_systems", "trust_architecture", "trust_gap_taxonomy", "autonomy_spectrum", "three_tier_autonomy_frame", "aux_evolution_curve", "human_agent_interaction_hai", "generative_ui", "multi_agent_orchestration", "agent_orchestration_ux", "autonomous_user_experience", "intelligent_interface_design", "agent_native_ux", "agent_driven_interfaces", "agent_enablement", "run_forensics", "failure_cards", "evaluation_harness", "agency_brains", "agent_to_agent_advertising", "bounded_agents" ] }, "documents": { "llmsTxt": "https://auxfirst.com/llms.txt", "manifesto": "https://auxfirst.com/manifesto.md", "services": "https://auxfirst.com/services.md", "patterns": "https://auxfirst.com/patterns.md", "heuristics": "https://auxfirst.com/heuristics.md", "knowledge": "https://auxfirst.com/knowledge.md", "agenticApiExperience": "https://auxfirst.com/agentic-api-experience.md", "aiProductExperience": "https://auxfirst.com/ai-product-experience.md", "agentProcessDesign": "https://auxfirst.com/agent-process-design.md" }, "pillarGuides": [ { "name": "Agent-First Design", "url": "https://auxfirst.com/agent-first-design.html" }, { "name": "Agent-First Design Patterns", "url": "https://auxfirst.com/agent-first-design-patterns.html" }, { "name": "Agent-First Design for Advertising Platforms", "url": "https://auxfirst.com/agent-first-design-advertising-platforms.html" }, { "name": "Agent-First Design for Legacy SaaS Retrofits", "url": "https://auxfirst.com/agent-first-design-legacy-saas-retrofits.html" }, { "name": "Agentic Family Office", "url": "https://auxfirst.com/agentic-family-office.html" } ], "references": [ { "name": "Managed Agent Platforms", "url": "https://auxfirst.com/managed-agent-platforms.html" }, { "name": "Agent-Native Platforms", "url": "https://auxfirst.com/agent-native-db.html" }, { "name": "AI Dictionary", "url": "https://auxfirst.com/ai-dictionary/" }, { "name": "Newsletter / Dispatches", "url": "https://auxfirst.com/newsletter.html" } ], "engagement": { "intake": "https://auxfirst.com/index.html#contact", "method": "discovery_call", "duration": "30_minutes", "outcome": "Engagement format recommendation (Seminar · Workshop · Audit · Sprint · Retainer) with scope and timeline.", "preReadRecommended": [ "https://auxfirst.com/manifesto.md", "https://auxfirst.com/services.md" ] } }