Execution layer for OpenClaw: 4 patterns of parallelism,
proven by experiments and science.
Not one "parallelism" — four distinct patterns with different economics. The user thinks in tasks, we choose the pattern.
Self-delegation: agent decides — direct reply or spawn sub-agent. Routing rules in AGENTS.md. The simplest pattern.
/pc-ask — N models simultaneously, majority voting. Consensus for critical decisions. +17.9% GSM8K.
/pc-scan — Sharded throughput. Stateless HTTP parallel calls. 43× speedup on batch tasks.
/pc-do — Decomposition into stateful sub-agents with tools. Stanford: 7B beats 671B via specialization.
Key point: Talk While Work is a full-fledged pattern, not a "foundation layer." All four patterns are independent — you can use any without the others.
Self-delegation via AGENTS.md — agent decides: fast reply or spawn sub-agent for heavy work.
talk-while-workinstall.py adds a routing section to AGENTS.mdsessions_spawnGreetings, translations, simple questions, concept explanations, single-context debugging. Fast, cheap, no overhead.
Code generation, web research, artifacts (presentations, reports), multi-step business tasks. Heavy work runs in background.
Routing rules (few-shot examples): "hello" → direct · "translate" → direct · "build a bot" → spawn · "compare 5 DBs" → spawn · "why is it slow" → direct · "I need a script" → spawn
Consensus voting — one prompt to N models simultaneously, majority voting, judge synthesis.
Science: Self-Consistency (Wang et al.) — multiple reasoning paths + majority voting = +17.9% GSM8K, +11.0% SVAMP. Debate vs Voting (arxiv 2508.17536): majority voting explains nearly all quality gains, debate is 3× more expensive and not much better.
Honest disclaimer: Multi-Mind is the most expensive pattern per query. For every one Multi-Mind task, there are 20-50 Multi-Shard tasks in real work. Don't overuse it.
Sharded throughput — N chunks of work as parallel stateless HTTP calls.
Economics: 87 pages via Haiku = $0.17, 8.2s vs 4 min sequential. 43× speedup. The most underrated pattern. Batch processing revolution.
Why HTTP direct, not sub-agents: 10 HTTP calls = 26 sec. 10 sub-agents = 2-3 min. Sub-agent overhead (15-18s init) kills throughput. HTTP direct is 7.5× faster.
Decomposition of a complex task into N full stateful sub-agents with tools, context, and memory.
Science: Stanford Self-Play (arXiv:2604.20209) — 7B model beat 671B DeepSeek-Prover-V2 through 3-agent specialization (Solver + Conjecturer + Guide). Proof: parallel specialization beats monolithic scale.
Real numbers from this chat. Measured on HydraGPT + OpenClaw.
10 sub-agents in parallel in 4.2 sec. Wall-clock speedup 3.7× vs sequential. 8/10 returned results.
Fixed overhead per sub-agent: 15-18 sec (initialization + planning). For simple tasks, overhead exceeds useful work.
10 HTTP calls: 26 sec. 10 sub-agents: 2-3 min. HTTP direct is 7.5× faster → smart routing is mandatory.
model="..." in sessions_spawn is ignored. System picks from fallback chain. Workaround: pool-based round-robin.
Persistent sub-agents in Telegram direct chats are impossible. Sub-agents = fire-and-forget background workers, not chat companions.
Sub-agents can use web_search, exec, read, write. Verified. Contrary to GitHub issues.
Conclusion: Smart routing is mandatory. Simple tasks (<30 sec) → HTTP direct. Complex (research, code gen, file ops) → sub-agents with tools. Model override bug requires pool-based routing.
Not marketing. Concrete numbers from peer-reviewed papers.
Multiple reasoning paths + majority voting. +17.9% GSM8K, +11.0% SVAMP vs single-path. ParallelClaw: /pc-ask prompt --models 5.
Parallel path exploration beats single-chain for complex tasks. /pc-do task --branches 3 --depth 2.
Naive chunking → aggregator noise. Good MapReduce requires structured outputs (evidence + confidence) + intelligent reduce.
7B beats 671B through 3-agent specialization. Proof: parallel specialization beats monolithic scale.
Debate vs Voting (arxiv 2508.17536): Majority voting explains nearly all quality gains. Debate is 3× more expensive and not much better. Fan-out + voting = default. Debate only for adversarial use cases.
Honest: what works, what doesn't, and why it doesn't kill the product.
| Limitation | Workaround | Communication |
|---|---|---|
| Model override bug | Pool-based round-robin | "Any available from config" |
| Thread binding unavailable | Fire-and-forget subagents | "Background workers" |
| Sub-agent overhead 15-18s | HTTP direct for simple tasks | "Parallel for 3+ subtasks" |
| Sequential within session | Multiple sessions | "Multiple sessions for true parallel" |
Markdown instructions + slash commands. Zero infrastructure, open source. /pc-ask, /pc-scan, /pc-do.
Key vault, cost ledger, connection pool (AIMD), query cache. Observability and cost control.
Thompson sampling bandit. Personal benchmark. Warm-start from aggregated priors across all users.
Ready skill, packaged as .skill file. Works through AGENTS.md.
Install the talk-while-work skill — install.py adds routing section to AGENTS.md.
System prompt receives 70+ examples: when to direct, when to spawn. Works immediately.
True parallelism requires: model RPM limits, maxChildrenPerAgent in gateway config.
Hobby: 1 model, sequential fallback. Pro: multi-model pool. Power: adaptive routing + bandit.
Key difference from other patterns: Talk While Work is a pattern inside the agent itself (self-delegation). Multi-Mind/Shard/Agent are patterns for external orchestration. Different levels, not a hierarchy.
From landing page to daily use.
GIF: /pc-ask → 4 opinions + voting. Copy install command.
claude plugin install parallelclaw — 10-15s. Wizard: auto-detect API keys.
4 models, voting, synthesis. $0.018, 5.2s. Mass intelligence — impossible in Cursor or ChatGPT.
87-page PDF in 8.2s. Batch processing revolution.
/pc-ask for important decisions, /pc-scan for batch, /pc-do for deep research.
Saved time, win-rate, cost breakdown. Visible value.
After 100 calls — offer to enable bandit. "For Go code, you choose Sonnet 73% of the time."
Honest scope. Each week — a deliverable for the user.
Provider detection + wizard + keychain vault. pc.consensus, pc.shard, pc.compose. Skills /pc-ask, /pc-scan, /pc-do. SQLite cost ledger. Landing page with GIF.
/pc-dd "Company" — 7 branches, all 3 orchestration patterns. /pc-compete, /pc-research. YAML definitions. Benchmark verification.
Retry with exponential backoff + graceful degradation. CONTRIBUTING.md + architecture.md. GitHub labels. 30-sec demo video. ClawHub publication.
HN / Reddit / Habr launch. Outreach to adjacent tool authors. 5 design partners. Install-to-first-value < 10 min. /pc-stats MVP.
talk-while-work
Self-Consistency (Wang et al.)
Stanford Self-Play
Debate vs Voting
ParallelClaw MVP v3 · 4 weeks to first users · parallelclaw.ai