Change the algorithm in the bot

Create a chat bot with an admin panel for a company that deals with solar panels for lead processing.

PP T.R.U.S.T — a Ukrainian transport company (freight transportation, KVED 49.41), Chernihiv. Fleet — 12 trucks, 18 employees.TRUST⚡️ — a Telegram bot for managing the operational activities of the fleet: drivers, vehicles, trailers, documents, trips, refuels, reports. The bot is the operational core of the company, Google Sheets is used as a data output and reporting layer.Some modules have already been implemented (see section 5), some are logically described but not coded. A developer is needed to complete and stabilize the entire system according to this technical specification. https://docs.google.com/document/d/1TyNC7tFKnNOn6imWczUHp-xgtcxUFyu8UtCM3UssVCY/edit?usp=sharing

A web application needs to be developed for automating order confirmation and offering additional products to customers via Viber. Orders are received in the CRM SalesDrive. General workflow: 1. The manager selects one or more orders from SalesDrive in the web application, chooses a scenario and one or more promotional flyers, and then initiates automatic processing. 2. A transactional message with order details: products, amount, payment method, city, delivery service, and branch/address is sent to the customer via Viber Business Messages. The customer chooses: — “Yes, continue”; — “Contact the manager”. 3. After confirmation, one or more selected promotional flyers with additional products are automatically sent to the customer. 4. The customer writes in free text form, in one or several messages, which products and in what quantity they want to add. After completing their selection, the customer presses the button “I have selected everything”. 5. The AI analyzes all customer messages and matches the selected items only with the products contained in the flyers sent to the customer. For each flyer, a structured list of products and their IDs/articles is stored in the system in advance. 6. The system generates a recognized list of products and quantities and sends it to the customer for confirmation. The customer chooses: — “Yes, correct”; — “Need to change”; — “Contact the manager”. 7. After confirmation, the selected additional products are automatically added to the existing order in SalesDrive. The amount and status of the order are changed. 8. After the order confirmation is completed, the customer is sent information about shipping times. Example: “The order will be shipped within 2 business days. If you need the fastest shipping, click the button below. We will take this into account when processing the order.” Buttons: — “All good”; — “Need fast shipping”. If the customer selects “Need fast shipping”, a corresponding mark or comment should be saved in the SalesDrive order. The automatic scenario continues at this point. 9. The customer is automatically sent an informational message about the need to inspect the package upon receipt. Example: “Please inspect the product at the delivery service branch upon receipt. If you find any mechanical damage, do not pick up the package and contact us. This will help us quickly resolve the situation.” Buttons and customer responses at this stage are not required. 10. The last message sent to the customer automatically includes the company's contact details. Example: “If you have any questions regarding the order, you can contact us: Phone: … Viber: … Telegram: …” Contact details and message text should be editable by the application administrator. 11. Non-standard cases are forwarded to the manager. For example: — the customer requests to change the order; — the customer chooses to contact the manager; — the AI could not unambiguously recognize the products or quantities; — the customer did not confirm the recognized list; — the customer sent an unsupported attachment; — a technical error occurred. After forwarding the order to the manager, the automatic scenario stops. Integrations: — SalesDrive API; — TurboSMS HTTP API / webhook, Viber Business Messages; — AI model API for recognizing free text from the customer. It is also necessary to implement: — a web panel for the manager; — receiving and displaying SalesDrive orders; — selecting one or more orders; — selecting a scenario; — selecting one or more flyers; — managing flyers and their associated products; — storing the state of the Viber dialogue; — dialogue history; — changing order statuses; — stopping the automatic scenario; — basic statistics. Incoming images, files, and voice messages from the customer do not need to be recognized. The AI works only with text. TurboSMS has confirmed the possibility of automatic message sending via HTTP API, receiving responses via webhook, and continuing the scenario within the Viber chat session. A detailed technical specification has been prepared and will be provided to the potential executor. In the response, please indicate: 1. The estimated cost of developing the MVP. 2. The estimated development time. 3. Whether there is experience with CRM/API/webhook integrations. 4. Whether there is experience integrating AI/LLM API for processing user text. 5. Examples of similar projects, if available.

A small project needs to be implemented for selling digital and physical products. What needs to be done: Develop a simple modern landing page for selling 3 products: PDF version; physical kit; paid Telegram subscription. Connect payment through WayForPay. Integrate orders with KeyCRM: transfer of customer data; creation of orders in KeyCRM; possibility of further automation. Develop a Telegram bot for managing subscriptions: automatic access issuance after payment; subscription activity check; subscription renewal; automatic access removal after payment completion; user administration through a simple panel or bot commands. It would be a plus: experience with WayForPay; experience integrating with KeyCRM; experience creating Telegram bots and paid subscriptions. Please specify immediately: exact cost; implementation timeline; examples of similar projects.

We need a chatbot with AI in Telegram for a webcam studio. The only task of the bot is to find suitable information in the studio's knowledge base and pass it to the models. It should not give its own advice or recommendations. If there is no suitable information in the knowledge base, it should direct the model to its curator. We have already assembled a bot on the suvvy.ai platform, and it generally works correctly. However, since we dealt with everything ourselves without specialists in this area, there are questions. We are currently looking for a specialist who could provide consultation/audit on this topic. The questions are approximately as follows: 1. We need a specialist to look at our settings and tell us if the cost of the bot's work is normal for us, or if it can be optimized (on this or another platform) to make the response prices lower? 2. How should we properly format the knowledge base for the bot? On Savvy, there is an option to add entries in the "question-answer" format or upload entire documents. It is important for us to clearly understand what information to upload in what format to the knowledge base so that the bot works as efficiently as possible, while also maintaining optimal response prices. There are also some minor clarifying questions about various functions, but these 2 are the most fundamental and essential) We are eagerly awaiting proposals regarding the cost and format of the consultation