AI for business - implementation of an AI agent for the warehouse (online store)

Budget: 18750 UAH Deadline: 7 days

I will create a pipeline: I will extract frames from the warehouse video, recognize shelves and labels on the shelves using Vision AI, match them with the SKU list, and generate an Excel file with addresses. Stack: Python, OpenCV for frames, GPT-4o Vision for recognition, openpyxl for export. Question: will there be one video or several by zones?

Budget: 27000 UAH Deadline: 14 days

Hello! I have production experience with AI processing of documents and images — I developed a pipeline for recognizing PDFs/scans with OCR + Vision AI + classification through LLM. The stack is fully relevant: Python, Claude/GPT Vision API, OCR, automatic generation of structured data.

Regarding the task of recognizing inventory through video — I want to be upfront about realistic expectations. Full automation (video → Excel without human involvement) for 3800 SKUs will have an accuracy of approximately 60-80% depending on the quality of the video and the readability of the labels.

I recommend a semi-automatic approach: AI recognizes and fills in 70-80% of the data, while the rest is verified by an operator through a simple web interface. This will provide 95%+ accuracy for the final result.

Architecture:
1. Video → frame breakdown by shelves
2. Vision AI recognizes the structure (zone → shelf → rack) and reads product labels
3. Matching SKUs with your product database
4. Generation of Excel + web interface for quick verification

Question: in what format is your product database currently (Excel, CRM, website)? This is needed for automatic matching of SKUs with the recognized text.

Estimated cost: $1000-1500. Deadline: 2 weeks. Expand

Budget: 25000 UAH Deadline: 20 days

Good day.
20 years of development experience.
Is all the product with barcodes or QR codes?
There are difficulties with the video in practice: the name or barcode is not visible, the AI hallucinated and misrecognized, etc.
----> I propose a more reliable option. 100% result.
I will develop a system (WEB + application) for a scanner linked to "room-row-shelf-cell."
There will be a mobile application, and one person will scan all 3800 items in one or two days.
It will be maximally reliable. No AI hallucinations.
There will be a product map and backward compatibility - search by article in the warehouse, in a convenient WEB interface, you enter the article/product code, and the scheme will show where the product is, on which shelf.
For example, if you need to collect an order - 20 products, a person goes with this application on their phone, a list of products and "addresses," and clearly picks up the product from each cell.
This will be guaranteed, reliable, and save time/money on routine processes. Expand

Budget: 3000 UAH Deadline: 2 days

This is an extremely complex and ambitious project that combines two completely different technological spheres: Conversational AI (working with voice and text) and Computer Vision / Video Processing (video analysis of the warehouse).

To ensure the client chooses you, you need to demonstrate that you are not just going to "connect ChatGPT," but that you understand the architecture of each of these systems and know how to integrate them into a rather specific CRM (OneBox).

Here is a professional version of the description for the bid:

Description option for Freelancehunt
Topic: Implementation of a comprehensive AI agent (Voice + Text) and AI warehouse analysis system for OneBox CRM

Hello! Your technical specification is an example of a deep understanding of how AI should work in real business. The project consists of two powerful blocks: conversational commerce and computer vision. I have experience building such AI ecosystems and am ready to implement this architecture for your store.

How I propose to implement the technical part:

BLOCK 1: AI Agent (Sales and Communication)

Voice AI (Calls): For maximum naturalness of voice (UA/RU) and absence of delays, I will use specialized voice engines (for example, Vapi.ai, Bland.ai, or OpenAI Realtime API). They are capable of interrupting, listening, and speaking with intonation.

Integration with OneBox: I will set up two-way synchronization via API. The agent will check the order status, customer history (for Upsell/Cross-sell), and automatically create cards/tasks in the CRM.

Upselling and QA System: I will create a separate scenario based on LLM for basket analysis (if there is a form — we offer chocolate and dye). I will also set up a transcription system for conversations of live managers: AI will analyze the text and assign a score for adherence to the script.

Text Messaging: I will automate the generation of personalized email/SMS through logic modules (Make/n8n), which will segment the database according to RFM analysis (recency, frequency, monetary).

BLOCK 2: AI Warehouse Recognition (Computer Vision)

Video Analysis: I will use models with a large context window (for example, Gemini 1.5 Pro), which are capable of frame-by-frame analysis of video files, "reading" labels on shelves, and recognizing the geometry of shelves.

Table Generation: AI will match recognized products with the format [ZONE] - [SHELF] - [RACK] and automatically generate an Excel file for import, according to your KPI.

Work format and assessment:
Since the system is complex, I propose to break it down into sprints:

Sprint 1: Text communications, integration with OneBox, Upsell/messaging logic.

Sprint 2: Voice bot (Inbound/Outbound) and IP telephony.

Sprint 3: Warehouse video analysis module and Excel generation. Expand

Budget: 25000 UAH Deadline: 10 days

Ready to take on the work, write to me, I will do everything with quality. Experience of more than 14 years!

Budget: 19000 UAH Deadline: 18 days

Good day!

An interesting project — these are exactly the kinds of tasks I enjoy, where there is a specific result in the form of a file for import.

Here’s what I propose for each block:

📦 Warehouse numbering — I will develop the addressing logic [ZONE]–[SHELF]–[RACK] for your area of 300 m², the number of zones and racks. You will receive a ready-made marking scheme.

🧠 AI recognition through video — I will implement it via Vision API (GPT-4o): the script receives video, extracts frames, recognizes the numbers of racks/shelves and product names, and automatically builds a warehouse map. If there are barcodes — I will connect their scanning for more accurate matching with 3,800 SKUs.

📊 Excel format for import — separate columns: Product / SKU / Zone / Rack / Shelf / Full address. I will adjust the format to fit your accounting system (1C, Khoroshop, custom WMS).

I have experience with Python automation and AI tools — a similar agent for automating workflows is already in operation.

A few questions for an accurate estimate:
— Is there already physical labeling of the shelves?
— Into which system do you plan to import the result?
— Is there a ready list of SKUs in Excel format?

I am ready to start immediately. Shall we discuss the details in private? Expand

Budget: 2000 UAH Deadline: 3 days

Hello! After reviewing your project, I am ready to start working on it. I can offer optimal solutions to achieve the best result.

Budget: 27000 UAH Deadline: 14 days

Hello! I am the project manager of Business Atlas. This task is 100% our profile. We specialize in creating autonomous systems where n8n manages the AI logic for processing and structuring data, including visual data. Our workflows have already implemented similar tasks: from automatic data collection and clustering to complex integrations via API. Our turnkey technical solution: 1. AI Video Processing (Computer Vision): We will use GPT-4o Vision or Gemini 1.5 Pro models for step-by-step analysis of your video. The AI recognizes shelf markings and product names on packaging/price tags, automatically linking them to a specific location (Zone-Shelf-Shelf). 2. Structuring in n8n: Through a script in n8n (self-hosted), we will transform visual data into a clear hierarchy. The system will group products by bins on its own. 3. Generation of import file: At the output, you will receive an Excel/Google Sheets in your format (Product | SKU | Location), ready for upload into your accounting system. 4. Validation: We will set up an automatic verification stage where the AI highlights questionable positions (for example, where the text in the video is blurry) so that you achieve a KPI of 90%+ accuracy. Project indicators: • Cost of system development: $1,200 – $1,800 (depends on the quality of the video and the readability of the labels). • Term: 10–14 working days. • Result: You receive a ready Excel file for import and the system itself, which you can use for regular inventory checks. Message me privately — I will analyze a small fragment of your video and show you how the AI reads your shelves right now.

Budget: 700 UAH Deadline: 1 day

Good day, is a fully local solution (local computer vision model) needed or is it possible to use API services?

Budget: 10000 UAH Deadline: 7 days

Good day!

The task is clear — AI analysis of warehouse video, recognition of product addresses (zone-shelf-shelf) and formation of Excel.

Technically, I will implement it as follows:
— Break the video into frames (ffmpeg)
— Each frame is analyzed by Claude Vision — recognizes the labels on the shelves and links the product to the address
— Deduplication and structuring of data (Python)
— Ready Excel in the required format

With high-quality video where labels are clearly visible — accuracy 95%+.

Please clarify a few points:

1. Is the video already available or will you be filming it? Are the labels on the shelves clearly visible?
2. Are the labels textual or barcodes/QR?
3. Are the product names already in the system — do you just need to link them to addresses? Expand

Budget: 8000 UAH Deadline: 1 day

Good day. I am ready to complete this project as I have extensive experience in application development.