# OpenBot OpenBot combines persistent AI agents, shared chats, files and automations in one workspace. Models and execution environments can be selected for the task. Canonical profile: https://ki-agentenkompass.de/en/agents/openbot/ Editor: [Michael Meierhoff](https://ki-agentenkompass.de/en/about/) Sources reviewed: 2026-11-09 Basis: provider documentation. Examples are editorial suggestions, not hands-on product tests. ## Overview Independent project of the nightly-labs GitHub organization; this profile covers OpenBot at openbot.run. - Provider: Nightly Labs / Synthetify Labs - Product family: OpenBot - Profile type: Agent workspace with a desktop app and optional server operation. - Environment: Desktop for macOS, Windows and Linux, complemented by browser access and mobile companions. Servers can be self-hosted or managed team instances. - Connections: MCP servers and skills extend agents; browser and terminal support execution. The platform also lists Slack, Discord and Telegram. - Setup: Install the app, connect a model provider and create an agent. Configure its access mode, workspace and required connections before work. - Pricing model: Personal noncommercial self-hosted use is free; model charges are additional. Commercial licences and hosted servers are sold separately. ## Key features ### Let agents collaborate Agents can delegate work and collaborate with people in shared channels. Source: https://openbot.run/ ### Maintain memories Essential memories are supplied to the agent; additional entries can be searched when needed. Source: https://raw.githubusercontent.com/nightly-labs/openbot/main/src/backend/agent-memory-store.ts ### Add tools MCP connections and skills extend capabilities; the engine determines supported integration paths. Source: https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/skills-and-plugins.md ### Share browser control The built-in browser supports observation and manual takeover. Credentials can be supplied through a separate secure input path. Source: https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/browser.md ### Run routines Scheduled routines have persistent queues and their own execution state. Source: https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/agents.md ### Connect local models Ollama and LM Studio can connect through an OpenAI-compatible endpoint. Source: https://openbot.run/providers/local-models ## Example workflows ### Example: research with handoff - Your input: You give an agent a question and permitted sources. - Possible result: A research agent gathers material; another prepares a draft in the shared workspace. - What to check: You review sources, task allocation and conclusions before use. ### Example: recurring report - Your input: You configure a reporting routine with the required connections. - Possible result: The agent processes the task at the scheduled time and presents a result. - What to check: You check execution status, data freshness and approvals before further actions. ## Requirements and limits ### Check commercial licensing Requirement: The published PolyForm licence permits noncommercial use; business use requires a separate agreement. Source: https://github.com/nightly-labs/openbot ### Provide model access Requirement: Availability and costs depend on the connected provider or your own model infrastructure. Source: https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/providers.md ### Interfaces differ Limitation: According to the architecture documentation, some settings, including memories and routines, remain in the desktop app. Source: https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/browser.md ## A good fit if … - you want to coordinate multiple agents and people in one workspace - you want to switch providers or connect your own local models - you can assess access settings and commercial licence terms ## Less suitable if … - you need an open-source licence permitting unrestricted commercial use - you expect all administration settings in the browser Editorial assessment based on provider documentation. Not a product test. ## Usable AI models Open choice Providers and models are selectable; OpenRouter and local OpenAI-compatible endpoints are supported. Capabilities vary by engine. - Model families named: Anthropic (Claude), OpenAI (GPT), Google (Gemini) · Other models can be connected freely - Model choice: You choose - Your own API key: Yes - Router or custom endpoint: Yes - Local models: Yes Documented · Sources accessed on: 2026-11-09 - [OpenBot · Providers](https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/providers.md) - [OpenBot · Local](https://openbot.run/providers/local-models) - [OpenBot · Changelog](https://openbot.run/changelog) ## Europe & regional availability Product access, country availability and data processing are assessed separately. Reviewed scope: OpenBot · Documented product access Reviewed on: 2026-11-09 ### Access and regional conditions With requirements Desktop platforms and a Polish operator are documented. A complete country list or uniform EU storage location for hosted services and model providers is not publicly established. Region covered: EU/EEA, United Kingdom and Switzerland Documented reason: The conditions of the specific access route and connected services apply. Official source: https://openbot.run/ ## Accounts & permissions Based on provider documentation. Permissions depend on the account, integration and organization policies. Review the permissions actually requested before use. Configuration covered: OpenBot · Account and permissions Sources reviewed: 2026-11-09 ### Account used Documented with conditions: Agents have their own profiles and workspaces; team servers add human members and roles. Browser logins can use saved Vault entries; authorized people manage these credentials. Evidence: https://openbot.run/ Evidence: https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/browser.md ### Admin approval Documented with conditions: Access mode and automatic approvals are agent settings. Specific prompts depend on the engine and selected configuration. Workflows can contain explicit approval steps. Evidence: https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/providers.md Evidence: https://openbot.run/ ### Data access & team sharing Documented with conditions: Workspace only restricts writes to the workspace and additional allowed folders; broader modes expand access. MCP and skills expand reach according to configured tools and their credentials. Evidence: https://openbot.run/changelog Evidence: https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/skills-and-plugins.md ## Prices and included usage Pricing checked: 2026-11-09 Personal noncommercial self-hosted use is free; model charges are additional. Commercial licences and hosted servers are sold separately. ### Personal €0; model charges additional Personal noncommercial use on your own infrastructure. Price source: https://openbot.run/pricing ### Commercial Starter €10/month; new licences free until December 1, 2026 Commercial licence for one person; hosting separate. Price source: https://openbot.run/pricing ### Hosted Standard €50/month Hosted server; commercial licence additional where required. Prices exclude applicable VAT. Price source: https://openbot.run/pricing Pricing details: https://openbot.run/pricing ## Operation & oversight Based on public provider documentation. Assessments are editorial inferences from the stated properties. Actual time requirements and reliability have not been measured. Deployment covered: OpenBot · Execution and resumption Sources reviewed: 2026-11-09 ### Files and working state Documented: SQLite stores agents, conversations, queues and provider-session bindings. Evidence: https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/structure.md ### Sign-ins Documented with conditions: Provider logins reside in engine-specific environments; browser secrets can be entered separately from the model prompt. Evidence: https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/browser.md ### Starting further runs Documented: User requests and scheduled routines initiate work; routines are durably queued before execution. Evidence: https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/agents.md ### Human intervention Documented with conditions: People can take over the browser and answer approval requests. Exact intervention depends on the tool. Evidence: https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/browser.md ### Detecting errors Documented: Routines can explicitly report failures; their execution status reflects these reports. Evidence: https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/agents.md ### Resuming after failures Documented with conditions: Persisted routine queues can resume after a restart. Provider sessions have their own resume bindings. Evidence: https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/agents.md ### Editorial assessment Editorial assessment for the documented access route. Each criterion lists requirements and unresolved operating details. ## Memory and conversation continuity Scope: OpenBot: agent memories, channel and connection scopes; conversation archive is separate. Sources accessed: 2026-11-09 Persistent database entries with essential and searchable memories and automatic reviews. ### Storage and scope Documented: Memories are stored in SQLite; openbot.db is the central application database. Provider sessions have additional directories of their own. - [OpenBot · Structure](https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/structure.md) - [OpenBot · Memorystore](https://raw.githubusercontent.com/nightly-labs/openbot/main/src/backend/memory-store.ts) ### Selection and timing Documented: Automatic reviews are triggered by new conversation turns and idle time. Agents can stage memories; changes are committed only after a completed turn. - [OpenBot · Memoryreview](https://raw.githubusercontent.com/nightly-labs/openbot/main/src/backend/agent/memory-review.ts) - [OpenBot · Memorytools](https://raw.githubusercontent.com/nightly-labs/openbot/main/src/backend/agent/agent-memories.ts) ### Inspect, edit and delete Documented: Entries can be listed, created, updated, deleted and cleared together. Essential, searchable and user-controlled memories are distinguished. - [OpenBot · Memorytools](https://raw.githubusercontent.com/nightly-labs/openbot/main/src/backend/agent/agent-memories.ts) ### Agent and channel boundaries Partly documented: The storage model distinguishes agent, channel and messaging-connection scopes. A comprehensive security guarantee for their separation is not publicly established by this. - [OpenBot · Memorystore](https://raw.githubusercontent.com/nightly-labs/openbot/main/src/backend/memory-store.ts) ### Retrieval Documented: Essential memories are supplied in context. Searchable agent memories use SQLite full-text search with BM25 ranking; conversation search is a separate tool path. - [OpenBot · Agentmemory](https://raw.githubusercontent.com/nightly-labs/openbot/main/src/backend/agent-memory-store.ts) - [OpenBot · Agents](https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/agents.md) ### Session reset and compaction Partly documented: Database memories exist independently of individual chats. Context reset and visible compaction are documented; exact detail retention depends on the engine and is not uniformly established. - [OpenBot · Structure](https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/structure.md) - [OpenBot · Changelog](https://openbot.run/changelog) ## Context The published PolyForm licence permits noncommercial use; business use requires a separate agreement. Official product documentation reviewed. Workflows and assessments are editorial; hands-on product tests are outside the scope of this directory. ## Sources - https://openbot.run/ - https://github.com/nightly-labs/openbot - https://openbot.run/terms - https://openbot.run/pricing - https://openbot.run/changelog - https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/providers.md - https://openbot.run/providers/local-models - https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/structure.md - https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/agents.md - https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/skills-and-plugins.md - https://raw.githubusercontent.com/nightly-labs/openbot/main/docs/architecture/browser.md - https://raw.githubusercontent.com/nightly-labs/openbot/main/src/backend/memory-store.ts - https://raw.githubusercontent.com/nightly-labs/openbot/main/src/backend/agent-memory-store.ts - https://raw.githubusercontent.com/nightly-labs/openbot/main/src/backend/agent/agent-memories.ts - https://raw.githubusercontent.com/nightly-labs/openbot/main/src/backend/agent/memory-review.ts - https://openbot.run/brand ## Provider The terms identify Synthetify Labs sp. z o.o. in Poland as operator. Source code is published under PolyForm Noncommercial; commercial use requires a separate licence. - Operator: Nightly Labs / Synthetify Labs - Organization: Software provider - Website: https://openbot.run/ - Reviewed: 2026-11-09 - Provider source: https://openbot.run/ [Sources and method](https://ki-agentenkompass.de/en/sources/)