The AI work platform for professional services.
Twelve apps. AI staff that actually do the work. Client intelligence that compounds.
Helm is an AI work platform for professional services businesses. Three layers: twelve operational apps (CRM, projects, email, invoicing, documents, time tracking, and more), autonomous AI agents that act on the same data, and a compounding client intelligence layer. Per-workspace pricing. MCP-native. Built for agencies, freelancers, and service firms.
Three layers, built to work as one.
The apps your business runs on. The AI that runs them. The intelligence that gets sharper with every project, invoice, and client conversation.
Layer 1
Operations Platform
Twelve apps your team runs the business on. CRM, projects, email, sales pipeline, documents, time tracking, invoicing, estimates, calendar, workflows, forms, analytics. One data layer, no sync.
Layer 2
AI Staff
Autonomous agents with defined roles, configurable autonomy, and scoped access to every app. They work from inside the platform, not around it.
Layer 3
Client Intelligence
Every invoice, email, meeting, and project accretes into a non-transferable knowledge layer about your clients. It compounds. It stays when people leave.
Twelve apps. All included.
Everything your team needs to run the business, on one platform. No paywall between you and the app you already paid for.
Agents that actually do the work.
Pre-built roles with defined responsibilities and scoped tool access. Or build custom agents. Each runs in observe, suggest, or auto mode.
Bring your existing stack.
300+ integrations via Composio. Your agents read data, take action, and keep everything in sync. No rip-and-replace required.
Stripe
Slack
GitHub
HubSpot
Shopify
QuickBooks
Google
Figma
Klaviyo
Jira
Salesforce
Asana
Notion
LinearQuestions about the platform.
Why a platform and not best-of-breed tools?
Best-of-breed stacks win on features. They lose on context. A project in Asana doesn't know the client's invoice history. A CRM in HubSpot doesn't know the project is burning hours. AI agents on top of a disconnected stack spend most of their tokens piecing context together. Helm is one data layer, so agents get full context by default.
Is pricing really per-workspace and not per-seat?
Yes. Workspaces pay a flat plan fee. Seats are unlimited. Add your whole team, add your clients on the portal, add as many AI agents as your plan supports. The pricing page shows the current plan tiers and limits.
Can I bring my existing stack?
Yes. Helm connects to 300+ external tools via Composio (Stripe, Slack, GitHub, HubSpot, Shopify, QuickBooks, and more). Agents read and act on external data so you don't have to rip and replace. Over time, the apps you use in Helm displace the ones you don't need to keep.
How do AI agents work?
Each agent has a defined role, scoped tool access across the 12 apps, and a configurable autonomy level (observe, suggest, or auto). They're reachable on the surfaces your team uses: in-app chat, Slack, email, and MCP. Under the hood they run on Claude with full workspace context.
Where does my data live?
Your workspace data lives in Helm (Supabase Postgres, US region). External integrations connect via Composio with OAuth, and agents respect the permissions you grant them. No training on your data. Client portals share only what you explicitly expose.
Is Helm available for self-serve, or only agencies?
Both. Agencies are our beachhead (and get agency-specific features like white-label and rev share). But any professional services business, from solo freelancers to established firms, can sign up directly and self-serve. Per-workspace pricing means team size doesn't punish you.
See the platform in action.
Per-workspace pricing. No per-seat tax. Every app included.