Back to vendors
H

Helio

Also known as: helio.im, Helio AI teammates

Visit site
Entry price$20/mo (Pro; $16/mo billed annually)Full pricing detail

A shared workspace where persistent AI teammates own recurring work end to end, each with a name, a role, memory, and a lane, running scheduled jobs in channels, tasks, coding sessions, and email with explicit human approval on sensitive actions.

Helio is a workspace built around the idea that AI should hold a job instead of answering a prompt. Each AI teammate carries a persistent identity with a name, a defined role, an assigned model, and a lane of recurring work it is accountable for, and its memory holds team standards, relationships, and past decisions across sessions so it never restarts from zero. Teammates pick up recurring jobs on their own schedule without being prompted again and post results back into the channels and boards the team already watches.

Coding tasks open isolated persistent sessions running Claude Code or Codex with a real terminal, a per workspace filesystem, and reviewable diffs. Email style work runs through a native inbox with real addresses at the team domain, and outbound sends pass a human approval card before anything leaves. The control model is central to the design, not an afterthought. Sensitive actions such as external email, deploys, or messages sent on a person's behalf route to a human first, and closing a task is a human only step.

A library of installed skills gives teammates a set method per kind of work, and finished output publishes as persistent linkable artifacts. Helio is in public beta from a nine person San Francisco team, with web, macOS, and Windows clients, transparent credit based pricing from a free tier, and some capabilities still marked as in progress.

Vendor details

Canonical URL

https://www.helio.im

Category

Multi-agent platform

Subcategory

AI teammates workspace

Funding status

Not publicly disclosed

Company status

independent

Use cases & customers

Primary use cases

Recurring operational work owned end to end by persistent AI teammatesScheduled reporting, research, and ops tasks that run without re promptingCoding tasks executed in isolated Claude Code or Codex sessions with human reviewEmail triage, drafting, and routing with approval gated outbound sendsMeeting capture that turns decisions into tracked follow up tasks

Target customers

small teamsstartupsSMBoperations and engineering teams

Deployment options

SaaSmacOS desktop appWindows desktop app

Integrations

Connected tools the teammates read from and act in, including Slack, Microsoft Teams, Lark, Discord, GitHub, Linear, Vercel, Zoom, Google Meet, and Docker per the vendor; email via Gmail and SES; coding runtimes Claude Code and Codex; per user custom MCP tools persist across sessions.

In practice

An ops lead briefs an AI teammate once on a weekly vendor report; it runs every Monday, checks the channels, files the report as an artifact, and posts the task update on its own.

A small engineering team routes bug tickets to an AI colleague that opens a coding session, ships a reviewable diff, and waits for a human to approve and close the task.

Inbound email to a shared team address syncs into the right channel, an AI teammate drafts the reply with role aware context, and a human approves the send with one click.

Agentic Index coverage score

8.5 / 14 capabilities · 61%

Integrations & Tool Calling Full

Teammates work inside connected tools through channels, an inbox and integrations and write results back. Integrations cover chat with Slack, Microsoft Teams, Lark and Discord, developer tools with GitHub, Linear, Vercel and Docker, meetings with Zoom and Google Meet, and email through Gmail and SES. Google connections use OAuth, and tools can be scoped per teammate. A job can be limited to specific tools, such as HubSpot and Google Sheets, and uses only the channels and tools connected to it.

SourceHelio, helio.im/product and /securityread 2026-10-05

Workflow Orchestration Full

Teammates own lanes of recurring work, run it on schedule and return results to task boards and channels. Earlier versions also described atomic task claiming, so two agents cannot race, and a todo, in progress, in review and done lifecycle, but Helio no longer shows them. A teammate is briefed with a name, a role and an area of responsibility, much like a new hire, and posts status and results back to the board the team tracks.

SourceHelio, helio.im/productread 2026-10-05

Knowledge Grounding & RAG Partial

Teammates work with shared workspace files (5 GB to 500 GB by plan) and channel history. Helio offers no knowledge base or search index and no controls over what teammates draw on. Connected data is used only to carry out the team's instructions.

SourceHelio, helio.im/pricing and /securityread 2026-10-05

Human Oversight & Guardrails Full

Sensitive actions such as external email, deploys or messages sent on a person's behalf route to a human approval card before they land, live changes require approval, admins set boundaries where Helio stops every time, closing a task is a human only step, and approval threads keep requests and decisions visible. A boundary is set once, such as asking every time before a live change, and the request, progress and approval stay in one thread.

SourceHelio, helio.im/security and /productread 2026-10-05

Security, Identity & Governance Partial

Workspace roles carry approval authority, admins can remove access, each teammate's channels and tools can be scoped, and credentials sit in an envelope encrypted store. Helio names no attestation, certification or SSO. Google OAuth tokens sit in that store, and disconnecting a tool deletes them. Model providers are barred from training on customer data, and Google user data is not used for advertising, sale or general model training.

SourceHelio, helio.im/securityread 2026-10-05

Observability & Auditability Partial

Task status and channel history stay visible to the team, approval threads keep requests, progress and decisions in view, and coding sessions produce reviewable diffs, but there is no audit log, trace or run analytics. Artifacts publish as live pages that link to the latest version, and the full thread of a teammate's work stays open for review.

SourceHelio, helio.im/security and /productread 2026-10-05

Memory & State Persistence Full

Each teammate keeps the context for its job, holding standards, relationships and past decisions across sessions so recurring work does not start from scratch. Memory is tiered, Basic on the Free and Pro plans and Full on Max. The memory is searchable and also holds preferences, and full memory across conversations comes with Max.

SourceHelio, helio.im/product and /pricingread 2026-10-05

Deployment & Data Residency Not documented

Helio is hosted only. The web, macOS and Windows clients reach the same service, and there is no self hosting, VPC, region or residency option. Helio does not say where the service is hosted.

SourceHelio, helio.im/securityread 2026-10-05

Prebuilt Agents, Templates & Packs Partial

Installed skills give teammates a set method for kinds of work such as research, documents, email and support, and scenario roles are shown, but Helio offers no named prebuilt teammates or templates. Outbound agent templates were earlier marked as in progress.

SourceHelio, helio.im/productread 2026-10-05

Triggers & Channel Coverage Full

Work starts from daily, weekly or event triggered schedules without prompting again, from channel messages, from inbound email and from meetings a teammate attends. Plans cap scheduled tasks at 3, 20 and 100. A recurring job is scheduled once, and Helio handles every run after that.

SourceHelio, helio.im/product and /pricingread 2026-10-05

Model Flexibility & Routing Full

Each teammate is configured with a model, GPT and Claude models are included on Pro and Max, and coding sessions run Claude Code or Codex. There is no automatic routing. Advanced models are not included on the Free plan, and credits measure AI use at 200 credits to the dollar.

SourceHelio, helio.im/pricingread 2026-10-05

APIs, SDKs & MCP Extensibility Partial

Custom MCP tools persist per user across coding sessions, and skills and plugins have their own docs, but there is no public API or SDK for driving Helio.

Sourcehelio.im/productread 2026-09-27

Testing, Debugging & Optimization Not documented

There is no testing, evaluation or optimization tooling for teammate behavior, and review happens through human approval of outputs. Installed skills give each kind of job a set method, but Helio does not check how well teammates follow it.

SourceHelio, helio.im/productread 2026-10-05

Browser & Computer Use Not documented

Coding sessions provide a terminal and filesystem, which runs code but does not operate an interface, and there is no browser or computer use. Teammates work through the channels and tools a team connects and publish reports and dashboards as live pages.

SourceHelio, helio.im/productread 2026-10-05

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Pricing

$20/mo (Pro; $16/mo billed annually)

credits

Free tier

Included quota

Pro includes 9,000 credits per month (4,000 base plus 5,000 early pricing bonus), 20 AI teammates, 100 channels, 20 scheduled tasks, 50 GB storage; Max tiers scale to 750,000 credits and unlimited teammates

What is public

Complete self serve price list: Free, Pro $20/mo, Max tiers $100 to about $1,000/mo (all 20 percent lower billed annually), per plan credit allowances, the pay as you go top up rate, and per plan capability gates

Billing mechanics

Monthly or annual subscription carrying a monthly credit allowance; credits meter model usage across all teammate work at 200 credits per dollar equivalent; boosters top up one time; annual billing is one charge at 20 percent off with credits still arriving monthly

Cost watchouts

Early pricing bonus credit rates are flagged as changeable for new signups; Full memory and AI email are gated to Max tiers from $100/mo; unused monthly plan credits expire; Max usage tiers reach roughly $1,000/mo list at 50x

Variable cost rationale

All teammate activity (channel replies, task runs, meeting attendance) meters against the credit balance, so heavy months cost more, but the plan includes a substantial quota, the top up rate is published, and exhaustion pauses work rather than billing silently

Overage / add-ons

Boost top ups at a published 200 credits per dollar (from $20); when credits run out teammates pause and notify; plan credits expire at cycle end while booster credits last 12 months

Sales call required

No, self serve available

Free / trial

Free plan with a one time 1,000 credit grant, no card required

Lowest paid plan

Pro

Commercial notes

Cancel anytime from billing settings with the plan active through the paid period; no enterprise or sales led tier is published; public beta with early pricing bonuses explicitly subject to change

Key ambiguities

The pricing page FAQ describes Pro as including 8,000 credits while the plan card shows 9,000 (4,000 base plus 5,000 bonus); treated as early pricing drift and recorded from the plan card

Missing data

No enterprise tier, no published SLA, no security attestations; credit consumption rates per unit of real work (per task, per meeting) are not published, so forecasting a workload in credits requires trial usage

Agentic Index verified 2026-09-27

Alternatives to Helio

The closest documented capability profiles to Helio among multi-agent platforms tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.

  • Mindra9.5 / 14Adds documented Testing, Debugging & Optimization
  • Pokee AI8.5 / 14Adds documented Deployment & Data Residency
  • Agent Zero10.5 / 14Adds documented Deployment & Data Residency and Testing, Debugging & Optimization, among others
  • Akka12.5 / 14Adds documented Deployment & Data Residency and Testing, Debugging & Optimization
  • Artian11.0 / 14Adds documented Deployment & Data Residency and Testing, Debugging & Optimization
  • CrewAI13.0 / 14Adds documented Deployment & Data Residency and Testing, Debugging & Optimization, among others

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.