WorkBuddy AI Desktop Agent Guide

Workflow Playbook & Supported Model Pricing Reference

WorkBuddy AI operates directly inside your authorized macOS or Windows folders to plan multi-step tasks, process spreadsheets and PDFs, coordinate multi-agent Expert Groups, and ship finished decks and code. Explore practical recipes and model cost tiers below.

Best Model: GPT-5.4 / Claude Sonnet 4.6

1. Raw CSV Folder to Executive Slide Deck

Point WorkBuddy at a local data folder to clean metrics, surface anomalies, and generate a finished PowerPoint deck.

Target folder: ./quarterly-data/
Goal: Aggregate regional revenue retention and churn across all CSVs,
identify the top 3 growth bottlenecks, and generate a 10-slide executive
presentation at ./output/Q4-Business-Review.pptx plus a summary memo.
Best Model: Claude Sonnet 4.6 + DeepSeek-R1

2. Parallel PM + Architect + Engineer + QA Sprint

Launch a collaborative Expert Group where specialized agents divide specification, coding, and automated test verification.

Mode: Expert Group (PM + Architect + Engineer + QA)
Workspace: ./projects/subscription-service/
1) PM Agent: Draft edge-case requirements in ./specs/billing-prd.md
2) Architect: Define API schema and state transitions
3) Engineer & QA: Implement endpoints and run unit tests until 100% pass
Best Model: Gemini 2.5 Pro (1M Context)

3. Multi-PDF Contract & Policy Comparison Matrix

Leverage a 1M-token context window to extract structured clauses across dozens of local PDFs into an audited Excel matrix.

Workspace: ./legal-agreements/
Task: Inspect every PDF in the folder, extract Renewal Window, Liability
Cap, SLA Uptime Credits, and Governing Law, flag non-standard clauses,
and export a comparison workbook to ./audit/contract-matrix.xlsx.
Best Model: DeepSeek-V3.2 / Qwen3-Max

4. Reusable Domain Skill for Low-Cost Daily Briefs

Pair modular team Skills with high-efficiency models to run recurring competitive and operational reports at minimal credit cost.

Use Skill: @competitive-landscape-brief
Input context: ./strategy/product-roadmap-2026.md
Deliverable: Benchmark top 5 category competitors across pricing tiers,
agent autonomy, and local tool integrations; save to ./briefs/q4-landscape.md.
Best Model: Auto Router / Gemini 2.5 Flash

5. Trigger Desktop Tasks from Mobile Messaging

Use WorkBuddy Remote Assistant mode to dispatch local workstation tasks from Slack, Telegram, or Discord while away from your desk.

[Remote Assistant Task]
Analyze the latest benchmark logs in ~/Workspaces/perf-suite/results/,
generate a latency comparison chart PDF, and reply in this thread with
the top three regression root causes.
Best Model: Claude Opus 4.6 / Sonnet 4.6

6. Autonomous Repository Audit & Patch Verification

Let WorkBuddy inspect a local repository, eliminate N+1 queries, and verify fixes against your local test suite.

Workspace: ./services/checkout-api/
Goal: Audit database access layers for N+1 queries and unhandled timeouts,
apply minimal patches, execute the full pytest suite, and write a before/after
verification report to ./reports/query-audit.md.
Plans & Model Pricing

WorkBuddy AI Subscription Plans & Supported Model Rates

Run tasks using built-in TokenHub credits or connect your own API keys (BYOK) and local models via Settings → Models.

Free Plan
$0 / month

For individual evaluation and light desktop automation

  • 100 base monthly credits + daily active bonus (~30 credits/day)
  • Automatic model routing across built-in models
  • Local file workspace access & core Skills
Team Plan
$40 / seat / month

For collaborative teams sharing Skills and pooled credits

  • 1,000–2,000 credits per seat pooled across the organization
  • Shared custom Skills repository & unified billing
  • Admin Console & centralized model routing policies
Estimated Monthly Usage 1,200 Credits / mo
Recommended Tier Pro ($10/mo)
Model Provider Context Window Input / 1M Tokens Output / 1M Tokens Credit Rate Recommended WorkBuddy Use Case
Claude Sonnet 4.6 Anthropic 200K $3.00 $15.00 1.0x Standard Multi-step coding, Expert Group orchestration, complex tool chains
Claude Opus 4.6 Anthropic 200K $15.00 $75.00 2.5x Frontier System architecture design, deep codebase security & logic audits
GPT-5.4 / GPT-4.1 OpenAI 128K – 1M $2.50 $10.00 1.0x Standard Executive slide decks (.pptx), spreadsheet modeling, structured reports
Gemini 2.5 Pro Google 1M $1.25 $10.00 0.8x Long-Ctx Batch multi-PDF contract extraction, full-repo context analysis
Gemini 2.5 Flash Google 1M $0.15 $0.60 0.2x Fast High-speed file sorting, instant Remote Assistant summaries
DeepSeek-V3.2 DeepSeek 128K $0.27 $1.10 0.2x Economy Cost-optimal daily CSV cleaning, markdown documentation batching
DeepSeek-R1 DeepSeek 128K $0.55 $2.19 0.4x Reasoning Financial reconciliation, mathematical verification, root-cause analysis
Qwen3-Max / Coder-Plus Alibaba 256K $0.40 $1.20 0.25x Economy Multilingual spreadsheet automation, structured JSON/script tasks
Hunyuan / Kimi K2 / MiniMax Built-in / BYOK 256K $0.30 – $0.80 $1.00 – $2.40 0.2x – 0.5x Long-document synthesis, cross-model research benchmarking
Local Ollama / Custom API BYOK (Settings → Models) Model Dependent $0 (Direct API) $0 (Direct API) 0 Credits Zero platform credit burn via custom OpenAI-compatible or local endpoints

Ready to automate multi-step work on your desktop?

Download WorkBuddy AI for macOS & Windows or view full plan details on the official website.

Go to www.workbuddy.ai ↗