In Meru, work with Agents to build anything.
Whether it's software engineering, scientific research, or even building a company.
00 / LIVE WORKBENCH · CLOSED-LOOP DEMO
STAGE 01 / 06
Auto Pilot
meru-studio
/
main · 4 agents online
YOU · OWNER
PROMPT #01
“Checkout sometimes freezes—fix it and add a dark mode.”
MERU · AGENT
SYNCED · 0.2s
Got it. Found the sync billing lock in checkout/pay.ts—splitting into 2 parallel sandboxes now.
01 Request → 02 Parallel Split
2 Sandboxes
Isolated billing lock; 2 worktrees coding async checkout & dark tokens in parallel.
03 Verify → 04 Data Observation
128/128 Passed
Caught 1 low-contrast dark border, auto-returned & fixed. All checks green.
RELEASE APPROVAL
Ready · Awaiting your approval
Describe what you want to build or run...
↑
⇄
Auto-returned to Step 02 on failed check
STEP 01
Request
Locate Lock · Split 2 Ways
target: checkout/pay.ts · dark.css
STEP 02
Parallel Split
2 Worktrees Code in Parallel
isolation: worktree-a ∥ worktree-b
STEP 03
Verify
Verify & Auto-Fix
audit: contrast 2.1:1 → fixed 4.8:1
STEP 04
Data Observation
Data Observation · Next Loop
next: Apple Pay share +38% ↺
↺
Live production metrics automatically spawn the next cycle
DATA OBSERVATION
0 Alerts · LIVE
P95 Latency
0.4s-89%
Night Conversion
+14.2%▲
HUMAN-AGENT LEVERAGE
STAGE 01 / 06
Your Input
1 sentence (12s)
Agent Closed-Loop Output
6 files · 128 checks passed
↺
Post-release data observation triggers the next iteration
State your request—Meru locates the billing lock and splits the work into 2 parallel sandboxes.
01 / CLOSED-LOOP ENGINE
One request in, a complete closed loop out.
From a single feature or daily task to an entire project.
01
Request
Describe your goal
02
Parallel Split
Run subtasks in parallel
03
Verify
Auto-test and fix issues
04
Release Approval
One-click human approval
05
Data Observation
Metrics drive the next step
MECHANISM 01 · PARALLEL SANDBOXES
Parallel Split & Execution
Zero Wait
sandbox-a / API refactor
100%
sandbox-b / UI interaction
100%
sandbox-c / E2E benchmark
100%
MECHANISM 02 · AUTOMATED VERIFICATION
Automated Verification & Fix
Verified Before Release
Code & Diffs
Parallel output
Independent Verify
Catches & fixes bugs
128 / 128
Ready for approval
MECHANISM 03 · RELEASE APPROVAL
Human Release Approval
Clear Boundary
Split · Code · Verify · Observe
Auto
Prod Release · External Send · Billing & IAM
Your Approval
MECHANISM 04 · DATA OBSERVATION
Post-Release Data Observation
Data Loop ↺
↺ NEXT TASK FROM DATA
Global Apple Pay Share +38% · Auto-Scoped
02 / SCALABLE AGENT TEAMS
Scale with resident Agent teams—from solo builders to full organizations.
Whether you are an individual developer, a one-person company, a research lab, or a scaling team.
INDIVIDUALS · LABS · TEAMS
You & Your Team · Multiple Resident Agent Teams Online
Human Input Today: 3 requests
Release Approvals: 2 clicks
Closed-Loop Deliveries: 19 items
Engineering Team
v2.4 Live
“Fix the checkout freeze and add a dark mode.”
01 Request
02 Parallel Split
03 Verify
04 Observe
Checkout P95 Latency
0.4s-89%
Research Team
200 Runs Done
“Run 200 ablation benchmarks and find the fastest, lowest-cost inference recipe.”
01 Request
02 Parallel Split
03 Verify
04 Observe
Unit Inference Cost
-42%99.4% Acc
Operations Team
62 Drafts Ready
“Audit all refund emails today and prepare 1-to-1 reply drafts for my approval.”
01 Request
02 Parallel Split
03 Verify
04 Observe
Audit Time per Ticket
4.0s0 Overreach
03 / FRONTIER ECOSYSTEM
Powered by the world's frontier models and Agents.
Zero vendor lock-in—frontier models and mainstream Agents share one unified context.
Frontier Foundation Models
Plug & Play · Multi-Model
Anthropic
Claude Opus 5.5
Deep architectural decomposition & complex systems engineering
OpenAI
GPT-6 Astra / GPT-6.1 Sol
Rigorous formal reasoning & cross-model adversarial verification
Google DeepMind
Gemini 4 Argon
Multi-million context retrieval & full-repository telemetry analysis
Open Weights
DeepSeek 4.1
High-concurrency sandbox execution & private self-hosted deployment
Native Agent Harnesses
Shared Context Memory
Claude Code
Repository-scale code reading, refactoring & multi-worktree execution
Native
OpenAI Codex
Sandboxed verification, automated test generation & security auditing
Native
Gemini CLI & Custom Agent (MCP)
Mount any custom Agent or internal toolchain via standard MCP
Native
04 / GET STARTED
Start building today.
Self-host the open-source monolith locally, or launch instantly on Meru Cloud.
OPEN SOURCE · APACHE-2.0
meru
Open-source monolith—run locally or on your own private servers with one command.
$curl -fsSL https://meru.sh/install | sh
github.com/Matthewoho/meru
View on GitHub →
MANAGED CLOUD · ZERO OPS
merucloud
Zero-ops managed cloud—your resident Agent organization stays online 24/7.
Elastic Parallel Sandboxes
All Frontier Models & Agents
Approve Key Gates from Mobile
Instant Setup · Pay-as-you-go
Launch Meru Cloud →