MultAI

One skill. Seven AI platforms. Instant synthesis.

/multai is a Claude Cowork/Code plugin skill that submits your research prompt to Claude.ai, ChatGPT, Microsoft Copilot, Perplexity, Grok, DeepSeek, and Google Gemini simultaneously — then synthesizes the results into structured deliverables. Market landscape reports, capability comparison matrices, product deep-dives, or a direct answer from all seven platforms at once.

---

How It Works

You → /multai → 7 AI Platforms in parallel → Synthesized report

You type one prompt. /multai figures out what you need, runs it across all platforms, and hands back a consolidated result. No flags, no routing decisions, no platform management.

Capability	Detail
Parallel submission	All 7 platforms run concurrently
Intelligent routing	Analyzes your intent and selects the right workflow automatically
Market landscape reports	9-section structured reports — top 20 commercial + OSS, positioning matrices, trends
Product deep-dives	Capabilities, integrations, pricing, competitive context, XLSX scoring
XLSX comparison matrix	Capability matrix auto-scored and reordered across platforms
DEEP mode	Activates Deep Research on each platform where available
Rate limiting	Per-platform budget tracking across sessions; never silently skips a platform
Agent fallback	Vision-based fallback via browser-use when a UI selector fails
Login retry	Real-time sign-in notification; 90-second countdown + automatic retry for platforms that need login
Popup dismissal	Auto-accepts browser dialogs; dismisses cookie banners, GDPR notices, and modal overlays
Chat readiness	Detects unexpected UI states (error pages, redirects) and hands control to browser-use for recovery
Verified install	Playwright import + headless Chromium launch verified on setup; cached for fast subsequent runs
Tab reuse	Existing browser tabs reused across runs; --followup continues open conversations
Report viewer	Ālo Design System report viewer with light/dark toggle, gradient accents, and interactive charts

---

Supported Platforms

Platform	Notes
Claude.ai	Pro plan recommended for DEEP mode
ChatGPT	Plus plan for Deep Research
Microsoft Copilot	Free tier works
Perplexity	Pro for Deep Research
Grok	X/Twitter account required
DeepSeek	Free tier works
Google Gemini	Google account required

---

Quick Start

1 — Prerequisites

• Claude Code v1.0.33 or later — check with claude --version, update with brew upgrade claude-code or npm update -g @anthropic-ai/claude-code
• Python 3.11+, Google Chrome

2 — Install

# Register the marketplace (one-time):
/plugin marketplace add alo-exp/multai

# Install:
/plugin install multai@multai

Run /reload-plugins if /multai doesn't appear immediately.

Python dependencies (playwright, openpyxl, Chromium) are installed and verified automatically on first session start via a SessionStart hook. The engine confirms Playwright imports correctly and Chromium launches headlessly — no manual setup required.

Agent fallback (optional): For the vision-based browser-use fallback:

bash "$(find ~/.claude/plugins/cache -name setup.sh | head -1)" --with-fallback

Alternative — Local / Dev Install

git clone https://github.com/alo-exp/multai.git
cd multai
bash setup.sh            # creates .venv, installs deps + Playwright Chromium
# optional agent fallback:
bash setup.sh --with-fallback

claude --plugin-dir ./multai

3 — Log in to platforms

Open Chrome and sign in to each platform. The engine reuses your existing Chrome profile — no credentials are stored.

4 — Set optional API keys

# ~/.zshrc or ~/.bashrc
export GOOGLE_API_KEY="..."      # free from aistudio.google.com — enables Gemini agent fallback
export ANTHROPIC_API_KEY="..."   # from console.anthropic.com — enables Claude agent fallback

5 — Use the skills

/multai — research, landscape analysis, direct multi-AI queries, and matrix operations:

/multai Run a market landscape analysis on DevOps platforms for SMBs
/multai Research humanitec.com
/multai Add Harness to the comparison matrix
/multai What are the main trade-offs between Rust and Go for backend services?

/comparator — standalone head-to-head comparisons without a prior research run:

/comparator Compare Humanitec vs Port.io
/comparator Which is better for a startup — Backstage or Cortex?
/comparator Compare these two products and give me a weighted score

/consolidator — merge any set of content sources into a unified, structured report:

/consolidator Consolidate these three research papers into a summary report
/consolidator Summarize these five customer interview transcripts into themes
/consolidator Combine these meeting notes from four teams into a single overview

All skills announce their plan before acting — you can always override or adjust.

---

What /multai Can Do

Market landscape reports

"Run a landscape analysis on API gateway platforms" "Give me a market map for observability tools for startups"

Produces a 9-section structured Market Landscape Report: market definition, size & CAGR, competitive positioning (2×2, Wave-style, Value Curve), key trends, top 20 commercial + OSS solutions, buying guidance, and future outlook.

Output: reports/{task-name}/{Category} - Market Landscape Report.md + auto-launched browser preview

Product deep-dives

"Research humanitec.com" "Evaluate Backstage" "Analyze Port.io — how does it compare to Cortex?"

Deep research on a specific product — capabilities, integrations, pricing, competitive context — optionally scored in the comparison matrix.

Output: reports/{task-name}/{Product} - Consolidated Intelligence Report.md

Head-to-head comparisons — /comparator

"Compare Humanitec vs Port.io" "Which is better for SMBs — Backstage or Cortex?" "Compare these two products and score them"

Standalone skill for comparing any two (or more) solutions. Derives a capability framework from available evidence (CIRs, documents, or LLM knowledge), optionally lets you set feature priorities, scores each solution with priority-weighted ticks, and produces both an XLSX matrix and a readable Markdown summary with per-category winners and key differentiators. No prior research run required — works from LLM knowledge alone if needed.

Output: reports/{domain}/{domain}-matrix.xlsx + reports/{domain}/{task-name}-comparison-summary.md

Can also be triggered via /multai — it routes automatically when comparison intent is detected.

Multi-source consolidation — /consolidator

"Consolidate these three research papers into a summary" "Summarize these five customer interviews into themes and recommendations" "Combine these meeting notes from four teams into one overview"

Standalone skill for synthesizing content from any set of sources — documents, transcripts, notes, URLs, pasted text, or AI platform responses — into a unified, well-structured report. Detects the content type and auto-derives an appropriate report structure (research synthesis, theme extraction, decision log, etc.), or follows a consolidation guide you provide.

When invoked from within a /multai workflow, operates in AI-Responses mode and produces a CIR (Consolidated Intelligence Report) from raw platform outputs.

Output: [Topic] - Consolidated Report.md (generic) or [Topic] - Consolidated Intelligence Report.md (AI-Responses mode)

Comparison matrix operations

"Add Harness to the comparison matrix" "Update the score for Cortex on the developer portal capability" "Reorder the matrix by score"

Maintains an existing XLSX capability matrix — adding platforms, updating scores, applying combo columns, reordering, and verifying coverage.

Direct multi-AI queries

"What are the emerging consensus patterns for LLM memory management?" "Summarize the current state of WebAssembly for server-side workloads"

For anything that isn't a landscape, deep-dive, or matrix operation, /multai submits directly to all 7 platforms and synthesizes a consolidated answer.

---

Project Structure

multai/
├── .claude-plugin/
│   ├── plugin.json           ← Plugin manifest
│   └── hooks.json            ← SessionStart hook (auto-installs deps)
├── skills/
│   ├── orchestrator/         ← /multai skill — router + engine owner
│   │   ├── SKILL.md
│   │   ├── platform-setup.md
│   │   └── engine/           ← Playwright automation engine
│   │       ├── orchestrator.py
│   │       ├── config.py
│   │       ├── rate_limiter.py
│   │       ├── agent_fallback.py
│   │       ├── collate_responses.py
│   │       └── platforms/    ← claude_ai.py chatgpt.py copilot.py …
│   ├── consolidator/         ← /consolidator skill — multi-source synthesis + CIR
│   ├── landscape-researcher/ ← Market landscape workflow (internal)
│   ├── solution-researcher/  ← Product deep-dive workflow (internal)
│   └── comparator/           ← /comparator skill — head-to-head comparisons + XLSX matrix
├── domains/                  ← Shared domain knowledge (enriched per run)
├── reports/
│   └── preview.html          ← Report viewer
├── docs/                     ← Architecture, SRS, test & CI/CD plans
├── tests/                    ← pytest suite
├── setup.sh                  ← Bootstrap — venv, deps, Playwright Chromium
├── pyproject.toml
├── requirements.txt
├── USER-GUIDE.md
└── CONTRIBUTOR-GUIDE.md

---

Rate Limiting

The engine tracks per-platform usage across sessions and warns when a budget is low, but never skips a platform based on budget alone. A platform is excluded from a round only if:

• A sign-in page is detected (needs_login — 🔑)
• The platform is unreachable (network error)
• Actual quota exhaustion is detected on-page

---

Resilience

Agent Fallback

When a Playwright selector fails, a browser-use vision agent takes over automatically:

1. ANTHROPIC_API_KEY set → Claude Sonnet is the agent LLM
2. GOOGLE_API_KEY set → Gemini 2.0 Flash (free tier at aistudio.google.com)
3. Neither key → fallback disabled; Playwright exception propagates

If all Playwright steps fail for a platform, a full agent-driven run is attempted as a last resort.

Login Handling

When a platform requires sign-in, the engine notifies you immediately (not after all platforms finish) and retries automatically after a 90-second countdown. No manual re-runs needed.

Popup & Dialog Handling

Browser alert()/confirm() dialogs are auto-accepted. CSS overlays (cookie banners, GDPR notices, sign-up modals) are dismissed automatically via scoped selectors targeting modal and consent containers. The engine handles up to 3 layered popups per lifecycle step.

Chat Readiness Check

Before interacting with any platform, the engine verifies the chat UI is in the expected state — checking for sign-in redirects, HTTP error pages (404, 500, 502, 503), and blank tabs. If the UI is unexpected and browser-use is available, the agent takes over to navigate back to the chat interface.

---

Documentation

Document	Description
USER-GUIDE.md	Installation, usage, viewing reports
CONTRIBUTOR-GUIDE.md	CLI flags, platform internals, tests, CI/CD
docs/Architecture-and-Design.md	System topology and design decisions
docs/SRS.md	Software Requirements Specification
CHANGELOG.md	Version history

---

Requirements

Requirement	Version
Python	≥ 3.11
Google Chrome	latest
Claude Code	≥ v1.0.33

---

License

MIT — see LICENSE.

---

alo-exp · User Guide · Contributor Guide