Conversation AI

Please make the conversation ai send image messages automatically, it is not friendly user too many restriction

Leads are distributed to different sales users in our instance and then our conversation AI tries to book an appointment with the sales rep. I can do the multiple calendar booking but it seems to only route by intent in the conversation. I would love it to be able to book the contact on their assigned owner's calendar. Maybe these should also be able to book based on other contact properties as well. I've tried to use a round robin calendar with the "Always book with the contact's assigned user" turned on, but the bot does not seem to respect that setting. It books with the first calendar on the list.

For product sellers using HighLevel for e-commerce, or those who have uploaded their products to HighLevel, it makes sense for the Conversation AI to access this product information directly, rather than requiring a separate table upload. The reason for this is that the product listings not only contain information about the products and their prices, but they also include inventory data. If we rely on a table to store this information for the AI, we would have to manually update both the table and the product listings, meaning we would need to maintain two separate records for the same product. Since the products are already uploaded with images, descriptions, price, promo price, variations, and quantity, it would be great to allow Conversation AI to read it as the KB.

Let us examine why the AI responded incorrectly and what factors contributed to that response. The Agent Log provides limited information; it only indicates whether the AI retrieved data from the knowledge base. It's frustrating to constantly wait for the Support Team to investigate issues when we need immediate responses for our clients. All we can tell them is, "We've logged a case," but the Support Team doesn't respond quickly either. Often, when they do get back to us, their feedback is vague, simply stating that the prompt needs to be amended. However, we lack insight into which prompts will work and which will not. The log is not very helpful if we cannot conduct our own investigations. What would be helpful is a log that shows what the AI did and how it decided to do it. Something like the attached example.

Once we connect a custom action or API call in AI voice agent, is it not possible for conversational AI to have access to the same API set up instead of setting it up again right from the start?

I think it would be super cool, if when creating a new conversation AI bot, if you could put a website in so that the bot understands what business it is being created for. So it would look at the website and then it would be able to understand something like the services that the company offers. That way it could use the website as a reference to answering questions

I INTEGRATED AN AI AGENT FOR A DISTRIBUTOR THAT HAD APPROXIMATELY 200 CUSTOMERS. THE IDEA WAS THAT I SERVE FROM MONDAY TO FRIDAY, TAKE ORDERS AND CONFIRM. I HAD TO MANAGE A CATALOG WITH 821 PRODUCTS, BUT I COULD NEVER DO IT, I ALWAYS HAD SOME MISTAKE, EITHER INVENTING PRODUCTS OR PRICES. WE PERFORM DIFFERENT PROMPTS AND UPLOAD DIFFERENT FILES IN DIFFERENT FORMATS TO THE KNOWLEDGE BASE, TO BE PRECISE. IT WAS NEVER POSSIBLE FOR THE AGENT TO BE PRECISE IN HIS ANSWERS. THIS LED TO THE CUSTOMER CANCELING MY SUBSCRIPTION. IN CONCLUSION, THE AI AGENT LEAVES A LOT TO BE DESIRED, NO MATTER HOW MANY LAPS YOU GIVE IT, IT DOESN'T END UP BEING EFFICIENT AND IT'S NOT SUITABLE FOR COMPLEX JOBS.

At the moment, there is not any option to set a time range to answer messages with AI conversations because there's moments when AI answers in a inappropriate moment.

With multiple team members, it's important to also know who made any changes to an AI Bot and allow notes before saving, this way we know a change was made by adding some context.

Summary / Problem: Currently, Conversational AI / Bot workflows struggle when extracting data for structured custom fields (dropdowns, radio buttons, single/multi-select checkboxes). When the AI attempts to collect this information: It does not automatically present the defined list of options to the user, leading to guesswork. If the user replies with freeform text or slight synonyms (e.g., replying "he's my dad" instead of the dropdown choice "Parent"), the update either fails validation, requires complex manual regex/webhook workarounds, or leaves the CRM field blank. This limitation applies across both Contact and Opportunity custom fields, making conversational qualification and data hygiene difficult to maintain. Proposed Solution: Enable the Conversational AI engine to natively interface with CRM field schemas for structured fields: Dynamic Option Listing: When the AI bot prompts for a custom field configured as a single/multi-choice (dropdown, radio, checkbox), allow it to dynamically retrieve the field's allowed options and render them inline (e.g., as a numbered or bulleted list: 1. Parent, 2. Sibling, 3. Friend). Fuzzy Intent Matching: Allow users to answer either by choosing the number/option directly (1) OR by answering conversationally ("he's my father"). The AI should use semantic matching to map the answer back to the exact allowed enum value (Parent) and commit it directly to the CRM. Contact & Opportunity Support: Extend this capability to both Contact-level and Opportunity-level custom fields so appointment setting, pipeline stage qualification, and lead enrichment can happen entirely inside the chat. Validation Fallback: If the user’s response cannot be reasonably mapped to an allowed option, the AI should gracefully reprompt: "Please select one of the following: [Options]". Key Use Cases & Benefits: Flawless Data Hygiene: Completely eliminates invalid values, failed workflow conditions, and dirty CRM data. Frictionless End-User Experience: Leads can simply tap/reply with a single digit (2) or talk naturally, reducing drop-offs. No Complex Workarounds: Eliminates the need for 10-step branch logic, custom code, or third-party webhooks to map a simple dropdown. #ConversationalAI, #CustomFields, #Workflows, #Opportunities.