Boost.ai Prompts

Act as a digital lead at [COMPANY] ([INDUSTRY]). Boost.ai's Automator can build a virtual agent from our website and chat logs in under 10 days. Run the sprint: day 1 to 2, feed it [WEBSITE URL] plus [NUMBER] months of chat logs from [SYSTEM]; day 3 to 5, review the generated intents and merge the duplicates ([EXAMPLE DUPLICATES]); day 6 to 8, wire the top [NUMBER] intents to real actions ([ORDER LOOKUP] via [API]); day 9 to 10, test with [NUMBER] real customer questions and fix the misses. Include the daily standup questions, the definition of done (resolution rate above [PERCENT] percent on the test set), and what we deliberately defer to phase 2 ([DEFERRED SCOPE]).

Act as a CX lead at [BANK/INSURER] in [COUNTRY]. Boost.ai ships with thousands of ready intents for financial services. Configure ours: enable the intent packs for [ACCOUNT INQUIRIES], [CARD BLOCKING], [CLAIM STATUS], and [BRANCH/ATM FINDER]; connect the core banking/insurance system ([SYSTEM]) for balance and status lookups with auth via [METHOD]; set the compliance guardrails ([REGULATION]: every advice-adjacent answer includes [DISCLAIMER], full audit logging on); define the handoff to human advisors ([CONDITION], with the conversation transcript and customer profile attached). Include the security review with [SECURITY TEAM] and the pilot ([NUMBER] percent of digital traffic, [WEEKS] weeks).

Act as a quality manager at [COMPANY] where our Boost.ai virtual agent handles [NUMBER] conversations/month. The new AI-powered CX Insights auto-reviews every conversation. Set it up: define what 'good' looks like for [INTENT 1] and [INTENT 2] (resolution confirmed, tone [TONE], no unnecessary steps); review the weekly auto-generated findings ([NUMBER] conversations flagged); turn the top [NUMBER] recurring issues into backlog items ([EXAMPLE FIX]); and track the trend (repeat-issue rate down [PERCENT] percent per quarter). Include the meeting rhythm ([MINUTES] minutes weekly: review findings, assign owners) and how to share wins with the team ([CHANNEL]).

Act as a service director at [COMPANY] ([CALL VOLUME] calls/month, [INDUSTRY]). Extend our Boost.ai virtual agent to voice with conversational IVR (no 'press 1' menus). Map the top [NUMBER] voice intents ([INTENT 1], [INTENT 2], [INTENT 3]); write the natural opening ('Thanks for calling [COMPANY], what can I do for you?'); handle the messy realities (background noise: ask to repeat once, then offer [FALLBACK]; authentication via [METHOD]); and define the transfer to human agents (warm transfer with a [NUMBER]-sentence spoken summary). Include the pilot ([NUMBER] percent of calls), the containment target ([PERCENT] percent), and the [REGULATION] recording consent script.

Act as a team lead at [COMPANY] with [NUMBER] support agents nervous that the virtual agent replaces them. Reframe with Boost.ai's no-code builder and AI trainer certification: select [NUMBER] agents as trainers; train them in [WEEKS] weeks to build and improve intents using the drag-and-drop conversation builder (no coding); give each trainer ownership of [INTENT AREA] ([EXAMPLE]); set the weekly improvement ritual ([NUMBER] failed conversations reviewed, intents fixed, re-tested). Include the career narrative ('you become the person who teaches the AI'), the certification milestone, and the metric that proves the model works (trainer-owned intents improving [POINTS] points/quarter).

Act as an operations manager at [COMPANY] where [NUMBER] human agents handle the complex [PERCENT] percent that the virtual agent escalates. Deploy Boost.ai agent assist: during live chats/calls, the AI suggests answers grounded in [KNOWLEDGE SOURCE] and recommends next-best actions ([ACTION 1], [ACTION 2]); agents accept, edit, or dismiss with one click; every dismissal is logged as training signal. Include the rollout (start with [TEAM], [WEEKS] weeks), the acceptance-rate target ([PERCENT] percent by month [N]), and the feedback loop (dismissed suggestions reviewed every [FREQUENCY] to improve the underlying intents).

Act as a CX lead at [COMPANY] expanding to [COUNTRY 1] and [COUNTRY 2]. Boost.ai supports 30+ languages. Plan it: launch [LANGUAGE 1] and [LANGUAGE 2] first ([PERCENT] percent of our volume); build a [NUMBER]-question test set per language covering [INTENT 1], [INTENT 2], and our trickiest jargon ([TERM]); have native speakers ([NAMES/ROLES]) validate tone ([TONE]) and correctness; set the fallback (below [CONFIDENCE], route to [LANGUAGE]-capable humans). Include the rollout order for languages [NUMBER] to [NUMBER], and the parity target (non-English resolution within [POINTS] points of English by [DATE]).

Act as an automation architect at [COMPANY]. Our virtual agent must do real work: read PDFs via OCR ([DOCUMENT TYPE]), trigger RPA bots in [UIPATH/OTHER] for [LEGACY SYSTEM] updates, and create tickets in [TICKETING SYSTEM]. Design the integration layer: map each intent to its backend action ([INTENT] triggers [ACTION] with parameters [PARAMS]); set the confirmation pattern (the agent states what it will do and waits for 'yes' before [IRREVERSIBLE ACTION]); handle failures (RPA timeout after [SECONDS]: tell the customer [MESSAGE] and open a ticket automatically). Include the security scoping (which credentials the agent holds, rotation every [DAYS] days) and the end-to-end test script.

Act as a digital services lead at [PUBLIC ORG] in [COUNTRY]. We need a virtual agent for [SERVICE, e.g. permit applications] serving [NUMBER] citizens/month, with strict compliance. Plan the Boost.ai deployment (100+ public organizations use it): define the in-scope services ([SERVICE 1], [SERVICE 2]) and the explicit out-of-scope list ([OUT OF SCOPE]); meet the accessibility standard ([WCAG LEVEL]) and language requirements ([LANGUAGES]); set data residency ([REGION]) and audit logging per [REGULATION]; and design the citizen-friendly tone ([TONE], plain language, no jargon). Include the stakeholder sign-offs ([STAKEHOLDER 1], [STAKEHOLDER 2]), the pilot ([CITY/DEPT], [WEEKS] weeks), and the success metric (task completion rate above [PERCENT] percent).

Act as a CX director at [COMPANY]. Our Boost.ai agent resolves [CURRENT] percent of requests; the goal is 90 percent+. Build the optimization program: analyze [NUMBER] escalated conversations and bucket them ([MISSING INTENT], [WEAK ANSWER], [SHOULD-ESCALATE]); fix the top [NUMBER] buckets (build [NUMBER] new intents for [TOPIC], rewrite [NUMBER] weak answers with [SME]); deliberately keep [CATEGORY] escalating (some things should stay human); and re-measure monthly. Include the dashboard (containment, CSAT of contained versus escalated, top [NUMBER] failure intents), the monthly review ritual, and the guardrail metric (CSAT must stay above [SCORE]; containment at the cost of satisfaction is failure).