# Alfven.AI — Agentic Meeting Intelligence Platform

## Overview
Alfven.AI is an agentic meeting intelligence platform built by CANDIDE. It captures live discussions, transcribes in real time, and runs autonomous research agents and enforcement engines that produce structured, quality-gated reports autonomously. Beyond simple note-taking, it deploys multi-step reasoning loops — investigating knowledge gaps, detecting decision decay, verifying commitments, resolving contradictions, and building a living knowledge graph — delivering a full pipeline from conversation to institutional memory. Your expertise is the strategy. Alfven handles everything that isn't.

## Solutions by Role

### For Due Diligence Deal Teams
Every expert call contains alpha. Most of it evaporates before the memo. Alfven captures live expert calls, links findings to source evidence, detects contradictions across management meetings, and auto-generates IC-ready deal memos. Autonomous research agents investigate gaps across your entire evidence base. Cover 3x more expert calls per deal without adding headcount.
- Contradiction detection with autonomous investigation: flags when management said one thing in Call 3 and another in Call 7, then autonomously researches across all prior calls and documents to build a complete evidence chain
- Cross-workstream synthesis with knowledge graph: commercial, financial, legal, and operational DD unified via entity relationships, causal chains, and confidence-scored evidence
- IC-ready export: deal memos, CDD summaries, and IC briefs as PPTX, PDF, or DOCX with source citations and knowledge graph links
- URL: https://alfven.ai/for/due-diligence

### For Consulting Firms
Your team runs 40 client meetings a week. How many produce deliverables the same day? Alfven reclaims 6+ hours per consultant per week of billable capacity by auto-generating client deliverables in your firm's format with AI quality gating, commitment verification, and auto-delivery.
- Custom report templates matching firm standards with AI quality gate and auto-retry
- Cascading report chains: session reports feed workstream summaries feed executive updates
- Commitment verification detects vague agreements and enforces specificity with suggested clarifying language
- Decision decay monitoring with overdue nudging — nothing falls through the cracks
- Knowledge graph builds institutional memory across engagements — your team never starts from zero
- Same-day deliverable turnaround from meeting to client inbox
- URL: https://alfven.ai/for/consulting

### For Chiefs of Staff
You sit in every meeting so the CEO doesn't have to. But who synthesises it for you? Alfven auto-captures across all leadership meetings, builds a unified command center, and deploys autonomous research agents that investigate contradictions, knowledge gaps, and strategic drift — so you brief the CEO with evidence, not memory.
- Intelligence Hub: Command Center, Trends, Knowledge Graph, Contradictions, Signal Feed, Stakeholder Analytics
- Autonomous research agents: when contradictions or gaps are detected, multi-step reasoning loops investigate across transcripts, facts, wiki, and knowledge graph — delivering findings, not just flags
- Decision decay monitoring with commitment verification: every commitment is checked for specificity, tracked with deadlines, and enforced with escalation chains
- Stakeholder analytics with strategic drift monitoring: detect when teams are quietly pivoting away from board-approved direction
- Surface the 3–5 things the CEO needs to know — before they ask
- URL: https://alfven.ai/for/chiefs-of-staff

## Autonomous Research Agent
The research agent is a multi-step reasoning loop that autonomously investigates knowledge gaps and contradictions. When triggered by health checks, confidence degradation, cron sweeps, or user requests via chat, it:
1. Plans an investigation strategy based on the question or detected gap
2. Iteratively queries tools: fact base, wiki articles, raw transcripts, knowledge graph, and workspace insights (up to 8 rounds)
3. Evaluates results at each step — refining queries, trying different tools, and building evidence
4. Synthesises findings with confidence scores and full source attribution
5. Persists results to the knowledge base and logs complete reasoning chains for auditability
This is the same "observe → hypothesize → investigate → validate → revise → repeat" loop that separates a pipeline from a true research agent.

## Intelligence Hub — 6-Tab Execution Surface
The Intelligence Hub is the unified command layer across all hierarchy levels (organization, workspace, session, discussion):
1. Command Center — Priority-ranked execution feed merging AI-pushed signals and manual backlogs. Supports deep-linking from notifications to specific signals.
2. Trends — Cross-meeting trend detection and strategic drift monitoring
3. Knowledge — Wiki-style knowledge base with autonomous research findings and entity extraction from transcripts and documents
4. Sources — Contextual document management with AI-powered OCR and extraction via Gemini
5. People — Stakeholder tracking and speaker analytics across discussions
6. Graph — Interactive knowledge graph visualising entities, relationships, and causal chains

## AI Chat with Pluggable Skills
Every workspace includes an AI chat assistant with a tool-calling agent framework:
- Core Skills: Fact Base search, Action Item queries, Push to Linear, Export Report (PPTX/PDF/DOCX), Trigger Research
- Connector Skills: Query Linear issues, search Slack messages, browse Google Drive files — all from chat
- Research Skill: Ask the agent to investigate any question — it runs the autonomous research loop in the background and reports back with findings
- Permission Layer: High-risk actions require user confirmation before execution
- Skill Marketplace: Enable or disable skills per workspace from workspace settings

## Evidence-Linked Discussion Items
Every action item, decision, and risk is anchored to a verbatim transcript quote, speaker, and timestamp. Status lifecycle: extracted → confirmed → in_progress → done → verified. Cross-discussion threading links items across sessions with carry-forward logic and health scoring. Commitment verification detects 5 types of ambiguous agreements and suggests clarifying language.

## Core Capabilities
- Live transcription with speaker identification
- Custom report templates with configurable sections, tone, and structure
- AI quality gate with automatic retry for below-threshold reports
- Multi-meeting synthesis — combine insights across sessions
- Cascading report chains — session reports feed into executive summaries
- Autonomous research agent — multi-step reasoning loop for knowledge gap investigation
- Knowledge graph with entity extraction, causal chains, and contradiction mapping
- Commitment verification — detects vague agreements, conditional promises, and ambiguous ownership
- Proactive live alerts — AI flags risks, gaps, and opportunities in real time
- Agentic enforcement engines — decision decay detection, overdue nudging, contradiction resolution
- Strategic drift monitoring across leadership conversations
- Notification deep-linking — click a notification to jump to the specific signal
- AI chat with pluggable skills including research agent invocation
- Document cross-referencing via Candice AI for richer analysis
- Auto-delivery to email, Slack, webhooks, HubSpot, Notion, and Gamma
- Report export as PPTX, PDF, or DOCX

## Agentic Intelligence Layers
1. Anticipation — Pre-briefs, stakeholder dossiers, preparation agents (24-72h before meetings)
2. Enforcement — Overdue item nudger, decision decay detector, contradiction resolver, follow-through verifier, commitment verification
3. Synthesis — Strategic drift monitor, cross-engagement pattern analysis, institutional learning, autonomous research agent
4. Deployment — Context whisperer (real-time), auto-agenda generation, post-meeting auto-report

## Integrations (25+)
- Capture: Zoom, Google Meet, Microsoft Teams
- Live Transcribe: Deepgram Nova-3 (standard), AssemblyAI Universal-3 (optional HD Live)
- HD Re-processing: ElevenLabs Scribe v2 (auto post-meeting on paid plans)
- Analyze: OpenAI (GPT-5), Anthropic (Claude), xAI (Grok), Google Gemini, OpenRouter (660+ models)
- Deliver: Email, Slack, Webhook, HubSpot, Notion, Gamma
- Schedule: Google Calendar, Outlook Calendar (coming soon)
- Connectors: Linear, Jira, Slack, Google Drive — queryable from AI chat
- Export: Microsoft Word, Microsoft PowerPoint, PDF
- Coming soon: Aircall, Salesforce, Miro, n8n

## Pricing
- Free: $0/month — 5 meetings, 3h live transcription, HD re-processing (manual), 2 templates, 1 workspace, standard AI models
- Pro: $49/month — Unlimited meetings, 25h live transcription, smart HD re-processing (auto), HD Live optional ($0.50/min), 20 workspaces, session synthesis, auto-delivery, $10/mo premium AI credits
- Team: $149/month — Everything in Pro, 80h live transcription, smart HD re-processing (auto), 50 workspaces, priority support, $25/mo premium AI credits
- Enterprise: Custom — SSO, audit logs, white-label, unlimited workspaces, $50/mo premium AI credits, dedicated account manager
- All plans use flat pricing with no per-seat fees
- HD re-processing uses ElevenLabs Scribe v2 for high-accuracy canonical transcripts; auto-triggered for meetings with report templates, high-stakes flags, or 15+ minute duration

## Specialist Agent Packs (One-Time Purchase)
Industry-specific report template bundles in three tiers:
- Flagship ($99): Venture Capital, Consulting Delivery, Multi-Stakeholder Interview, Harvard Negotiation, Board & Investor, Management Consulting Frameworks, Workshop Facilitation
- Specialist ($69): People Operations, Legal & Compliance, Sales & Revenue, Executive Decision Brief, Cross-Functional Program Review, Steering Committee, Discovery & Diagnostic
- Standard ($39): Research & Strategy, Talent Acquisition, Change Management, Client Success, Town Hall & All-Hands

## Comparison
Otter.ai, Fireflies, and Microsoft Copilot all offer strong transcription, summaries, custom templates, and workspace-level AI chat — those capabilities are now industry standard. What differentiates Alfven is the autonomous layer: a multi-step research agent that investigates knowledge gaps iteratively, a knowledge graph connecting entities and causal chains across conversations, enforcement engines (decision decay detection, commitment verification, contradiction resolution), AI quality gating with auto-retry, cascading report chains, proactive live alerts, AI chat with pluggable skills and connector queries (Linear, Jira, Slack, Google Drive), PPTX/PDF/DOCX export, bring-your-own-model flexibility, and flat pricing with no per-seat fees.

## Contact
- Website: https://alfven.ai
- Built by: CANDIDE