Botpress Prompts

Act as a conversation designer at [COMPANY]. I need a customer support agent handling [INTENT 1], [INTENT 2], and [INTENT 3] for [PRODUCT]. Walk through Botpress Studio: sketch the conversation as a flowchart on the drag-and-drop canvas (greeting, intent routing, [NUMBER] resolution branches, escalation node); layer LLM reasoning on top of the structured flows for open-ended questions; connect the knowledge base ([DOCS URL]) for RAG-grounded answers; and add the human handoff node ([CONDITION] triggers it, carrying the transcript). Include the test script ([NUMBER] conversations covering each branch plus [NUMBER] off-script questions), and the pre-launch checklist (fallback responses written, analytics events tagged, [STAKEHOLDER] sign-off).

Act as a knowledge manager at [COMPANY] with docs scattered across [GOOGLE DRIVE/NOTION/SHAREPOINT], our website ([URL]), and [NUMBER] YouTube tutorials. Configure Botpress knowledge base RAG: connect the sources (OAuth for Drive, Firecrawl crawl of the site with sitemap detection, transcript ingestion for the videos); set the chunking and the retrieval limit ([NUMBER] chunks per answer); define the freshness rule (re-index [SOURCE] every [FREQUENCY]); and write the fallback ('I could not find that in our docs; here is how to reach [TEAM]'). Include the [NUMBER]-question grounding test (each answer must cite its source chunk) and the monthly audit (who reviews unanswered questions and adds the missing docs).

Act as a platform engineer at [COMPANY] with strict data sovereignty ([REGULATION]). Deploy Botpress Community Edition self-hosted at zero licensing cost: provision [INFRA, e.g. a VPS with 4 vCPU/8GB RAM]; install via [DOCKER/COMPOSE or binary] following the official guide; configure [POSTGRES/REDIS] persistence and backups (daily to [LOCATION], retention [DAYS] days); put it behind [REVERSE PROXY] with TLS ([CERT METHOD]); and connect your own LLM keys ([PROVIDER]) since you manage model costs directly. Include the hardening checklist (no public admin port, secrets in [VAULT], log retention [DAYS] days), the upgrade procedure, and the monitoring alerts ([ALERT 1], [ALERT 2]).

Act as a CX lead at [COMPANY]. Our customers live on web chat, WhatsApp, Telegram, Messenger, Instagram, and Slack. Deploy one Botpress agent across all six: configure each channel integration (WhatsApp via [PROVIDER], Messenger/Instagram via Meta app [APP ID], Slack via [WORKSPACE]); adapt the greeting per channel ([WHATSAPP: short], [WEB: rich cards]); respect each channel's constraints (WhatsApp template rules for [USE CASE], Instagram's [LIMITATION]); and unify the conversation history so a user switching from Instagram to web keeps context. Include the per-channel test matrix ([NUMBER] tests each), the rollout order ([CHANNEL 1] first because [REASON]), and the analytics view comparing containment per channel.

Act as a support director at [COMPANY] ([NUMBER] agents, [TICKETING VOLUME]/month). Replace the patchwork with Botpress's AI-native helpdesk: enable AI ticket classification (intent, urgency, sentiment) on every incoming conversation; let the agent autonomously resolve the action-required tickets it can handle ([REFUNDS], [ACCOUNT CHANGES], [MULTI-STEP WORKFLOW]); configure the hot handoff (AI stays active through escalation, no cold drop, full context carried to the human); and give agents the workspace with conversation summaries. Include the autonomy boundaries (the agent never [FORBIDDEN ACTION] without approval), the $0-per-seat Team plan math for [NUMBER] agents, and the weekly calibration ([NUMBER] handoffs reviewed).

Act as a developer at [COMPANY]. The Botpress visual builder covers [PERCENT] percent of our [USE CASE] agent, but we need custom logic: call [INTERNAL API] with [AUTH], transform the response, and branch on [CONDITION]. Use the Node.js extensibility layer: write the custom action in [FILE/PATH] following Botpress conventions; handle the [NUMBER] error cases ([TIMEOUT], [AUTH FAILURE], [BAD RESPONSE]) with graceful fallbacks; expose the action as a Studio node so non-developers can use it; and document it in [DOC LOCATION]. Include the local test harness (mock the API with [TOOL]), the code review checklist for bot actions, and the versioning rule (actions versioned with the bot at [VERSION SCHEME]).

Act as an AI engineer at [COMPANY]. Our Botpress agent needs live data from [TOOL 1] and [TOOL 2], and Botpress can act as an MCP server for Claude Desktop, Cursor, and ChatGPT. Set it up both ways: expose the bot's [CAPABILITY] as MCP tools for [CLIENT, e.g. Claude Desktop] (configure the MCP endpoint, test with '[TEST QUERY]'); and connect external MCP servers into Botpress so the agent can call [EXTERNAL TOOL]. Include the auth model (which credentials flow where, rotation every [DAYS] days), the scoping (the agent may read [DATA] but never write [SENSITIVE DATA]), and the latency check ([TOOL] calls must complete within [SECONDS] seconds or the agent falls back to [FALLBACK]).

Act as a global CX lead at [COMPANY] ([COUNTRIES] markets). Botpress auto-translates to 100+ languages. Plan the launch: start with [LANGUAGE 1], [LANGUAGE 2], [LANGUAGE 3] ([PERCENT] percent of traffic); build the per-language test set ([NUMBER] questions covering [INTENT 1], [INTENT 2], plus our jargon: [TERM 1], [TERM 2]); have native reviewers ([ROLES]) validate tone and correctness; keep the canonical flows in [LANGUAGE] and let translation handle the rest. Include the quality gate per language (resolution within [POINTS] points of English before launch), the fallback (route to [LANGUAGE]-speaking humans below [CONFIDENCE]), and the expansion queue ([NEXT LANGUAGES]).

Act as an agency owner at [AGENCY] deploying Botpress agents for [CLIENT TYPE] clients. Use the white-label option (Plus/Team/Enterprise): remove Botpress branding from the webchat widget; apply the client's brand ([COLORS], [LOGO], [FONT]) to the widget styling; serve from the client's domain ([DOMAIN]) with custom API access; and package the analytics dashboard as a client report ([FREQUENCY], metrics: [METRIC 1], [METRIC 2]). Include the per-client setup checklist ([NUMBER] steps, [TIME] to deploy), the pricing model you charge ([MODEL]: setup [FEE] + monthly [FEE]), and the handover doc so the client's team can request changes without breaking the white label.

Act as a finance-minded founder at [COMPANY]. Botpress bills a base plan plus separate AI Spend (model fees at provider rates, zero markup, since the May 2026 update). Build our budget: estimate monthly conversations ([VOLUME]) and the average LLM cost per conversation ([COST]) for our [USE CASE]; pick the plan (Free: 100 conversations; Plus $150/mo; Team $750/mo with RBAC and real-time collab); model three scenarios ([VOLUME LOW], [VOLUME BASE], [VOLUME HIGH]); and set the alerting (notify [ROLE] at [PERCENT] percent of budget, auto-throttle [NON-CRITICAL FEATURE] at [PERCENT] percent). Include the monthly review ([MINUTES] minutes: spend versus forecast, top [NUMBER] cost-driving intents, optimization actions) and the rule for when to renegotiate the plan.