The Muffin Company

TECHNICAL TASKImplementation and Integration of BAS Accounting + BAS Trade Management with Websites and Services1. General Information It is necessary to combine and configure the joint operation of the BAS Accounting and BAS Trade Management systems, as well as connect 3 websites to the BAS with correct data synchronization.2. Project Tasks2.1. BAS Integration Combine BAS Accounting with BAS Trade Management Set up correct data exchange between the systems Ensure a unified logic for accounting sales, inventory, and finances2.2. Website Connection Connect 3 websites to the BAS: Website on WordPress Website on OpenCart Website on the Khoroshop platform Integration functionality: transfer orders from websites to the BAS synchronize order statuses account for payments synchronize customers (optional) synchronize products and prices2.3. Inventory Accounting Currently, there is 1 general warehouse in the BAS It is necessary to divide it into 3 separate warehouses Set up: correct accounting of balances linking orders from websites to the corresponding warehouses write-off of goods from the warehouse during sales2.4. Service Connection Checkbox: fiscalization of sales transfer of receipts correct operation with the BAS Nova Poshta (NP): creation of TTH from the BAS transfer of order data synchronize delivery statuses

It is necessary to set up a connection for vapi.ai through make.com to the site on prom.ua. The bot should inquire about the price, availability, and description of the product on the site.

Automation / AI Content Factory for Creating Short Videos Looking for a specialist to develop an automated system for mass creation of AI content for TikTok / Instagram Reels. The main task is to create not a template that generates similar videos each time, but a full-fledged content factory capable of regularly producing different videos: with new plots, scenes, locations, actions, and visual solutions. ### How the process should roughly work: 1. Idea and Script Generation - the system generates a new theme/idea for the video; - creates a script; - breaks it down into separate scenes/frames; - for each scene, a description of what should happen is formed. It is important that the videos differ from each other each time, and the system does not constantly repeat the same ideas, scenes, and structure. 2. Image Creation for Scenes - for each scene, an image is automatically generated; - the images must correspond to the overall story of the video; - there should be the ability to work in two modes: Mode 1 — Different Characters and Different Stories Each video can have new heroes, a new visual style, new locations, and a plot. Mode 2 — AI Blogger / Constant Character One character is fixed, who maintains a recognizable appearance between videos, but at the same time changes: - clothing; - locations; - situations; - actions; - plot; - environment. So, there should be a conditional AI blogger who constantly appears in new content. 3. Bringing Scenes to Life + Voice Preferably use AI models that can immediately: - bring the image to life; - create speech / lip-sync; - generate or synchronize voice. That is, it is not necessary to do voiceover as a separate stage if the chosen AI service can generate video with speech immediately. But the architecture should consider the cost of generation: if separate voice generation + subsequent animation turns out to be significantly cheaper or of higher quality, that option can also be used. Language of speech: - Russian; - Ukrainian. 4. Video Assembly The output should be a set of ready scenes or a full vertical video. Additionally, the following can be implemented: - automatic scene connection; - subtitles; - music; - sound effects; - transitions; - automatic editing. Editing is not a mandatory part of the first stage of the project — it depends on the cost and complexity of implementation. 5. Auto Posting Optionally: - automatic publication on TikTok; - automatic publication on Instagram Reels. Auto posting can also be separated into a distinct stage. ### Important Principle of the System It must be a scalable AI content factory, not a rigidly fixed workflow for generating identical videos. The system should be able to create a large amount of different content through: - generating new ideas; - different scripts; - different scenes; - different locations; - different character actions; - different visual compositions; - controlling repetitions of already used ideas. At the same time, there should be the ability to fix certain parameters, such as: - one constant AI character; - specific visual style; - channel theme; - video format; - language; - video duration. ### Example Logic: Theme / Niche → New Idea → Script → Scene Division → Image Generation → Scene Animation with Speech → Video Assembly → Saving / Publication ### Regarding Implementation Preferably use n8n, but open to other solutions if the performer can offer a more stable architecture. The system should be modular so that in the future it can replace: - LLM; - image generation model; - image-to-video / talking image model; - voice service; - editing service; - publication method. It is desirable to provide for the storage of the history of already created content so that the system can take into account previous videos and reduce the number of repeating ideas. ### What I want to receive from the performer in the response: - your experience with AI automation and n8n; - examples of similar projects; - what architecture you propose; - which AI models / APIs you suggest using; - how you propose to solve the issue of content diversity; - how you propose to ensure the consistency of the constant AI character; - approximate cost of one generated video; - cost of automation development; - what can be implemented within the first MVP. I am ready to consider phased implementation: Stage 1: working content factory until the generation of ready scenes / video. Stage 2: automatic editing. Stage 3: automatic publication on TikTok / Instagram.

There is a Corp. license. It is necessary to configure the control center for 1C and SQL to understand where the performance drops and where the hangs occur.

General description: We are looking for a performer to solve the task of automating the update of data on supplier stock balances. First of all, we consider the search and adaptation of an existing ready-made solution. If a ready-made solution that meets the key requirements is not found, the development of an extension from scratch will be considered.The main goal is to set up stable and automatic loading of balances. An additional (optional) task is to set up the loading of price lists. Consideration may be given separately or in combination.1C Configuration:Program: Business Automation Software for small company. PROFVersion: 1.6.21.3Requirement: The solution must be implemented exclusively as an extension, without making changes to the main configuration.Approach to implementation (Important note for performers):We are open to your proposals! If you have:A ready-made extension that partially meets the requirements — offer it with a description of the necessary modifications.Your vision or alternative approach to implementation — we will gladly consider it. Our goal is to obtain an effective working tool, and we are ready to be flexible in ways to achieve it. Main functional requirements:1. Data retrieval:Source: Gmail mailbox.Method of retrieval: The system must be able to process: Emails with direct attachments (files).Emails that contain a direct link to download a file (without the need to enter logins/passwords on third-party portals).Attachment filtering: The ability to specify a "mask" for the file name or extension for each supplier (for example, *stock*.xlsx), so that the system automatically ignores other attachments in the email (invoices, images with signatures, PDF catalogs).File formats: Excel (.xls, .xlsx), CSV, XML. 2. Data configuration and processing:It is necessary to implement a flexible configuration interface for each supplier separately.Item matching logic: Matching of goods must occur strictly by article number.Processing of unmatched items: If the article from the file is not found in the 1C database, the system should not create a new item. Instead, a report should be generated listing such items for further analysis by the manager.Processing of incomplete files: The system must correctly process files that contain only balances (without prices) or only prices (without balances), updating in 1C only the data that is present in the file.Memory optimization: Parsing large Excel files should be maximally optimized (for example, reading through a spreadsheet document on the server side or using specialized libraries) to avoid crashing the 32-bit client application due to lack of RAM.Units of measurement and currency (optional): The ability to specify the currency of the input file (with auto-conversion) and set the packaging coefficient for correct calculation of basic units of measurement. 3. Data loading logic:Main task: Loading balances This is a key and mandatory part of the functionality.Critical requirement for clearing: Clearing previous balances of a specific supplier must occur only after successful reading and full validation of the new file. If the file is corrupted, empty, or the column structure has changed — old balances are not deleted, and the system simply generates an error. This is necessary to avoid accidental zeroing of the database and to guarantee data availability. Additional (optional) task: Loading pricesThis functionality will be an advantage but is not mandatory at the first stage.Static prices: The ability to load prices into three defined price types in 1C: "Purchase", "VIP price", "Retail".Calculated prices: It is necessary to provide a flexible mechanism for calculating prices according to formulas, which is adjustable for each supplier. For example: Option A: The retail price from the file is the base, while the purchase and VIP prices are calculated as a percentage of it.Option B: The purchase price is the base, while the retail and VIP prices are calculated using a markup.There must be the ability to set the base for calculation and the formula for each price type. 4. Storage and display of data in 1C:Storage: It is recommended to store data in a new Information Register. However, we are open to your proposals if you see a more optimal architectural approach.Display: Current information about balances should be displayed in the following interfaces:Document form "Customer Order".Form "Product Selection".Item card form.Important: If the same article is available from several suppliers, the interface must clearly show the balance for each of them separately (for example, a separate information panel, a pop-up window on hover, or additional columns). 5. Automation:Loading must be performed automatically using a server scheduled task (or another more optimal approach).There must be the ability to set different start times for different suppliers (for example, 8:00, 10:00).There must be the ability to manually start the loading process.Notification system: In case of critical errors during automatic loading (absence of file in the email, unknown format, reading error), the system must generate notifications for the responsible manager (in the form of a task in 1C, message, or email). Work result:Extension file (.cfe), or installation package of the ready-made solution.User manual for configuration and use. P.S. Examples of similar modifications: 1) https://ingenum.ua/analiz-praysov-postavshikov-dlya-bas-ili-1c2) https://pdk.com.ua/ua/features/dev/monitor_zagruzki-detail/3)https://nct.ua/catalog/raboty_dlya_1s_8/Automatic-loading-of-price-lists-of-suppliers-and-the-balance-of-goods-from-suppliers/