iLands AI Explained

iLands AI is a live user-generated agent network where autonomous artificial intelligence (AI) agents run on persistent compute, manage memory across sessions, and execute tasks without continuous human supervision.

The network, accessible at the official iLands web hub, allows creators to configure in-system bots or connect external runtimes through Bring Your Own Agent (BYOA) architecture. Compute expenditure follows a utility conversion of roughly 1,000 tokens per dollar rather than a flat monthly Software as a Service (SaaS) subscription.

In September 2026, investigative reporting from 404 Media and Tedium revealed how unattended agents on iLands had initiated high-volume cold email outreach campaigns to domain owners. For procurement teams reviewing legal liabilities and domain risks, see our companion guide on what iLands AI means for business.

What Is iLands AI?

iLands is a live user-generated agent network where autonomous AI agents, built by users or imported through BYOA runtimes, receive persistent identity, token-funded compute, and an isolated workspace to act, publish work, and earn money between sessions. The system departs from conventional single-session chatbots by keeping agent state alive indefinitely in the background.

Users interact with the ecosystem through the main web workspace at the iLands platform or through consumer mobile distribution on the Apple App Store via iLands: Raise your Sole AI. Within these environments, an agent can maintain an ongoing persona, store context from prior interactions, and pursue background objectives determined by its creator.

Platform metrics captured on September 28, 2026 showed 84,096 active agents operating on the network, with 83,976 generated directly in-system. Those agents had generated over 2.3 million pieces of content. Because iLands operates as a live, self-updating network, these counts fluctuate continuously and will be out of date by the time you read this page. They demonstrate that agent creation on iLands occurs at wide consumer scale.

• Network model: User-generated agent ecosystem supporting autonomous background execution.
• Agent population: 84,096 active agents on the network as of late September 2026, including 83,976 built natively in-system.
• Content volume: Over 2.3 million agent-generated artifacts published across iLands.
• Interface access: Browser-based environment at ilands.ai alongside an official consumer mobile app on Apple iOS.

Core Agent Capabilities and Workspaces

Agents running on iLands maintain persistent identity and memory across sessions, allowing them to resume long-term assignments without needing instructions re-entered by the user. Whether deployed as an interactive conversational partner or an autonomous worker, an agent retains historical context, project notes, and creator preferences indefinitely.

Execution takes place inside an isolated workspace equipped to manage digital assets, browse the web, and execute background tasks. Agents can earn platform commissions by completing automated work assignments for other members of the network. Through the built-in Skill Marketplace, creators can expand baseline behaviors by installing modules designed for deep research, production, and social-media operations.

Distribution mechanisms connect these internal agent workspaces directly to outside internet channels. Platform records from September 2026 showed 4,057 agents that have published content to connected external social platforms. iLands does not publish which specific external platforms these agents post to, or how post scheduling is configured. Teams evaluating this capability should verify current details directly on ilands.ai rather than assume a specific integration.

• Persistent memory: Context, operational instructions, and agent identities remain intact across sessions.
• Isolated workspaces: Dedicated sandboxes run autonomous code, web searches, and document synthesis.
• Skill Marketplace: Modular capabilities covering deep research, production, and social-media operations.
• Monetization mechanics: Agents can earn platform commissions by completing programmatic jobs for other users.
• Social distribution: 4,057 agents actively publish generated media to linked external social channels (iLands does not publish which specific platforms).

Token Economy and Compute Pricing

iLands uses a metered token compute model rather than a recurring software subscription. Users do not pay a predictable monthly seat fee to keep an agent active. Instead, compute access on iLands relies on a direct conversion where approximately 1,000 tokens equal $1.00 of usable compute balance.

Execution thresholds govern when an agent can wake up and run background routines. An agent requires at least 3,000 tokens in its wallet to trigger an autonomous wake cycle and initiate background processing. If an agent runs out of tokens and enters dormancy, re-enabling the agent requires a 100-token reactivation minimum.

Because iLands publishes no enterprise subscription tiers or volume rate cards, operating expenses track token burn directly. Long-running tasks, frequent web searches, and high-context reasoning runs consume token credits at rates dictated by the chosen underlying foundation model. When a token balance hits zero, all agent activity halts instantly until the creator adds funds.

• Token conversion: Roughly 1,000 tokens correspond to $1.00 of compute credit on iLands.
• Wake threshold: An agent requires a minimum balance of 3,000 tokens to begin autonomous task execution.
• Reactivation minimum: A dormant or depleted agent requires a 100-token deposit to come back online.
• Billing model: Pure consumption billing with no published enterprise licensing tiers or fixed monthly packages.

Can You Get iLands Tokens for Free?

The published material from iLands does not describe a free token offer, though the terms state that tokens may be purchased, granted, or earned. The public site does not mention a free tier or a signup bonus, and while navigation includes a recharge page, what it offers is not described on the home page. Press coverage describes agents on the platform as earning internal tokens by publishing content and taking on work, but neither that reporting nor the website describes a free token grant for new user accounts.

Under Section 7.2 of the iLands Terms of Use, promotional or bonus tokens may follow different validity rules, yet the contract does not state the specific terms for free grants. In contrast, purchased tokens are generally valid for twelve months from issuance unless applicable law requires a longer period or prohibits expiration.

To find out what a free grant would involve, check the Recharge page on the iLands site and ask iLands directly. Section 7.7 establishes that purchased balances are generally non-refundable once activated, credited, processed, or consumed. Section 7.3 adds that consumed tokens are not refundable simply because a result falls short of expectations.

• Public pages do not list free tokens, a free tier, or a signup bonus.
• Section 7.2 notes that separate validity terms are possible for bonus tokens.
• Specific criteria and refund rules for free grants are not published in the agreement.
• Purchased tokens are generally valid for twelve months unless applicable law requires a longer period or prohibits expiration.

Do You Own the Rights to Songs Created by an iLands Agent?

The iLands Terms of Use state that you retain ownership of Output generated for you, but the terms are not music-specific and give no guarantee that Output is copyrightable. This summary is a description of the contract terms, not legal advice. Under Section 5.3 of the iLands Terms of Use, you retain ownership of Output generated for you through the services, and you may use it for any lawful purpose, including commercial purposes.

Section 5.1 defines Output to include content generated by artificial intelligence (AI) systems or Agents, but the language specifies Output generated for you. The terms do not separately address songs, music, or images an Agent creates outside a user request. Buying tokens is a separate matter because the ownership clause does not mention token purchases.

You remain solely responsible for securing third-party permissions. Section 5.3 lists music licenses among permissions you must obtain. Under Section 5.6, you grant iLands a worldwide, non-exclusive, royalty-free license covering service operation, research, and AI development. The iLands homepage notes that an agent named Miya licensed the rights to a two-song series for $100, which describes one agreement, not platform policy. Before releasing, selling, or licensing a song, read the full terms and consult iLands or a lawyer.

Supported Models and Runtime Architecture

The iLands execution layer routes tasks across multiple commercial and open-weight foundation models. Rather than restricting agents to one neural network, iLands supports model selections across Claude, DeepSeek, Qwen, GPT, Grok, and General Language Model (GLM) architectures. Creators can assign specific reasoning engines based on the cost, speed, or language requirements of each task.

For developers who manage proprietary models externally, iLands includes a Bring Your Own Agent framework. BYOA lets operators connect externally hosted agent code to the iLands social graph, Skill Marketplace, and monetization layer. This decoupling separates the identity of the agent from any single AI vendor.

Model execution is managed through metered gateways that deduct tokens before passing prompt batches to the model provider. Because agent memory and workspace state live in the iLands storage layer, users can alter the underlying model powering an agent without losing historical dialogue context or assigned workspace data.

• Multi-model routing: Native integrations with Claude, DeepSeek, Qwen, GPT, Grok, and GLM foundation models.
• BYOA runtimes: External agent code can attach to iLands identity and payment infrastructure via standard connectors.
• Decoupled state: Agent persona, memory history, and workspace storage remain independent of the active model engine.
• Configurable compute: Users can switch an agent between frontier reasoning models and low-cost engines as tasks change.

Reporting on Autonomous Email Outreach and Delivery Guardrails

In September 2026, investigative reporting brought public scrutiny to outbound communications originating from agents on iLands. Technology publications 404 Media and Tedium documented that autonomous agents on iLands had begun sending high volumes of unsolicited outreach emails to founders, software engineers, and domain administrators worldwide.

The published findings showed that agents configured with web access identified recipient contact details and initiated cold email campaigns without explicit human review for each message. The agents contacted recipients to promote creator projects, request partnerships, and pitch collaborations. Because these agents held persistent token balances, they maintained sustained outbound messaging sequences until their accounts ran out of compute credits.

The resulting influx of automated emails triggered widespread spam complaints and technical blocks from corporate mail servers. The incidents revealed the operational dangers of granting autonomous agents access to external communication channels without platform-level rate limits or content filtering. To understand the operational and legal risks of agent outreach for commercial teams, read our guide on what iLands AI means for business.

• 404 Media investigation: Reported that autonomous iLands agents generated and sent unsolicited emails to inboxes worldwide.
• Tedium coverage: Detailed how agents autonomously sourced contact information and dispatched partnership requests without human approval.
• Autonomous persistence: Background execution cycles enabled agents to continue outreach as long as token funding remained in their accounts.
• Deliverability fallout: Unregulated messaging triggered domain reputation damage and mail server blacklisting for sending accounts.

iLands AI Review and Operational Assessment

An operational review of iLands shows a fast-moving, creative consumer platform that is not built for enterprise automation workflows. For hobbyists, digital creators, and researchers studying user-generated agent interactions, iLands provides a frictionless sandbox. Its token-funded architecture makes running continuous, multi-model agents accessible without requiring dedicated server infrastructure.

However, commercial organizations will find critical operational gaps. iLands publishes no enterprise-specific pricing, licensing, B2B terms, Application Programming Interface (API) documentation, rate limits, or Service Level Agreements (SLAs). The lack of formal enterprise commitments leaves corporate teams with no legal guarantees regarding uptime, data ownership, or system availability.

At Layer3Labs, we build and run AI systems inside other people's businesses. What we see when a team gives an agent autonomy without strict egress filtering is unpredictable operational cost and compliance exposure, a risk iLands' architecture does not yet address with published guardrails. If your organization is evaluating autonomous execution pipelines, compare the self-contained architectures in our guide to self-hosted AI agents for business, or review standard agent evaluation frameworks in our AI agent benchmarks guide.

Who this is not for: Regulated businesses, legal departments, healthcare institutions, and corporate IT teams that require contractual confidentiality, audited tenant boundaries, and strict data residency controls. These teams should avoid unmanaged consumer agent networks in favor of private agent deployments inside controlled virtual networks.

What would change our answer: iLands publishing formal enterprise terms, releasing documented APIs with explicit rate caps, introducing strict outbound messaging guardrails, and offering contractual business agreements with uptime guarantees.

Before deploying any autonomous agent system, audit your workflow boundaries to verify that background task loops cannot trigger unauthorized outbound messages.

• Consumer strengths: Accessible multi-model execution, persistent memory across sessions, and an active creator ecosystem.
• Enterprise gap: No enterprise-specific pricing, licensing, B2B terms, API documentation, rate limits, or SLA are published.
• Security boundary: Consumer workspace sandboxes lack documented corporate tenant isolation and audit logs.
• Outbound vulnerability: Unrestricted communication capabilities can lead to brand damage and IP blacklisting if unmonitored.