Carly
Also known as: Carly AI, usecarly, CalBot, Calbot Service
Email-native agent platform where every agent has its own address, memory, files and scoped tool access, sending and following up on its own instead of drafting, with customer-authored workflow graphs and a 122-tool MCP server for external AI clients.
Carly is an email-native agent platform, and the interface choice is the product. A user emails or copies an agent to delegate work in natural language, and every agent built from the dashboard carries its own email address, its own instructions, its own memory, its own files and its own configured tool access. Because an agent has a real address it can be looped into a thread with a client the way a colleague would be, and can reply to that client directly. Agents are also reachable through Slack, Telegram, WhatsApp and text.
What separates it from the inbox and meeting tools it sits beside is that it acts instead of drafting. The company draws the line itself. Many AI tools prepare a reply and wait for the user to press send, whereas Carly sends and follows up on its own. In appointment setting that means a missed meeting produces a same-day follow-up offering to rebook, and a request to move a time reopens the negotiation, finds a slot and updates the calendar unattended. Recurring instructions work the same way, from an hourly check of the inbox to a Monday reminder.
Work is composed as workflow graphs a customer authors or starts from a template, with triggers, actions and branches, and a lifecycle of test, run, pause, resume and archive. Template galleries cover agents, workflows across thirteen categories and a prompt library. A published MCP server exposes 122 tools for email, calendar, CRM, workflows, files, notes and booking pages to Claude, ChatGPT, Cursor and any other MCP client, authenticated with OAuth 2.1 and PKCE and bound to one person's account, with read-only tools running freely and writes asking first. A CLI and a Chrome extension sit alongside it. There is no REST API or SDK.
Integrations span property, recruiting, construction, finance and CRM systems by name, with anything uncovered reached using the customer's own API key, and a lightweight built-in CRM populates contacts automatically from email and calendar activity with natural-language search over that history. Deployment is cloud only with no published region or residency option, no model provider is disclosed, and the company publishes no security attestation. Pricing is self-serve, with a free tier containing deterministic workflows and no AI, and the assistant itself sold from a low monthly rate metered against an LLM credit allowance.
Vendor details
Canonical URL
https://www.usecarly.com
Category
Enterprise operations agent
Subcategory
Email native executive assistant and workflow agent
Funding status
Private, formerly CalBot before rebranding to Carly AI, with no funding detail found.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Carly gives its integration count variously as 120+, 200+ and 260+ across 45+ categories. Native CRM coverage spans HubSpot, Salesforce, Attio, Dynamics 365, Zoho and Pipedrive, alongside email (Gmail, Outlook), calendars, file storage (Google Drive, Dropbox, Box), messaging (Slack, WhatsApp, Telegram), video conferencing (Zoom, Google Meet, Microsoft Teams) and accounting (QuickBooks, Xero, FreshBooks). Each agent is given its own configured tool access instead of inheriting a global connection set.
In practice
A prospect misses a booked meeting and the agent follows up the same day offering to rebook, rather than the thread dying.
A lead goes quiet after a first conversation. The agent sends the follow up from your mailbox, updates the contact in the CRM, adds a to-do and later runs a scheduled check in on the dormant contact.
A team copies an agent on an email thread the way they would a colleague, and its work appears in the same thread everyone can already see.
Sources & related URLs
Related / legacy domains
Agentic Index coverage score
8.0 / 14 capabilities · 57%
| Integrations & Tool Calling | Full |
|---|---|
|
Named integrations run deep. In property alone Carly connects to AppFolio, BoldTrail, Buildium, CINC, DoorLoop, Entrata, Follow Up Boss, Guesty, Hospitable, Hostaway, Lofty, Propertyware, Real Geeks, RealPage, Rent Manager, Sierra Interactive, Wise Agent and Yardi. Bullhorn, CompanyCam, Deel, Greenhouse, JobTread, Outreach, Procore, Ramp, Salesloft, Smartsheet and Teamwork sit alongside them, and CRM coverage spans HubSpot, Salesforce, Attio, Dynamics 365, Zoho and Pipedrive. Anything without a built in connector is reached with the customer's own API key for that service. The customer pastes the key on the integrations page in the Carly dashboard, and the service then works the same way as the named integrations. The key belongs to the third party service and is not an API for Carly. Agents write as well as read. Through the MCP tools they send and reply from the customer's own mailbox, create, update, move and delete calendar events, add and update contacts, create and append to docs, append workbook rows, and create and share folders across Google Drive and OneDrive. Each agent gets its own configured tool access instead of inheriting a global set of connections. Carly gives its integration count variously as 120+, 200+ and 260+ across 45+ categories. Sourceusecarly.com/integrations/ named counterparties and own-key mechanism, usecarly.com/mcp/ tool tableread 2026-09-13 |
|
| Workflow Orchestration | Full |
|
Customers author workflow graphs or start from a template, then list, test, run, pause, resume and archive them, and schedule recurring tasks. Carly says it now does what Zapier does, with triggers, actions, branches and 200+ integrations, so a graph can branch on conditions. A workflows section offers an overview, a template gallery and thirteen categories, from inbox and calendar through sales, support and internal ops to simple (no AI). Customers also create named agents from a dashboard, each with its own instructions, configured tool access, permissions and email address. Work is divided between these specialized agents, and several can be copied onto the same thread. Appointment setting runs end to end without the user. A reschedule request reopens the negotiation, finds a new slot, updates the calendar and confirms, and a missed meeting produces a same day follow up offering to rebook. Instructions can be compound, such as "enrich new form leads and email me" and "pull the earnings reports and save them to a folder," and each chains retrieving, transforming and writing across different systems. Sourceusecarly.com/mcp/ Workflows tool group, usecarly.com/workflows/ template categories, usecarly.com agent modelread 2026-09-13 |
|
| Knowledge Grounding & RAG | Partial |
|
Agents can search mail, contacts, data tables, notes and files. They search and read mail, look up people, search, query and aggregate CRM records, create data tables and query rows, and save, search and delete notes. Through Google Drive and OneDrive they list and download documents in the customer's storage, and each agent also carries its own files. The built in lightweight CRM fills its contacts automatically from email and calendar activity and offers natural language search over that history. Carly maintains that CRM, and the customer does not curate it. Contact enrichment and web research reach beyond the customer's own data. There is no index, embeddings, vector store, semantic ranking, chunking or knowledge base product. Answers do not point back to the message, record or file they came from, and agents write to clients on the customer's behalf with nothing to stop an unsourced claim in an outbound email. Memory save and search work for each agent separately, and the connected tools fill these stores. Sourceusecarly.com/mcp/ tool table, usecarly.com CRM descriptionread 2026-09-13 |
|
| Human Oversight & Guardrails | Partial |
|
Only one path has an approval step. On email there is no approval step, by design. Carly sells that, saying many AI tools draft the reply and wait for you to hit send, "whereas Carly sends and follows up on its own." Through the MCP server, read only tools run without interrupting the user, and tools that write ask first. A person reviews and approves each write, one action at a time. Before an agent does anything, a person sets its configured tool access and permissions, and OAuth binds every call to the account that authorized it. When an agent decides on its own to send a follow up to a client from the customer's domain, there is no approval queue, confirmation step, risk tiering for each action or configurable autonomy level. Customers cannot require sign off before outbound mail goes out. Sourceusecarly.com/mcp/ authentication and tool-behavior statement, usecarly.com agent permission modelread 2026-09-13 |
|
| Security, Identity & Governance | Partial |
|
No SOC 2, ISO 27001, GDPR, HIPAA or CCPA claim is made, and there is no auditor, report, penetration test, trust center or compliance or security page. Legal terms come down to Terms and Privacy on calbotservice.com. Access runs on OAuth 2.1 with PKCE, with no API key to paste and no shared credential. Every connection is bound to one person's Carly account, and every call is scoped to the account that authorized it. Each agent also carries its own configured tool access and permissions, so authority is set for each agent instead of granted globally. Agents hold OAuth grants to a user's mailbox, calendar, CRM and cloud storage and send from the customer's own domain. There is no SSO, SCIM or RBAC. Sourceusecarly.com/mcp/ authentication section, usecarly.com full navigation and footerread 2026-09-13 |
|
| Observability & Auditability | Partial |
|
Each agent has its own email address and works in the thread, so everything it sends shows up in the inbox for everyone on that thread, including people outside the customer's organization. What a customer inspects is the actual email the agent sent, in the same place a colleague's would be. Workflows can be listed, paused, resumed and archived through the MCP tools as well as in the dashboard, so a customer can see what is configured and running and halt it. There is no run history, execution log, decision or reasoning trace, error or failure view, retry visibility or audit log of agent actions. The inbox shows what the agent sent, but not what it considered, what it read to decide, which tools it called, or what it tried and failed. Actions outside the inbox, such as a CRM record updated, a file moved or a tool call that errored, leave no trace. The lightweight CRM's contact history records the customer's business, not the agent's work. Sourceusecarly.com agent email-address model, usecarly.com/mcp/ workflow list, pause, resume and archive toolsread 2026-09-13 |
|
| Memory & State Persistence | Partial |
|
Each agent gets its own memory. Through the MCP server an outside caller can save, search and delete memory, manage to-dos, create data tables and query rows. Memory belongs to one agent, and OAuth binds every connection to one person's account, so state is kept separate instead of pooled. Deleting memory is one of the tools. Carly gives no lifetime or retention policy, so customers cannot tell how long memory lasts, whether it expires or what happens to it on cancellation. There is no trust or data handling page. Agents keep running after the user closes their laptop. A missed meeting produces a same day follow up, and a reschedule request can reopen a negotiation days later, so the agent carries the thread's state forward. Sourceusecarly.com/mcp/ tool table, usecarly.com agent modelread 2026-09-13 |
|
| Deployment & Data Residency | Not documented |
|
The product runs as cloud SaaS with no application to install, working over existing email and messaging channels. There is no region, region list, country, cloud provider or tenancy model, no self hosted or on premises option, no customer cloud path and no way to choose where it runs. Carly makes no residency claim. Carly works across three domains, usecarly.com for marketing, carlyassistant.com for the application and the MCP endpoint, and calbotservice.com for terms and privacy. None of them says where data is hosted, and there is no residency page. Sourceusecarly.com navigation and footer, usecarly.com/mcp/read 2026-09-13 |
|
| Prebuilt Agents, Templates & Packs | Full |
|
Agent Templates is a browsable catalog at usecarly.com/agents/templates/, beside a Prompt Library at /prompts/ and a Use Cases index. A separate workflow template gallery at usecarly.com/workflows/templates/ filters across thirteen named categories, each with a lasting query URL. The categories are inbox and email, executive assistant, calendar and meetings, personal assistant, sales and pipeline, customer and support, marketing and content, internal ops, daily newsletters, data briefings, booking pages, fun and team, and simple (no AI). Customers can start a workflow from a template, and what arrives is a configured graph that runs. Each template does its own job. The daily newsletter templates and the sales and pipeline templates are unrelated working units, and a client intake agent and a follow up agent each work on their own. Carly's Free, Personal and Enterprise tiers are pricing plans, not separate products. Free booking pages and the standalone utilities in the Free Tools menu are lead generation tools, not agents. Sourceusecarly.com/agents/templates/, usecarly.com/workflows/templates/ category filters, usecarly.com/mcp/ template toolingread 2026-09-13 |
|
| Triggers & Channel Coverage | Full |
|
Work starts in three ways. An inbound message can start it. Every agent has its own email address, and the pitch is "just CC Carly, she'll take care of the rest." Anyone on a thread can start agent work by replying, including people outside the customer's organization, so no separate app is needed. Agents are also reachable through Telegram, WhatsApp, Slack and text. A schedule can start it. Recurring instructions include "send a reminder every Monday to check in with ops" and "check my emails every hour and let me know if anything's important." Scheduling recurring tasks is a core operation in the MCP tools, and agents run scheduled check ins on dormant contacts. An event can start it. A missed meeting produces a same day follow up offering to rebook, a reschedule request reopens the negotiation automatically, and new form leads are enriched on arrival. All of this runs unattended. Carly sends and follows up on its own instead of drafting a reply for the user to send, and the missed meeting follow up goes out the same day whether or not the user opens anything. An outside caller can also drive the platform through the MCP server. Sourceusecarly.com homepage instruction examples and CC model, usecarly.com/mcp/ scheduling toolsread 2026-09-13 |
|
| Model Flexibility & Routing | Not documented |
|
The model behind Carly's own agents is not named, and neither is its provider or model family. There is no model selector, admin entitlement, model setting for each agent, disclosed routing, or way to bring your own model or key for inference. Customers can connect Carly to Claude, ChatGPT or any other MCP client, with setup instructions for Claude, Claude Code, ChatGPT, Cursor and VS Code and a tutorials category headed "ChatGPT & Claude (MCP)." In each case the customer's own assistant calls Carly through the MCP server. That choice is about which AI tool drives Carly, not which model Carly reasons with. Pricing is metered against an LLM credit allowance. The free tier contains no AI, so runs there are uncapped because nothing consumes credits. A customer's own API key, pasted on the integrations page, connects third party SaaS that has no built in connector. It is not used for a model endpoint. Sourceusecarly.com/mcp/, usecarly.com homepage, usecarly.com/integrations/read 2026-09-13 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
Carly's own MCP server does the work of an API or SDK, and its tools create and change the platform's own core objects. The endpoint is https://carlyassistant.com/mcp/, and auth is OAuth 2.1 with PKCE. Every connection is bound to one person's Carly account, with no shared credential. Its 122 tools are grouped by category across email, calendar, CRM, workflows, files and notes. The Workflows group authors workflow graphs or starts from a template, and lists, tests, runs, pauses, resumes, archives and schedules them, so an outside caller can build and run the platform's workflows. There is no REST or GraphQL reference, OpenAPI spec or SDK. Sourceusecarly.com/mcpread 2026-09-23 |
|
| Testing, Debugging & Optimization | Partial |
|
A workflow can be tested as a separate step from running it, so it can be tried before it goes on a schedule or a channel. An agent can also be tested in the browser before channels are added. Both let a customer try a configuration before it acts on live mail. A test shows only that the workflow ran, not whether the agent got it right. There are no test cases, scores, pass or fail verdicts, judges, golden sets, regression suites, or accuracy or quality metrics, and two versions of an agent cannot be compared. Credit and workflow run counts track consumption only. Sourceusecarly.com/mcp/ Workflows tool groupread 2026-09-13 |
|
| Browser & Computer Use | Not documented |
|
Agents do not drive a browser session, hosted or local, and have no desktop control and no remote or local computer control. Every route into another system is programmatic. The MCP server's 122 API backed tools cover email, calendar, CRM, workflows, files, notes and booking pages, and the customer's own API key reaches anything else. Carly presents API integration as the alternative to browser automation for recurring work. Carly offers a Chrome extension, which is a browser tool a person installs and uses. The agent does not operate a browser through it. The MCP tools can run sandboxed Python and pandas, which touches no interface. Agents send and read mail from their own address over a mail protocol, without driving a mail client's screen. Carly's premise is "no app needed." Agents live at the protocol layer and reach people through email instead of operating anything visual. Sourceusecarly.com/mcp/ tool table, usecarly.com/integrations/, usecarly.com no-app-needed positioningread 2026-09-13 |
|
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
AI agents from $35 per month, with Personal at $35 a month and Organization at $35 per seat per month. A free tier carries deterministic workflows and no AI, and Enterprise is custom.
A monthly subscription for AI agents, flat at $35 a month for one person on Personal and $35 per seat per month on Organization, with AI work metered against an LLM credit allowance. Workflow steps that call no model consume nothing.
Included quota
Personal includes 200 credits a month, and Organization includes a pooled allowance with usage visible per member. Workflow steps that do not invoke an AI model are free and unmetered.
What is public
Carly publishes four plans, Free at $0, Personal at $35 a month with 200 credits, Organization at $35 per seat per month with a pooled allowance, and Enterprise at a custom price. Free non AI workflow steps and free booking pages are stated across its product pages. The credit cost of an action and the overage terms are not published.
Billing mechanics
Self serve subscription. Non AI workflow steps are free, and AI agents are the paid layer. Personal is a flat $35 a month for one person, Organization bills $35 per seat per month with a pooled credit allowance and usage visible per member, and Enterprise is custom.
Cost watchouts
The free tier contains no AI. It is deterministic workflows and a calendar view, so it does not exercise the agents, and assessing them takes a paid plan. The $35 is a from price, and AI work is metered against an LLM credit allowance described only as generous, with 200 credits a month on Personal and no credit rate, pooled allowance figure or overage price published. Steps that call no model are free, so the plumbing half of an automation estate carries no marginal cost and exposure concentrates in AI work. Organization is priced per seat, so a team's bill grows with headcount. No security attestation exists at any tier, and advanced security and compliance sit in the custom priced Enterprise tier. Deployment is cloud only with no region or residency option to buy.
Variable cost rationale
AI work is metered in credits. Personal includes 200 credits a month and Organization a pooled allowance across its seats, while workflow steps that call no model are free and unmetered, so the bill concentrates on genuinely AI work. On Organization the bill also grows with each seat, at $35 per seat per month. No credit rate per action, overage price or top up is published, so what happens when an allowance runs out is not stated. Pooling on Organization spreads the allowance across the team, and the free tier carries no credits at all.
Additional watchouts
The bigger commitment is trust rather than price, because the agent sends on its own.
Overage / add-ons
Not published.
Sales call required
No, self serve available
Free / trial
Free plan at $0 with deterministic workflows, a multi calendar view and free booking pages, but no AI agents or credits
Lowest paid plan
Personal at $35 a month for one person, with 200 credits a month
Commercial notes
Carly charges only where a model is actually invoked, so it meters the intelligence and not the plumbing.
Key ambiguities
AI agents start at $35 per month on usecarly.com. Personal is $35 a month for one person, while Organization is $35 per seat per month, so a ten person team pays $350 a month. Personal includes 200 credits a month, but the size of the pooled Organization allowance, the credits an action consumes and what happens when credits run out are not published. Enterprise adds the customer's own email domain for agents, custom branding and advanced security and compliance, and whether it also carries API access is not stated.
Missing data
The size of the pooled Organization allowance, the credit cost of an action, overage terms, agent volume limits and Enterprise pricing.
Related vendors
- 11th Estate — Agentic platform whose AI engine scans markets, matches an…
- Abacum — FP&A platform for growing companies whose AI Modeler builds, edits…
- Adopt AI — AI native accounting and bookkeeping firm whose agents run…
- Aera Technology — Decision intelligence platform where an always on agent executes…
- Akro AI — On premise operational intelligence that automates document heavy…
- alfred_ — AI executive assistant that triages the inbox overnight and scores…
Alternatives to Carly
The closest documented capability profiles to Carly among enterprise operations agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Blue Yonder9.0 / 14Adds documented Deployment & Data Residency
- Infor7.0 / 14A lighter documented profile than Carly
- Cleavr8.5 / 14Fuller documented coverage on Human Oversight & Guardrails and Observability & Auditability
- Sales Layer8.5 / 14Fuller documented coverage on Human Oversight & Guardrails and Security, Identity & Governance
- Vooma8.5 / 14Adds documented Browser & Computer Use
- Celonis9.0 / 14Fuller documented coverage on Knowledge Grounding & RAG and Security, Identity & Governance
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded