Feature Request

Business Problem: Currently, there's no way to test an AI Agent before deploying it in a live environment. This lack of a sandbox or preview mode creates friction during setup, increases the risk of misconfigured responses, and slows down iteration. Teams are forced to activate the agent to see how it behaves, which can lead to awkward customer experiences or internal confusion. Desired Outcome: A Test Mode for AI Agents that allows admins to simulate conversations, preview AI behavior, and fine-tune logic — all without affecting live customers. The test interface should reflect how the AI will respond in real scenarios. Use Case (Build + QA) Before Launch : Managers and admins can preview how the AI handles FAQs, product inquiries, or assignment flows. While Tuning : Teams can safely test tweaks to tone, logic, or integrations without needing to go live first. With a Test AI Agent mode, teams can confidently build, experiment, and perfect their AI experiences — without flying blind or breaking things in production 🔧🧪✅

Business Problem Currently, AI Agents created in one Workspace cannot be easily replicated or reused in another Workspace. For organizations managing multiple Workspaces (e.g., for different regions, brands, or business units), this creates duplicated effort. Admins must manually rebuild AI Agents in each Workspace, reconfiguring prompts, which is time-consuming and prone to errors. Desired Outcome Enable an Export / Import functionality for AI Agents, allowing Workspace Admins to: Export an AI Agent’s configuration into a transferable file or package. Import the AI Agent into another Workspace seamlessly. This would: Save time by avoiding repetitive manual configuration. Ensure consistency of AI Agents across Workspaces. Reduce errors and mismatches when scaling standardized AI-driven support or sales processes.

Business Problem: Currently, the Test AI Agent feature only supports text input, making testing less realistic. In real conversations, customers share files or images, but users can’t preview how the AI Agent will handle them before going live. This forces testing in live environments, risking poor responses. Users also refine prompts on the spot based on agent behavior, but without file support, they can’t fully validate or iterate quickly. Desired Outcome: Enable file uploads in Test AI Agent so users can: Validate how the AI Agent handles shared files Spot gaps or unexpected behavior before going live Adjust prompts or instructions before deploying live

Platform Problem Currently, AI Agents are not able to reply to specific messages and keep them in a thread. Desired Outcome To allow AI Agents to reply to specific messages and keep them in a thread for applicable channels (e.g. WhatsApp Business Platform (API), WhatsApp Cloud API, Telegram)

Business problem: Currently, when the AI agent provides an incorrect response, there is no immediate way to correct and train the agent, which could perpetuate the same errors. Desired outcome: Implement a feature that enables direct corrections to AI agent responses within the platform, enhancing the agent's learning process and improving its future accuracy. Use case: Users would have the ability to flag incorrect responses and provide the correct ones, which the AI agent would then learn from to enhance future interactions.

Business problem: Writing high-quality AI agent instructions is difficult and unforgiving: tiny changes in grammar, ordering, punctuation, list nesting, or casing can change AI behaviors entirely. Users must juggle formatting rules, variable placeholders, token budgets, and multilingual templates—without clear feedback—leading to inconsistent outputs, regressions, and lots of trial-and-error in production. Grammar/format sensitivity: small edits lead to different behaviors. Order-dependence: rule placement changes entire AI Agent outcomes. Placeholder risk: undefined/mismatched variables; template leaks. Token limits: verbosity drops critical rules. Ambiguity/drift: mixed tone/modes; vague refusals. Multilingual fragility: first vs. follow-up drift; placeholder mishandling. No feedback loop: no preview/diff/probes pre-ship. Desired outcome: Provide a chat-to-build prompt builder that prevents grammar/formatting errors to result in clean compiled instructions. This allows users to chat with another AI to build AI Agent prompts easily. This results in: Format-safe inputs: typed fields → clean and reliable prompts. Real-time linting: conflicts, placeholders, links, currency, length. Validator integration: one-click checks + fixes.

Business problem: AI Agents are often setup with instructions that look fine in setup but fail in production because there’s no post-setup QA: gaps, conflicts, and unsafe promises slip through, creating bad answers, compliance risk, and extra human handoffs. Teams end up debugging by visual checking instead of a repeatable checklist, for example: Frequent failure modes: conflicting rules (e.g., “auto-detect language” vs. hardcoded first message), undefined placeholders/links, and broken URLs. Operational gaps: pricing shared without delivery country, multi-currency in a single reply, vague handoff criteria, and missing refusal scripts (delivery/stock/payment). Resulting impact: lower trust/adoption, higher support load, slower time-to-live, and escalations that could have been prevented. Desired outcome: Provide a one-click “Validate instructions” action (right after Agent setup/edit) that scans the instruction set, flags issues with clear severity, and offers concrete, safe-by-default fixe — so teams can trust AI Agents confidently. Readiness checklist: must-haves for handoff rules, consent copy, stop/unsubscribe variants, product links, and technical notes on demand instructions. UX & reporting: inline validator with warnings for conflicting instructions

Business Problem Currently, the AI Agent cannot add internal comments to conversations. As a result, when a conversation is escalated, human agents must manually read through the entire conversation history to understand what happened. This slows down escalations and increases the risk of missing critical context, such as actions already taken, reasons for failure, or why an escalation occurred. Desired Outcome Enable the AI Agent to add internal comments within conversations. These comments could: Summarize actions taken by the AI Agent. Explain the reason for escalations or failures. Provide guidance to human agents on next steps. Tag or mention specific users/managers when attention is required.

Currently, we have typing indicators for both contacts and users. Users can see a typing indicator when other users are replying to contacts, and contacts can see a typing indicator when agents are replying to them (in supported channels). We are now looking to implement typing indicator for AI Agent, where its behavior will mirror that of actual users. Business problem: As a user, I want to know when an AI Agent is generating a response to a contact, so that I can focus on other leads that need attention and avoid sending duplicate replies. Desired outcome: Typing indicator showing when AI Agent is generating response Reduce duplicated replies Use Cases: Checking if AI Agent is generating responses Allowing contacts to see when a reply is coming (in supported channels) Prioritizing leads requiring manual follow-up

Business Problem: Currently, our platform lacks the feature to qualify leads based on specific desired criteria. This absence results in a less targeted approach to lead management and can potentially lead to inefficiencies in the sales process, as the sales team may spend time on leads that do not meet the desired criteria. Desired Outcome: Implement a new AI Agent objective that allows for the qualification of leads based on predefined criteria. This feature would enable a more focused and efficient approach to lead management, ensuring that the sales team's efforts are directed toward the most promising and relevant leads. Use case (Convert): This objective clearly supports the Convert phase of the sales cycle. It ensures only the most relevant leads are passed on to sales, optimizing time and resources while improving close rates. Benefits include: Shorter sales cycles Better pipeline hygiene Higher quality lead handover to human sales reps