OpenClaw Productivity Use Cases: 15 Personal and Team Automation Configurations

Most developers have a list of productivity automations they intend to build. The list grows. The automations don't.

The friction is rarely the idea. It's the setup: choosing the right model tier for the task, writing a system prompt that produces reliable output, and wiring a trigger that runs without daily supervision. Multiply that by 15 cases and the category feels overwhelming before it starts.

This page removes that friction. All 15 OpenClaw Productivity use cases are organized into 4 sub-themes, each summarized with the recommended Claude tier, system prompt structure, and configuration steps. The selection guide in Section 2 narrows the field to one starting case based on your current highest-priority pain point.

The 4 sub-themes and case counts:

• Personal Assistant (3 cases) — morning briefs, multi-channel assistants, phone-based assistant
• Project & Task Management (5 cases) — autonomous planning, state tracking, dashboards, Todoist, multi-agent teams
• Communication & CRM (4 cases) — customer service, inbox management, personal CRM, event confirmations
• Health & Household (3 cases) — health tracking, family calendar, second brain

One cost point worth stating upfront: 9 of the 15 cases run correctly on Haiku or Sonnet. Running all 15 simultaneously using correct tier routing costs a fraction of the equivalent Opus cost for every case. Confirm the arithmetic from https://wisgate.ai/models and calculate the annual delta before making routing decisions.

Most of these automations are running within one session. Open AI Studio and test your system prompt against sample inputs before committing to a cron schedule — it takes less time than writing the prompt. Get your API key at wisgate.ai/hall/tokens, trial credits included. By the end of this article you'll have a specific case selected, configured, and validated.

OpenClaw Configuration

Step 1 — Locate and Open the Configuration File

OpenClaw stores its configuration in a JSON file in your home directory. Open your terminal and edit the file at:

Using nano:

nano ~/.clawdbot/clawdbot.json

Step 2 — Add the WisGate Provider to Your Models Section

Copy and paste the following configuration into the models section of your clawdbot.json. This defines WisGate as a custom provider and registers Claude Opus with your preferred model settings.

"models": {
  "mode": "merge",
  "providers": {
    "moonshot": {
      "baseUrl": "https://api.wisgate.ai/v1",
      "apiKey": "YOUR-WISGATE-API-KEY",
      "api": "openai-completions",
      "models": [
        {
          "id": "claude-opus-4-6",
          "name": "Claude Opus 4.6",
          "reasoning": false,
          "input": ["text"],
          "cost": {
            "input": 0,
            "output": 0,
            "cacheRead": 0,
            "cacheWrite": 0
          },
          "contextWindow": 256000,
          "maxTokens": 8192
        }
      ]
    }
  }
}

Note: Replace YOUR-WISGATE-API-KEY with your key from wisgate.ai/hall/tokens. The "mode": "merge" setting adds WisGate's models alongside your existing providers without replacing them. To add additional models, duplicate the model entry block and update the "id" and "name" fields with the correct model IDs from wisgate.ai/models.

Step 3 — Save, Exit, and Restart OpenClaw

If using nano:

1. Press Ctrl + O to write the file → press Enter to confirm
2. Press Ctrl + X to exit the editor

Restart OpenClaw:

1. Press Ctrl + C to stop the current session
2. Relaunch with:

openclaw tui

Once restarted, the WisGate provider and your configured Claude models will appear in the model selector.

OpenClaw API Productivity: Which of the 15 Cases Should You Configure First?

Fifteen cases across four sub-themes is not a linear reading list — it's a reference. This section gives you one starting case based on the most common developer pain points.

Selection by current highest pain point:

Selection by available setup time:

• Under 30 minutes: Inbox De-clutter, Custom Morning Brief, Todoist Task Manager, Health & Symptom Tracker, Family Calendar & Household Assistant
• 1–2 hours: Personal CRM, Project State Management, Multi-Channel AI Customer Service, Dynamic Dashboard, Event Guest Confirmation, Second Brain, Multi-Channel Personal Assistant
• 2+ hours: Autonomous Project Management, Multi-Agent Specialized Team, Phone-Based Personal Assistant

The model tier principle for this category:

Haiku is the correct tier when the instruction is unambiguous and the output schema is fixed. Sonnet is correct when synthesis across multiple inputs or nuanced judgment is required. Opus applies to exactly one Productivity case — Autonomous Project Management — where planning quality determines the quality of every downstream execution step. Never route a task to a higher tier when output quality is indistinguishable from the tier below. Confirm pricing from https://wisgate.ai/models and calculate the cost arithmetic before making routing decisions.

AI Personal Assistant API: 3 OpenClaw Configurations for Daily Context and Reach

This sub-theme covers automations that act as a persistent daily assistant — delivering structured morning context, managing multi-channel communication, and handling voice-initiated queries. Two of three cases run on Sonnet; the Morning Brief runs on Haiku.

Case 1 — Custom Morning Brief

Model: claude-haiku-4-5-20251001

What it does: Aggregates calendar events, open tasks, a news digest, and weather into a single structured brief delivered before the workday begins. Output is fixed-format: today's priorities, schedule, and a one-paragraph context summary. Nothing is synthesized across ambiguous inputs — it formats and summarizes known sources.

Why Haiku: The instruction is unambiguous and the output schema is fixed. This is not a reasoning task — it is a formatting and summarization task. Haiku handles it correctly at the lowest per-request cost in the Claude family.

System prompt structure:

• Role: personal briefing assistant for a software developer
• Input sources: listed by name (calendar, task list, news feed, weather API output)
• Output schema: three sections with defined word limits — Priorities (max 5 bullets), Schedule (time-ordered list), Context (max 80 words)
• Delivery time: configuration variable

Configuration steps in OpenClaw:

1. Complete universal setup with claude-haiku-4-5-20251001
2. Paste the system prompt with your input sources and output schema defined
3. Connect inputs via a pre-processing step that formats calendar, task, and news data before passing to OpenClaw
4. Set cron trigger at your target delivery time (recommended: 07:00 daily)
5. Validate in AI Studio with sample aggregated input before activating

[Link to: Full Custom Morning Brief configuration →]

Case 2 — Multi-Channel Personal Assistant

Model: claude-sonnet-4-5

What it does: Monitors and responds across email, Slack, and SMS using a unified context window. Maintains awareness of ongoing threads across channels so responses are consistent — a follow-up email acknowledges context from the earlier Slack conversation.

Why Sonnet over Haiku: Multi-channel context synthesis is not a fixed-schema task. The model must track open threads across concurrent channels and maintain response consistency. Haiku degrades on cross-channel coherence at this input type.

System prompt structure: channel list with priority weighting; response style per channel (concise for SMS, formal for email); escalation rules for urgent items; open thread retention pattern.

Configuration steps in OpenClaw:

1. Complete universal setup with claude-sonnet-4-5
2. Define channel list, priority weights, and tone rules per channel in the system prompt
3. Test with a two-channel synthetic scenario in AI Studio — a question asked on Slack and a follow-up email about the same topic
4. Activate per channel, one at a time, verifying each before adding the next

[Link to: Full Multi-Channel Personal Assistant configuration →]

Case 3 — Phone-Based Personal Assistant

Model: claude-sonnet-4-5

What it does: Extends OpenClaw to voice-initiated queries via phone or SMS. The developer calls or texts; the assistant interprets the request, executes via the WisGate API, and returns a voice or text response. Useful for calendar queries, task capture, and quick information lookups while away from a screen.

System prompt structure: voice-accessible assistant role; response length constraints (voice responses max 2 sentences); supported command categories (calendar, tasks, reminders, search); fallback behavior for out-of-scope queries.

Configuration steps in OpenClaw:

1. Complete universal setup with claude-sonnet-4-5
2. Configure a telephony integration layer (such as Twilio) to pass transcribed voice input to OpenClaw as a user message
3. Define supported command categories and fallback response in the system prompt
4. Test with five representative voice query types in AI Studio — scheduling a meeting, capturing a task, asking for today's priorities — before activating the phone endpoint

[Link to: Full Phone-Based Personal Assistant configuration →]

Claude API Task Automation: 5 OpenClaw Configurations for Project and Task Management

This sub-theme contains the highest-complexity cases in the Productivity category — including the only Opus-tier case in all 15. It also contains three Sonnet cases and one Haiku case. Read the model rationale for each before selecting.

Case 4 — Autonomous Project Management

Model: claude-opus-4-5

What it does: Accepts a high-level project goal and autonomously plans, schedules, delegates, and tracks execution — generating subtask lists, monitoring progress, and replanning when a task is blocked. This is the only case in the Productivity category where Opus is the correct and operationally justified tier.

Why Opus: Planning quality determines execution quality for every downstream step. A wrong task decomposition from a lower-tier model produces ambiguous subtasks that block or misdirect every executor that follows. This is the one case where the Opus cost premium has direct operational justification.

System prompt structure:

• Project manager role with authority to create, assign, and close tasks
• Input schema: project goal, constraints, team/agent roster, deadline
• Output schema: task array with fields for id, description, assignee, dependency, estimated time, and status
• Replan trigger: when any task reaches status: blocked, output a revised plan for the blocked and dependent tasks

Configuration steps in OpenClaw:

1. Complete universal setup with claude-opus-4-5
2. Define the project scope, team roster, and output schema in the system prompt
3. Input your project goal; review the generated task plan before activating execution
4. Validate plan quality in AI Studio with a sample 5-task project before connecting to your task management system

[Link to: Full Autonomous Project Management configuration →]

Case 5 — Project State Management

Model: claude-sonnet-4-5

What it does: Maintains a persistent state file across multi-session projects. At session start, reads current state and provides a context summary. At session end, updates the state file with completed work, decisions made, and next steps.

System prompt structure: state reader/writer role; state file schema (current phase, completed items, open items, blockers, decisions, next action); distinct session-start and session-end prompt variants with different output instructions.

Configuration: Complete universal setup with claude-sonnet-4-5. Define your state file schema. Use the session-start prompt variant each morning and the session-end variant before closing. Store the state file in a consistent location that both prompt variants reference by path.

[Link to: Full Project State Management configuration →]

Case 6 — Dynamic Dashboard

Model: claude-sonnet-4-5

What it does: Generates a structured project or team dashboard on demand or on schedule, aggregating task status, blockers, metrics, and next actions into a consistently formatted report.

System prompt structure: dashboard generator role; named input data sources; output schema with defined sections (Active work, Blockers, Metrics, Next actions); refresh interval as a configuration variable.

Configuration: Complete universal setup with claude-sonnet-4-5. Define your data sources and output schema. Set the trigger as either on-demand (initiated by message in OpenClaw) or scheduled (cron job that passes current project data as user input).

[Link to: Full Dynamic Dashboard configuration →]

Case 7 — Todoist Task Manager

Model: claude-haiku-4-5-20251001

What it does: Interprets natural language task input and creates, updates, or closes Todoist tasks via the Todoist API. The developer types or says a thought; the agent formats it correctly and files it without friction.

Why Haiku: The task is parsing and formatting. The instruction is unambiguous — convert natural language to a structured Todoist task. This is a fixed-schema output task that does not benefit from Sonnet-level reasoning.

System prompt structure: Todoist integration role; task schema (title, project, due date, priority 1–4, labels); parsing rules for ambiguous due dates and priorities; confirmation format for created tasks.

Configuration: Complete universal setup with claude-haiku-4-5-20251001. Add your Todoist API token as a context variable in the system prompt. Test five representative natural language inputs in AI Studio — "remind me to follow up with the design team tomorrow morning" — before activating.

[Link to: Full Todoist Task Manager configuration →]

Case 8 — Multi-Agent Specialized Team

Model: claude-sonnet-4-5

What it does: Orchestrates a team of specialized agents — researcher, writer, reviewer, coordinator — each with a defined role, input format, and output schema. The coordinator agent receives the goal, delegates to specialists, and synthesizes outputs into a final deliverable.

System prompt structure: one system prompt per agent role; coordinator prompt includes the full role roster and delegation logic; each specialist prompt defines role, accepted input format, and required output schema.

Configuration: Complete universal setup with claude-sonnet-4-5. Create one dedicated OpenClaw conversation context per agent role. Run the coordinator first; pass its task assignments to each specialist conversation in sequence. The coordinator's synthesis call receives all specialist outputs as its final user message.

[Link to: Full Multi-Agent Specialized Team configuration →]

OpenClaw Use Cases — Communication & CRM: 4 Configurations for Inbox and Relationship Management

This sub-theme covers automations that manage inbound communication volume and relationship context. Two cases run on Haiku (structured, fixed-schema tasks); two run on Sonnet (nuanced tone matching and relationship synthesis).

Case 9 — Multi-Channel AI Customer Service

Model: claude-sonnet-4-5

What it does: Handles inbound customer queries across email, chat, and social channels. Classifies each query by type, generates a response draft with the correct channel tone, and flags complex cases for human review.

Why Sonnet: Customer-facing tone matching across channels — concise on chat, thorough on email, appropriately casual on social — requires nuanced judgment. Haiku produces acceptable classification output but degrades on tone-matched response quality for customer-facing messages.

System prompt structure: customer service agent role; supported channels with tone rules per channel; query classification taxonomy (billing, technical, general, escalate); response templates for common query types; escalation trigger conditions.

Configuration: Complete universal setup with claude-sonnet-4-5. Define your channel list, tone rules, and classification taxonomy. Test with five representative query types per channel in AI Studio before activating on live channels.

[Link to: Full Multi-Channel AI Customer Service configuration →]

Case 10 — Inbox De-clutter

Model: claude-haiku-4-5-20251001

What it does: Processes an inbox batch, classifies each email by required action (reply, archive, delegate, read-later, delete), and generates a prioritized action list. Runs on a schedule or on demand.

System prompt structure: inbox triage role; classification taxonomy with explicit definitions per category; priority scoring rules; output format as a classified list with one-line rationale per item.

Configuration: Complete universal setup with claude-haiku-4-5-20251001. Define your classification taxonomy with explicit edge cases (e.g., "newsletters go to read-later unless they contain a direct question"). Set a twice-daily trigger. Test with a 20-email sample batch in AI Studio before activating.

[Link to: Full Inbox De-clutter configuration →]

Case 11 — Personal CRM

Model: claude-sonnet-4-5

What it does: Maintains a structured contact database with interaction history, relationship context, and follow-up reminders. After significant interactions, the developer inputs notes; the agent updates the contact record and surfaces the next suggested action.

System prompt structure: CRM manager role; contact record schema (name, context, last interaction date, open follow-ups, relationship tier); update input format; next-action generation logic based on relationship tier and time since last contact.

Configuration: Complete universal setup with claude-sonnet-4-5. Define your contact record schema and relationship tiers. Initialize with existing contacts via batch input — paste formatted contact data as a single user message. Test update and retrieval flows in AI Studio before daily use.

[Link to: Full Personal CRM configuration →]

Case 12 — Event Guest Confirmation

Model: claude-haiku-4-5-20251001

What it does: Automates outbound guest confirmation messages, parses RSVP responses, updates the guest list, and generates a confirmation status report. Triggers follow-ups for non-responders after a defined wait period.

System prompt structure: event coordinator role; event details as configuration variables; confirmation message template with personalization fields; RSVP parsing rules (yes/no/maybe/no-response); follow-up trigger after N days of non-response.

Configuration: Complete universal setup with claude-haiku-4-5-20251001. Populate event details and guest list in the system prompt. Test confirmation message generation and RSVP parsing with synthetic responses in AI Studio. Activate the follow-up schedule before the first send.

[Link to: Full Event Guest Confirmation configuration →]

OpenClaw Use Cases — Health & Household: 3 Configurations for Personal and Family Management

This sub-theme contains the lowest-complexity cases in the Productivity category. All three run on Haiku and have the shortest setup time in the entire library. If you are new to OpenClaw and want a working automation within one session, start here.

Case 13 — Health & Symptom Tracker

Model: claude-haiku-4-5-20251001

What it does: Accepts daily health log inputs — symptoms, sleep quality, energy level, medications taken — via a conversational OpenClaw interface and maintains a structured log. On demand, generates a trend summary formatted for review or for sharing with a healthcare provider.

System prompt structure: health log assistant role; daily log input schema (date, sleep hours, energy 1–5, symptoms free-text, medications); trend summary output format (weekly averages, notable patterns, days flagged); privacy note (data stays local); explicit disclaimer that the agent is not a medical advisor.

Configuration: Complete universal setup with claude-haiku-4-5-20251001. Define your log schema. Run a 7-day sample log in AI Studio to verify trend summary quality and disclaimer placement before daily use.

[Link to: Full Health & Symptom Tracker configuration →]

Case 14 — Family Calendar & Household Assistant

Model: claude-haiku-4-5-20251001

What it does: Manages shared family scheduling, household task assignment, and recurring reminder generation. Accepts natural language input from multiple family members and maintains a unified household context — who has what commitment when, and what tasks are assigned to whom.

System prompt structure: household coordinator role; family member roster with names; shared calendar schema; task assignment rules (fair distribution or by person preference); recurring reminder definitions (weekly groceries, monthly bills, seasonal tasks).

Configuration: Complete universal setup with claude-haiku-4-5-20251001. Define your family member roster, task assignment rules, and recurring items in the system prompt. Test multi-member scheduling scenarios — two conflicting appointments on the same day — in AI Studio before daily use.

[Link to: Full Family Calendar & Household Assistant configuration →]

Case 15 — Second Brain

Model: claude-haiku-4-5-20251001

What it does: Captures, organizes, and retrieves knowledge from notes, bookmarks, and reading highlights. Accepts unstructured input — a raw note, a URL, a highlight — and returns a structured note with summary, tags, and connections to related concepts already in the knowledge base.

System prompt structure: knowledge management role; input types accepted (raw note, URL, reading highlight, idea fragment); output schema (title, summary max 60 words, tags, related concept list); retrieval query format for searching the knowledge base.

Configuration: Complete universal setup with claude-haiku-4-5-20251001. Define your note schema and tag taxonomy. Seed the knowledge base with 10 existing notes to establish initial context — paste them as a batch user message. Test retrieval queries in AI Studio with topic-based and keyword-based searches before daily capture use.

[Link to: Full Second Brain configuration →]

OpenClaw Use Cases: Productivity Category — Complete Model Routing Reference

Tier summary: 7 cases on Haiku · 7 cases on Sonnet · 1 case on Opus

Annual cost projection — correct routing vs. Opus-for-all-15:

Confirm all per-token pricing from https://wisgate.ai/models before publishing. Once confirmed, calculate:

• Sum of (estimated monthly token volume per case × correct tier rate × 12) for all 15 cases
• Sum of (estimated monthly token volume per case × Opus rate × 12) for all 15 cases
• Annual delta = Opus-for-all minus correct-routing cost

State the confirmed annual delta as a specific dollar figure. The 7 Haiku cases are the largest contributor to that delta — high-frequency, fixed-schema tasks run multiple times daily at Haiku cost versus Opus cost compounds significantly across a year. Confirm and publish the arithmetic before this article goes live.

OpenClaw Use Cases: Productivity — Your First Configuration

Fifteen complete Productivity automation configurations for OpenClaw via WisGate. Seven on Haiku, seven on Sonnet, one on Opus. All share the same base URL (https://wisgate.ai/v1), the same API key pattern, and the same 8-step setup sequence. Only the model ID and system prompt content change per case.

The Productivity category has the lowest average setup friction in the OpenClaw library. Every Haiku-tier case in this page can be configured, validated in AI Studio, and running on a cron schedule in under 30 minutes from a blank system prompt. The Sonnet cases take one to two hours. Only Autonomous Project Management requires a full planning session to define the project scope, team roster, and output schema correctly.

Your next step: Return to the selection guide in Section 2. Identify the case that matches your current highest pain point. Follow the configuration steps for that case. Run the output through AI Studio. Activate the trigger. Then repeat for the second case.

For the complete 36-case OpenClaw library across all 6 categories, return to the [OpenClaw Use Cases pillar page →].

Your first automation is one API key and one system prompt away. Go to wisgate.ai/hall/tokens — trial credits are included, no commitment required before your first scheduled run. Paste your system prompt into AI Studio, select the model tier from the routing table above, and validate the output before setting the cron schedule. Every case in this article uses the same base URL and API key. Switching between all 15 automations is one model ID parameter change and a system prompt swap. Pick the case that solves your most immediate problem today, run the validation in AI Studio, and activate it.

All per-token cost figures require confirmation from wisgate.ai/models before publication. Insert confirmed rates into the cost comparison table before this article goes live. Model pricing is subject to change. The Health & Symptom Tracker case includes a disclaimer that the agent is not a medical advisor — ensure this is preserved in all derivative content and in the deployed system prompt.