Sébastien Laframboise

ConvoViva WordPress Plugin: Integration and Comparison

AI-powered WordPress plugin

AI agents tailored to your WordPress site

ConvoViva for WordPress is a plugin that lets you add one or more AI agents directly to your site. It uses the OpenAI Responses API and can operate in memoryless chat for simple responses, or in chat with memory to retain conversation context.

Each agent can have its own instructions, OpenAI model, operating mode, access policy, its shortcode, its floating bubble, its appearance, sounds and, if needed, a knowledge base. Guided setup, human oversight, quality and results reports, and OpenAI usage tracking then help you manage the service from WordPress.

Buy the pluginSee options
  • Multiple AI agentsOne agent per need
  • With or without memoryDepending on the conversation
  • Knowledge baseWordPress files and content

Compare modes

With or without memory: choose based on need

Use a quick exchange for simple requests or retain context when a conversation needs to continue.

Simple queries

Memory-free chat

Ideal for quickly answering common questions, recommending a service, or directing a visitor to the right page.

How it works

  • Each message is handled separately.
  • No conversation history is stored on the AI side.
  • Responses use OpenAI’s Responses API.
  • Local responses can reply before calling OpenAI.
  • A knowledge base can be queried with File Search without retaining context between messages.
  • Model, instructions, appearance, sounds, and logs can be configured per agent.

Use for

  • Automated FAQ.
  • Simple queries.
  • Redirecting to a service, page, or product.
  • Light prequalification of requests.
  • Repetitive responses that can be handled locally without consuming the OpenAI API.
  • Documented responses that do not require conversational tracking.

Main benefit

This mode simplifies setup and is well suited to a site that primarily wants to answer common requests with minimal context to retain.

Tracked conversations

Chat with memory

Designed for longer exchanges: the agent retains context and responds more clearly as the messages progress.

How it works

  • Conversational memory managed with the Responses API.
  • A conversation ID allows context to be maintained.
  • Logged-in users can retrieve their context via their WordPress profile.
  • A new conversation button can be displayed to start fresh.
  • The same knowledge base via OpenAI Vector Store and File Search can also be used in this mode.

Use for

  • Long conversations.
  • Customer support more contextual.
  • Assistance before a quote request.
  • Recommendations based on the current discussion.
  • Responses supported by synced files or WordPress content.

Main benefit

This mode provides a more natural experience when a visitor exchanges multiple messages with the AI agent and responses need to take into account what has already been said.

Agents and customization

An AI agent for every need

Create distinct agents, prepare them with built-in assistants, and tailor their instructions, access, display, and appearance.

One agent per needYou can create separate agents according to your services, products, or visitor types.
  • Technical support agent.
  • Sales or prequalification agent.
  • Service-specific agent.
  • Name and slug editable for each agent.
Independent configurationEach agent can have its own settings to adapt its behaviour to the context in which it’s displayed.
  • Custom AI instructions.
  • Separate OpenAI model.
  • Stateless or memory mode.
  • Per-agent bubble settings, response formats, human handoff, quick buttons, sounds, theme and knowledge base.
Eight-step guided setupA resumable flow helps you validate services, prepare an agent, run a live test and choose how to publish it.
  • Validation of license, OpenAI source and effective model.
  • Choose or create an agent and optionally prepare its knowledge.
  • Save the essential look and test with a real OpenAI response.
  • Safety check then publish via bubble, shortcode or no immediate publication.
Assistant and instruction verificationPrepare, check and preview the actual instructions used before saving the agent.
  • Generate or improve from only the sources you authorize.
  • Choose profile, tone, purpose, contact collection and human handoff.
  • Draft verification against the model, allowed formats, knowledge, human handoff and active languages.
  • Read-only preview of the effective instructions by language; applying a proposal never automatically saves the agent.
Agent-specific shortcodeAn agent can be displayed anywhere in WordPress using a dedicated shortcode.
  • Display inside a page, post or any compatible area.
  • Ability to target a specific agent using its slug.
  • The role-based access policy configured on the agent also applies to the shortcode and public requests.
  • Language enforced when needed, with automatic detection on sites WPML or Polylang.
Floating bubble and advanced targetingDisplay the relevant agent based on location, audience, and the site’s current context.
  • Optional activation.
  • Basic display site-wide or only on selected pages.
  • Optional targeting by URL or pattern, content type, WooCommerce category, schedule, device, and language.
  • Match all conditions or at least one condition.
  • Access policy by role and button anchoring to the left or right.
Visual themesThe chatbot’s appearance can be adjusted without directly editing the plugin code.
  • Choose a theme per agent.
  • Built-in color editor.
  • Visual rendering preview.
  • Option to generate a theme with AI.
  • Built-in icon or media library image for the header and floating button.
CSS, sounds and interface detailsThe chatbot can be tailored to the site’s visual identity and desired experience.
  • Additional CSS per agent.
  • CSS selectors provided in the admin.
  • Optional sound effects for primary actions.
  • Customizable mic and Send button via SVG, emoji, or image.
  • Floating window movable, maximizable and resizable on compatible screens.
  • Configurable animated bar, footer and new conversation button.

Conversation experience

More helpful interactions for your visitors

Make conversations, multilingual support, and product recommendations easier while keeping control over responses.

Quick buttons and contextual suggestionsConfigured or suggested buttons throughout the conversation help the visitor move forward more easily.
  • Prepped questions or quick links configurable by agent.
  • Contextual response options suggested by the assistant.
  • Direct call to action when a link is more useful.
  • Text field always available so the visitor can reply freely.
Local greetings and responsesRespond instantly to common requests or greet the visitor before even calling OpenAI.
  • Rules based on configured keywords or phrases.
  • Triggers and responses adaptable to each language.
  • Automatic messages in the chat or near the floating bubble.
  • Fewer API calls for repetitive questions.
Languages and tailored contentLanguages are configured globally, then applied to agents and compatible interface elements.
  • Multiple languages can be enabled with one default language.
  • Automatic detection, including with WPML and Polylang.
  • Headings, footers, quick buttons, and local responses adaptable.
  • WooCommerce product button text translated according to the active language.
WooCommerce recommendationsOn a WooCommerce site, the agent can present relevant products directly in the conversation.
  • Cards with image, title, price, link, and translated button.
  • Supports simple and variable products.
  • Product search tailored to the visitor's language.
  • Contextual questions or action link options with recommendations.
Link trackingEasily measure visits generated by the chatbot’s links and recommendations.
  • Optional UTM settings for products, URL buttons, and suggested links.
  • Common settings or tailored per link type.
  • Preview tracked addresses before saving.
  • Internal tracking of clicks on the chatbot’s recommendations and actions.
Dictation, continuity and controlThe visitor enjoys a smooth experience while retaining the ability to reread, edit, or restart.
  • Voice dictation that can be reviewed and edited before sending.
  • Display history retained by agent and language.
  • Conversation continuity for visitors and signed-in members.
  • Optional new conversation button to start over.
Controlled message limitSet a common limit on messages sent to OpenAI in a conversation or choose unlimited operation.
  • Choose 10, 25, 50, 75, 100 or 200 messages, or unlimited operation.
  • Limit applied to modes with or without memory.
  • Local responses do not consume this OpenAI limit.
  • Clear message when the limit is reached and offer a new conversation if that option is enabled.

Content and search

A knowledge base unique to each agent

Link each agent to relevant WordPress files and content so it can search for the appropriate information.

Vector Store per agentWith or without memory, an agent can use a distinct OpenAI knowledge base.
  • Creation from WordPress.
  • Customizable Vector Store name.
  • Association specific to the configured agent.
  • Removable from the admin.
Uploaded filesDocuments can be added to the knowledge base to guide the agent’s responses.
  • Supported formats: PDF, TXT, MD, DOCX, HTML, and JSON.
  • Default maximum size: 20 MB per file.
  • List of files visible in WordPress.
  • Status, size, and deletion managed from the admin.
Synchronizable WordPress contentThe agent can also rely on content published on the site, without having to prepare separate files only.
  • Manual content selection.
  • Synchronization of all published content of the selected types.
  • Ability to exclude certain content.
  • List of synchronized content with their status.
Controlled synchronizationThe knowledge base can be maintained more easily as content evolves.
  • Manual synchronization from the admin.
  • Optional automatic synchronization when content is edited.
  • Last synchronization, status, and errors visible.
  • Remove WordPress content from the base as needed.
File SearchThe agent can search associated documents to use the most relevant information in its responses.
  • Search the agent’s Vector Store.
  • Maximum number of results configurable.
  • Adjustable score threshold.
  • Agent-specific documentation base.
Recommended useThe knowledge base is useful when responses need to be based on information specific to your site or organization.
  • Service documentation.
  • Important site pages.
  • Guides, policies, or procedures.
  • Sales or support information.

Operational handoff

Switch from AI to a person when needed

The visitor can request an explicit, consented follow-up, and your team processes the request in a protected flow without mixing the human channel with the OpenAI conversation.

Explicit, consented formEach agent can offer a configurable human follow-up request, disabled by default.
  • Name, email, phone, reason, contact method, and optional sharing of the history.
  • Explicit consent required before submission.
  • Deterministic button or AI-proposed structured prompt.
  • The prompt opens the form; only the visitor’s action creates the request.
Inbox and processing cycleTrack each request from the queue to resolution in a clear timeline.
  • Filters, counters, and controlled opening of useful details.
  • Assignment, reassignment, personal pickup, and re-queuing.
  • Start, resolve, or cancel according to an explicit cycle.
  • Protected timeline that limits conflicts between operators.
Secure human chatAn assigned operator can open a two-way text channel from the request.
  • Visitor and operator messages encrypted at rest.
  • AI responses suspended while the human channel is active.
  • AI resumes on close, expiration, cancellation, or resolution.
  • Human channel continuity when changing language or starting a new AI conversation.
Team, notes, and prioritiesThe ConvoViva Operator role provides processing tools without opening the plugin’s sensitive settings.
  • Dedicated WordPress role with limited operational permissions.
  • Priorities, approved tags, and encrypted internal notes.
  • Unread badges, read receipts, and typing indicators.
  • Contact details visible only to authorized persons.
Notifications and service organizationCustomize alerts, assignment, and SLA targets to match your internal workflows.
  • Optional email and in-browser notifications when the page is open.
  • Per-operator preferences for new requests, messages, emergencies, assignments, and delays.
  • Optional automatic assignment, first-response and resolution targets, followed by an operational report.
  • Optional outbound webhooks, signed and limited to metadata with no content or contact details.
Separate protection and retentionContact details, optional context, and human messages follow their own lifecycle.
  • Contact details and context encrypted and authenticated at rest.
  • No IP addresses retained in human follow-up requests.
  • Configurable retention periods for contact details and requests.
  • Support for WordPress tools to export and erase personal data.

Measurement, cost, and privacy

Track activity without losing control

Monitor usage, distinguish engagement from outcomes, track OpenAI calls, and apply appropriate retention rules.

Live conversationsWatch active and recently ended conversations in read-only mode.
  • View of ongoing and recent conversations.
  • Filters by agent and time period.
  • Display of visible responses, products, contextual choices, and errors.
  • Active conversation counter in the admin.
Usage and engagementIdentify the agents, languages, responses, pop-ups, and actions that drive the most activity.
  • Daily chart and exact range, including days with no activity.
  • Totals by agent, language, agent-language combo, and anonymous visitor within its retention window.
  • Popup performance and engagement calculated only on responses that offered an action.
  • Signed clicks on products, suggested links, and response choices, with filtered CSV exports.
Quality and outcomesMeasure explicit ratings and coded outcomes without treating a non-response or a simple click as success.
  • Yes/No resolution question, optional satisfaction from 1 to 5, and coded reason when Statistics consent is granted.
  • Final channel, end reason, and human follow-up outcome retained as separate dimensions.
  • Verified conversions recordable via a reliable server integration, without equating clicks to conversions.
  • Outcome coverage, silent exits, technical drop-offs, and aggregated CSV export.
OpenAI usage and costsView content-free telemetry on OpenAI calls and monitor an overall monthly budget.
  • Attempts, tokens, models, latency, retries, terminal failures, and daily trends.
  • Report-specific filters by agent, model, and status.
  • Local estimates in US dollars clearly flagged when incomplete.
  • Overall monthly budget and non-blocking email alerts at chosen thresholds.
Protected logsRecorded conversations are encrypted and their retention period stays under your control.
  • Encryption and authentication of stored conversations.
  • Masking of common sensitive identifiers in the live view.
  • Configurable retention for logs.
  • Live content automatically deleted after a short period.
Consent and lifecycleConfigure optional purposes and retain control of local data and OpenAI-managed resources.
  • Optional consent management that distinguishes required processing, site-retained history, and analytics.
  • Suggested privacy policy text plus WordPress exporters and erasers.
  • Non-destructive disablement and uninstallation by default; local purge only after explicit confirmation.
  • Resumeable assistant to clean conversations, Vector Stores, and referenced OpenAI files before local deletion.

Included features

Everything you need to run the chatbot

Essential features to install, customize, operate, and monitor your AI agents in WordPress.

Onboarding and access
  • Resumeable eight-step guided setup with a live OpenAI trial.
  • Shortcode per agent and language, or optional floating bubble.
  • Role-based access policy shared by the bubble, shortcodes and public requests.
  • Selected pages and advanced targeting by URL, content, category, schedule, device, or language.
OpenAI configuration
  • Use of the Responses API.
  • Per-agent OpenAI model selection.
  • API key encrypted in the plugin or WordPress OpenAI connector; no OpenAI Assistant ID required.
  • Two modes: memoryless or with memory.
  • Knowledge base and File Search usable in both modes.
Content control
  • Assistant for generating or refining instructions from authorized sources.
  • Draft checking and preview of effective instructions for each active language.
  • Structured formats allowed per agent: text, products, contextual choices, or prompt for human follow-up.
  • Local responses without API calls and knowledge base via files or WordPress content.
Visitor experience
  • Responsive interface; movable, maximizable, and resizable floating window on compatible screens.
  • Typing indicator and voice dictation when supported by the browser.
  • Quick buttons, contextual choices, and action links.
  • Configurable automated welcome messages.
  • Multilingual support, global message limit, and optional button to start a new conversation.
Human oversight and operations
  • Explicit, consented form, configurable per agent and disabled by default.
  • Inbox with assignment, priorities, tags, notes, and processing workflow.
  • Encrypted human chat that suspends the AI during handover.
  • ConvoViva Operator role with permissions limited to authorized tasks.
  • Notifications, automatic assignment, SLA targets, operational reporting, and optional outgoing webhooks.
Monitoring and analytics
  • Read-only live conversations.
  • Encrypted conversation logging.
  • Configurable retention period.
  • Usage, popup, and engagement metrics.
  • UTM tracking, signed clicks, resolution, satisfaction, quality, and outcomes.
  • Tokens, latency, reliability, estimated OpenAI costs, and budget alerts.
  • Filtered CSV exports of statistics, outcomes and logs.
Privacy and lifecycle
  • Optional consent management by purpose.
  • Encryption of stored secrets, conversations, contact details and human messages.
  • Masking of sensitive identifiers and configurable retention periods by data type.
  • Export and erasure via WordPress tools for identifiable personal data.
  • Default retention on uninstall, explicit local purge, and separate assistant for referenced OpenAI resources.
Maintenance
  • Updates included for 1 or 2 years, depending on the chosen plan.
  • Built-in licensing system.
  • Configured license key required to display the chatbot.
  • License usable on 1, 3, 5, 10 or 25 WordPress sites, depending on the chosen plan.
  • Compatible with WordPress 5.5+.
  • Compatible with PHP 7.4+.

Choose your plan

Three ways to proceed

Choose the level of autonomy and support that matches your site and technical comfort.

Option 1 · Self‑serve

Purchase the plugin only

This option is suitable if you're comfortable with WordPress, installing plugins, and configuring an OpenAI API key.

  • You purchase the plugin.
  • You install it yourself.
  • You use your own OpenAI account.
  • You configure your agents, their instructions, access, targeting, shortcodes, appearance, and sounds.
  • You can enable a knowledge base if your setup allows it.
  • You configure consent, human handoff, notifications, retention, and budget alerts as needed.
  • OpenAI fees are your responsibility.
  • Updates included for 1 or 2 years, depending on the chosen plan.
Buy the plugin
Option 2 · Assisted

Plugin with assistance

This option is suitable if you want to purchase the plugin but delegate installation, configuration, or adjustments.

  • Purchase of the plugin required.
  • Purchase of an additional block of hours.
  • Installation can be performed on your site.
  • Configuration of your OpenAI API key or the WordPress OpenAI connector.
  • Creation or adjustment of AI agents.
  • Configuration of AI instructions, access, targeting, shortcodes, appearance, sounds, and basic testing.
  • Knowledge base setup available, as required.
  • Consent configuration, human monitoring, and reporting available as per the agreed mandate.
  • Updates included for 1 or 2 years, depending on the chosen plan.
Buy the plugin
Add a hours block
Option 3 · Managed

Monthly managed service after setup

This option is suitable if you want me to handle the ongoing technical management of the chatbot after the initial integration is complete.

Custom annual plan, from 540 $ / year, or 45 $ / month

  • Purchase of the plugin required.
  • Plugin updates included for the duration of the chosen license.
  • Periodic checks to ensure the chatbot and configured agents are functioning properly.
See details
  • Maintenance of the API configuration used for the managed plan.
  • Basic monitoring of usage and estimated OpenAI costs in the built-in report.
  • Basic checks of active agents.
  • Basic checks of the floating bubble, shortcodes, and the configured knowledge base.
  • Initial setup not included.
  • Installation, integration, customization, drafting of AI instructions, creation of knowledge bases, and testing billed separately.
  • OpenAI usage included up to a limit equivalent to 50 % of the chosen monthly plan value.
  • If this limit is exceeded occasionally, the client will be notified.
  • If usage regularly exceeds this limit, a higher plan will need to be chosen or additional fees may apply.
  • Reasonable minor adjustments included after initial setup, as agreed.

The monthly plan does not cover the time required for a full or customized chatbot integration. Depending on needs, the initial setup may require several hours of work, sometimes 20 to 40 hours or more for an advanced project. These works are estimated separately before start.

Evaluate my project

Prerequisites

  • A WordPress site 5.5+ with PHP 7.4+ and plugin installation allowed.
  • A licensed key configured to display the widget or floating bubble.
  • A valid OpenAI API key, or a configured WordPress OpenAI connector, except in the case of a managed agreement where API access may be administered for you.
  • Credits or active billing on the OpenAI account in use.
  • AI agents configured according to your services, products, or response rules.
  • Relevant files or well-structured WordPress content if you want to use a knowledge base.
  • Clear choices for authorized roles, consent, retention and human review before production.

Please note before purchase

  • The annual managed service is offered after an initial configuration is validated. Needs analysis, custom integration, agent creation, drafting of AI instructions, preparation of knowledge-base files or content, advanced testing and major adjustments are billed separately.
  • OpenAI usage fees are not included in the price of the extension alone.
  • Installation, custom configuration and technical support are available for an extra fee, by hourly block or under a separate agreement.
  • Persistent conversations, use of a knowledge base, content synchronization and file searching can incur higher API costs than a simple chatbot without memory or knowledge, depending on usage volume and amount of content accessed.
  • Human oversight must be configured, assigned to authorized operators and governed by your own service and privacy rules.
  • Displayed OpenAI costs are local estimates; a verified conversion requires a reliable server integration and cannot be inferred from a single click.
  • The extension and its human oversight do not replace a qualified professional for complex, sensitive requests or those requiring professional judgment.
  • AI-generated responses depend heavily on the quality of the instructions provided, the files added and the WordPress content synchronized.
Read the disclaimer

Need help choosing?

Unsure whether to buy the extension only, add an hourly block or sign an annual managed agreement? I can help you choose the option best suited to your site, budget, level of autonomy, number of AI agents you need and the type of knowledge base to set up.

Discuss my project

Have a WordPress project in mind?

WordPress services available throughout Quebec.

Contact me for efficient WordPress service and lasting savings.

Contact me
WPentretien.ca — Visit the site

Sébastien Laframboise — WordPress Developer in Sainte-Adèle, Quebec
Maintenance, repair, technical support and development of WordPress sites in Quebec.
© 2026 Sébastien Laframboise. All rights reserved.