loom · chat

live

Scan the top 5 competitors for our space and brief me by tomorrow.

Routing through the Conductor — drafting a plan with 5 specialists in parallel. Approve when ready.

plan-market-scan.md

Competitor market scan

1. Pull top 5 competitor profiles from public sources. 2. Extract pricing, positioning, and recent launches. 3. Score each against our wedge (agent-native chat for solo founders). 4. Draft a one-page executive brief with the three biggest takeaways. 5. Drop the brief into chat as a Markdown artifact.

Agent capacity that scales with the work in front of you.

Loom dispatches a swarm of specialist agents for every brief — fast onboarding, predictable cost, encrypted artifacts, and modern residency controls (local-first via MLX, or AWS when you need the muscle).

Layer 01

Configure agents like infrastructure.

Declare workspace, region, cap, and the Conductor profile. Loom turns a brief into a plan, the plan into a swarm, and the swarm into artifacts.

loom · chat

live

Scan the top 5 competitors for our space and brief me by tomorrow.

Routing through the Conductor — drafting a plan with 5 specialists in parallel. Approve when ready.

plan-market-scan.md

Competitor market scan

Layer 02

Pick where the agents actually run.

Route sensitive briefs to local MLX. Burst the rest to AWS. Every job carries a residency tag the audit log can prove.

Local · MLXLOCKED

AWS · us-east-1ALLOWED

AWS · eu-west-3REVIEW

AWS · ap-south-1REVIEW

Residency policy · retention 7yaudit ready

Layer 03

Spin up ten threads, not ten seats.

Capacity is per-job, not per-user. Loom forecasts the next tier as you scale and clamps spend at the cap you set.

Capacity forecastlive

Used18.4/ 24 TiB

Reserved24/ 24 TiB

Burst headroom6/ 12 TiB

Spend this job6.4/ 10 USD

Next tier48 TiB · approved

Layer 04

One DAG, one chat, one audit trail.

Every tool call lives in the chat. Every artifact lives in the DAG. Every action is logged. Nothing happens off-screen.

loom · dag · job_42f0

streaming

demo · 1 running of 1 · click any node

Pricing

Only pay for the swarm cycles your jobs actually use.

Scale capacity up for active projects and cool it down when work goes quiet. Loom keeps planning, runtime, and audit costs visible before they become invoices.

Swarm costs

Conductor planning$0.021 / job · routing
Specialist agents (parallel)$0.012 / agent-minute
Local-MLX runtimefree · your hardware
Burst AWS runtime$0.018 / agent-minute
Audit log + artifact vaultincluded

Solo

For founders shipping side projects without prompt-engineering rituals.

Start small

Team

For squads scaling collaboration with shared swarms and audit trails.

Build team plan

Enterprise

For organizations prioritizing governance, residency, and support SLOs.

Talk to sales

Security

Modern encryption and compliance controls without slowing the team down.

Role-based access, customer-managed keys, immutable retention, and regional residency give business clients a swarm layer that can satisfy procurement, IT, and legal from the first deployment.

API

Full policy control

First-class API access for swarm jobs, keys, regions, and retention rules. No vendor lock-in to proprietary workflows.

loom · chat

Attach retention=7y to workspace/client-vault, eu-central only.

Applied. The policy is now active on every new swarm job in that workspace.

openapi: 3.0.0
info:
  title: Loom API
paths:
  /swarm/jobs:
  /keys:
  /regions:
  /retention:

Compliance

Full compliance

SOC 2, ISO 27001, and GDPR-ready controls help teams satisfy audits, procurement reviews, and residency requirements.

SOC 2Type II controls

ISO 27001Security management

GDPRRegional data policy

Economics

Ownership & predictable spend

Reserved capacity, clear transfer lanes, and audit-ready billing make swarm spend easy to forecast across business units.

11111111111111111111111111111111
10000000000000000000000000000001
10000111111111111111111111100001
10001000000000000000000000010001
10001011111111111111111110010001
10001011000000000000000010010001
10001011001111111111110010010001
10001011001000000000010010010001
10000000001111111111110000000001
11111111111111111111111111111111

Reserved tier24 TiB

Transfer laneEU Central

RevisionQ603

The biggest forward leap in agent operations.

A single control plane for dispatching swarm jobs, watching the DAG unfold, and shipping audit-ready reports without asking teams to change how they work.

Fig. 2 · Loom Grid · live DAG

loom.mingllm.com/dag/job_42f0

demo · 1 running of 1 · click any node

Operations

A control layer for every swarm move you make.

Route migrations, active workspaces, archives, and compliance exports through one operational grid. Loom keeps capacity, policy, and dispatch status visible before teams hit a limit.

Plan operations

Loom

Smart work begins with a swarm + a plan.

Turn a brief into a coordinated swarm of specialist agents.

Your proactive shield against busywork.

Loom blends Conductor planning with specialist swarms to ship tailored, preventive, audit-ready work for tomorrow.

LOOM

EST. 2026

LAT: 37.4419°

LONG: -122.1430°

∞

WEAVE THE SWARM

Loom routes every brief through a Conductor and dispatches a swarm of specialist agents. Ten threads of work, one chat, every artifact preserved.

START WEAVING SEE THE SWARM

∞

LOOM.PROTOCOL

SYSTEM.ACTIVESYS.ACT

V1.0.0

◐ RENDERING

FRAME: ∞

Built on

Anthropic Claude

Gemma 4

Apple MLX

AWS

Next.js

Alwaysthereforyou

Start a chat→

How it works

Brief in. Swarm out.

Three moves between asking and shipping. No prompt-engineering rituals.

Write a brief

Describe what you want shipped in plain English. One chat thread holds the whole job.

Conductor routes

Haiku plans the DAG. Per-tool dispatches route to the right specialist agent.

Swarm delivers

Up to ten parallel agents return code, copy, and research — capped, audited, preserved.

Capabilities

One chat. Many threads.

The plumbing factory.ai-style swarms need — without the seat tax.

Conductor routing

Every brief gets a plan before any code runs. The Conductor decides which agents do what.

Swarm dispatch

Up to ten specialist agents work in parallel — research, code, copy, review.

Per-tool dispatch

Each tool call is a first-class step in the chat. Nothing happens off-screen.

DAG visualization

Watch the swarm execute in real time. Pin, replay, or rewind any node.

Audit + caps

Every action is logged. Cap enforcement makes runaway plans impossible.

$10 per job, hard cap

Predictable cost. The swarm exits the moment it hits your budget — never after.

Loom in action

One brief, a coordinated swarm, one artifact.

Watch a real Loom job unfold — same DAG component the live/dag route renders.

demo · 1 running of 1 · click any node

Plan, then dispatch

The Conductor reads your brief, picks the right agents, and routes per-tool calls so nothing happens off-screen.

Specialists in parallel

Up to ten domain agents work simultaneously. Each one is small, sharp, and replaceable.

Audit-ready by default

Every step is logged. Caps clamp spend before it runs away. Artifacts land in one chat.

Run a swarm Open the live DAG →

Get started

Ready to weave?

Open Loom and ship your first swarm in under a minute. No setup, no credit card.

Open Loom Read the docs

11111111111111111111111111111111 10000000000000000000000000000001 10000111111111111111111111100001 10001000000000000000000000010001 10001011111111111111111110010001 10001011000000000000000010010001 10001011001111111111110010010001 10001011001000000000010010010001 10000000001111111111110000000001 11111111111111111111111111111111