Vitto Groups — Multi-Vitto Open Platform (Future Vision)

Status: Future spec — NOT for immediate implementation Depends on: Whistle (intra-space group chat) must ship first Session: S218 | Date: 2026-03-14

The Vision

Anyone can create a group with anyone — not limited to a single space. Each user brings their own Vitto into the conversation. Multiple Vittos coexist, each with their own personality, memories, expertise, and Dossier knowledge. Vitto becomes a helper within any group or conversation.

This is the startup play. Vitto stops being a product feature and becomes a platform.

What Makes This Different

Each Vitto is unique

• Franco's Vitto knows the codebase, deploy history, architecture decisions
• A client's Vitto knows their business requirements, timeline, budget
• A designer's Vitto knows the design system, accessibility guidelines, brand rules
• In a group, these Vittos complement each other with different knowledge domains

Vitto as representative

• When Franco is AFK, his Vitto can answer on his behalf: "Franco is busy, but based on his past decisions, he'd probably say..."
• Each Vitto has a confidence level about representing its user
• Vitto can defer: "I'm not sure Franco would agree — I'll flag this for when he's back"

Multi-Vitto interaction

• Vittos can debate: "Franco's Vitto thinks we should use PostgreSQL, but Alex's Vitto notes the team has more MongoDB experience"
• Vittos can negotiate: one Vitto advocates for its user's priorities, the other for theirs
• A "mediator Vitto" (platform-provided, neutral) can facilitate when user Vittos disagree

Use cases

• Client-vendor collaboration: Client's Vitto + developer's Vitto in one thread. Both have full context from their respective sides.
• Cross-company projects: Each company's team members bring their own Vittos with institutional knowledge.
• Community groups: Interest-based groups where everyone's Vitto contributes relevant expertise.
• Personal assistant groups: Family, friends, teams — each person's Vitto helps with scheduling, remembering, summarizing.

Architecture Considerations (not designed yet)

Identity

• Cross-tenant user identity (OAuth, email-based, or GoGrab account)
• Vitto identity tied to user account, not space
• Privacy controls: what does my Vitto share in external groups vs what stays private?

Multi-agent orchestration

• Turn-taking protocol: which Vitto responds? Based on relevance + recency + user preference
• Conflict resolution: when Vittos disagree, how is it surfaced?
• Token budget: N Vittos × context = expensive. Need tiered participation (only relevant Vittos active per message)

Data model

• Same ChatSession schema with groupType: 'open', spaceId: null
• Each participant has a vittoId linking to their personal Vitto configuration
• GroupConversation in vitto-api spawns per-user Vitto contexts

Monetization ideas

• Free: groups with up to 3 people + their Vittos
• Pro: unlimited group size, Vitto-as-representative (AFK mode), recording + transcription
• Enterprise: custom Vitto personalities, compliance controls, audit trails
• Platform: API for third-party apps to invite Vitto into their group chats (Slack, Discord, Teams)

Capabilities Vitto can offer in groups

• Recording + transcription: Vitto records the conversation, generates minutes
• Humor + personality: Vitto makes contextual jokes, lightens the mood (configurable)
• Action item tracking: Vitto tracks who committed to what, follows up
• Knowledge bridging: "Alex mentioned X last week in a different conversation — relevant here"
• Translation: Real-time translation between participants speaking different languages
• Summarization: "Here's what happened while you were away" — personalized per participant

What the Whistle Feature Builds Toward This

The intra-space Whistle feature creates the foundation:

• ChatSession.participants[] — works for both intra-space and open groups
• groupType field — 'intra-space' now, 'open' later, same schema
• Socket.IO multi-participant broadcast — same infrastructure
• Cross-channel delivery — already works via S217 bridge
• Dossier extraction from group context — same pipeline
• Vitto adaptive behavior (active/observer/facilitator) — evolves into multi-Vitto orchestration

The jump from Whistle to Vitto Groups is:

1. Remove the space boundary (spaceId: null)
2. Add per-participant Vitto context (each user's Dossier + personality)
3. Add multi-Vitto turn-taking protocol
4. Add cross-tenant identity resolution
5. Build the standalone product surface (web app, mobile app)

Next Steps (when ready)

1. Ship Whistle (intra-space) and validate the group chat UX
2. Gather feedback: do users actually whistle? How big are groups? What does Vitto do well/badly?
3. Design the multi-Vitto interaction protocol (this is the hard research problem)
4. Build cross-tenant identity (OAuth + email-based invites)
5. Launch Vitto Groups as a standalone product with free tier