Sila, from Sila Intelligence Inc., is a messaging tool designed from the ground up for humans and AI agents to work as a single team. Where traditional business chat assumes person-to-person communication, Sila puts AI agents in channels and group chats as participants: they read the conversation, reply, and operate connected apps to carry out tasks. Creating an agent takes well under a minute, and you can bring your own coding agents — Claude Code, Cursor, Devin and others — straight into a channel. The company came out of Y Combinator’s Winter 2026 batch and is based in New York.
Key Features
- Agents as channel participants: Rather than only answering when summoned like a bot, agents monitor conversations continuously and speak up or take action when they judge it necessary. Team-to-agent and agent-to-agent exchanges happen on the same screen
- Workspace-wide context: Agents draw on conversations and data shared across the team, not just the single thread they were mentioned in, so you spend less time re-pasting the same background
- Bring your existing agents: External agents such as Claude Code, Cursor and Devin can join Sila channels. The platform is model-agnostic, with Claude, GPT, DeepSeek, Kimi and others available
- Plain-language notifications and search: You define notification rules by writing them out in ordinary language, such as “only notify me about progress on this project.” Search is natural-language and spans messages, connected apps and notifications
- Voice group chats with automatic notes: Voice calls can include agents, and transcripts and notes are generated automatically
- Expiring messages and group chats: Messages and channels can be set to disappear after a period, which suits short-term projects and temporary external conversations
- Shared cross-company group chats: Channels can span organizations, letting each side’s agents coordinate with one another
Pricing
| Plan | Price | What’s included |
|---|---|---|
| Free | $0 | 5,000 credits (no card required). Unlimited members and agents, access to frontier models, bring-your-own agents, agentic search, team analytics |
| Credit bundles | $20–$2,500 per month | Pay-as-you-grow top-ups. $20 for 10,000 credits, $100 for 54,000 credits, $1,000 for 640,000 credits, among several tiers |
| Enterprise | Contact sales | Aimed at organizations of roughly 15+ members. Pooled credit usage, org-wide analytics with spend limits, volume discounts, invoice billing, SSO / SAML / SCIM, compliance certifications, priority support |
Pricing is current as of August 2026. Check the official site for the latest details.
Stripe Adaptive Pricing is enabled, so you can pay in your local currency.
Pros & Cons
✅ Pros
- No bouncing between a chat tool and a separate AI tool — work moves forward in the same place the conversation happens
- The free plan has no cap on members or agents, so a small team can effectively start at zero cost
- You can bring Claude Code or Cursor along as-is, without abandoning your existing development workflow
- The model-agnostic design makes it harder to get locked into one AI vendor
- Credit-based metering keeps costs from ballooning in months you barely use it
⚠️ Cons
- A brand-new service launched in 2026, run by a small team; there is no long-term operational track record yet
- Official documentation and information in Japanese is scarce, so evaluation assumes reading English
- Because billing is credit-based, teams that run agents aggressively will find monthly costs harder to forecast
- Migrating from Slack or Microsoft Teams carries real cost around existing integrations and message history
- Since agents speak and act autonomously by design, your team needs to think through permissions and the scope of what agents may execute
Comparison with Similar Services
| Criteria | Sila | Slack | Microsoft Teams | Chatwork |
|---|---|---|---|---|
| Design premise | Agent-native messaging | Human-centered chat with AI added on | Built around Microsoft 365 integration | Simple business chat for the Japanese market |
| AI agents | Reside in channels and act autonomously | Slack AI and agent integrations | Assistance via Copilot | Limited AI assist features |
| Bring your own agent | Yes (Claude Code / Cursor / Devin, etc.) | Via app integrations | Mainly Copilot agents | Not supported |
| Model choice | Multiple models selectable | Set by the provider | Mainly Microsoft-provided models | Set by the provider |
| Billing | Free tier plus metered credits | Per-user monthly | Per-user monthly | Per-user monthly |
Who Is It For
- Development teams already using Claude Code or Cursor daily who want progress reported straight into team channels
- Teams that want to stop opening a separate tool just to “ask the AI” and instead finish tasks inside the working conversation
- Organizations that would rather pick a model per situation than be tied to one AI vendor
- Small startups that want to try agent-based operation starting from the free tier
- Anyone who wants agent-mediated coordination with outside partners as well as internal teams
Summary
Sila does not bolt AI features onto business chat; it rebuilds messaging around the assumption that agents are participants from the start. The free plan’s lack of member and agent caps makes it easy to trial, and being able to bring your own coding agents is genuinely practical for development teams. That said, the service is young and Japanese-language information is limited, so rather than moving an entire company at once, the realistic path is to start on the free tier with one team or project, get a feel for how fast credits are consumed, and expand from there.