Create an AI agent for inventory management

Budget: 1000 UAH Deadline: 1 day

Hello!
We are the Business Atlas team.
The essence of your project is not just in automation; the entire engineering and value lie in three critical nodes that you have laid out very precisely. The first is automatic parsing of QlikView reports every 30 minutes without expensive code. The second is the independent calculation of shortages and the logic of movements/production under the individual conditions of 40+ suppliers from China and Ukraine. The third is complete process management through a Telegram chat and automatic task creation for performers in CRM via API.
These elements form the architecture of smart procurement, and this is exactly our area of expertise. We build such industrial AI applications based on n8n/Make, deploying them on your VPS, which completely eliminates the human factor, protects against shortages, and prevents money from being frozen in excess goods.
We propose to move step by step. First, we will create the initial end-to-end slice (pilot): fully training the AI agent for 2-3 suppliers, connecting QlikView, setting up the order logic, and integrating with Telegram/CRM. The timeline for the pilot is 3-4 weeks of work, and we will fix the budget immediately after a brief discussion of the details.
Message me privately — we will discuss the structure of your reports and agree on a time for a call to detail the roadmap for the future AI purchaser!

Budget: 27000 UAH Deadline: 27 days

Hello! The project is right at the intersection of my two strengths — AI agents on Claude and automation (browser, bots, integrations).

What I see: the agent pulls data from your QlikView report and, according to the rules of each supplier, generates orders and transfers between warehouses, assigns tasks (the same cuts by lengths), and all of this is managed from Telegram by several employees, while tasks are sent to your CRM via API.

From experience, it fits:

— I built multi-agent pipelines on Claude (generation → verification → decision, with rules and validation) — so “an agent trained under conditions” is my territory, not just prompts;

— I worked with several suppliers in one engine (my own rules/compilers for each) — exactly your case “for each supplier separately”;

— browser automation (Playwright/Selenium) + Excel parsing + Telegram bots with multiple users + API integrations — I’ve done it all.

I won’t push for a specific platform just for the sake of it — I’ll suggest based on the task (Claude Agent SDK / Cowork or any other custom on VPS), depending on how it will be easier to connect to QlikView and your CRM.

A pilot with 2–3 suppliers is the right direction; we will test all the logic on it.

For all questions: choice of environment, data collection from QlikView, CRM and API, logging, automatic and semi-automatic actions and their layer of rules — I suggest we discuss in correspondence.

I’m ready to start right today! Thank you for your attention! Expand

Budget: 6000 UAH Deadline: 14 days

Good day! The first thing I will mention is the platform, as everything else depends on it.

We use OpenClaw daily for our own tasks: the agent runs on a VPS, is controlled via Telegram, fetches data in the browser, and writes to external APIs on a schedule. Your stack is not new to us.

I see a pilot with 2-3 suppliers:
1. An agent on your VPS. A separate set of rules for each supplier (processing times, logistics), rather than one instruction for all.
2. Data from QlikView. I would start with exporting to Excel, as it is more stable than clicking the report in the browser each time.
3. Order formation and movement between warehouses. The agent prepares the solution, a person confirms it.
4. Telegram: several employees, each with their own rights. The task goes to a specific executor.
5. Tasks in your CRM via API.

The question that affects the architecture: is there access to QlikView via API or scheduled export, or does the agent need to open the report in the browser under your user? If there is an export, the pilot will be more reliable and cheaper to maintain. And what CRM do you have?

The pilot duration is 14 days, the price is 6000 UAH. I am starting with this: gathering initial feedback on Freelancehunt, although my experience is not entry-level. I propose the rest of the suppliers and the sales analysis agent as a separate stage, so you pay for what is done, not for what is promised.

There are live bots in operation, ready to share in the chat so you can test the flow yourself.

Petro Pankov, BotCraft Group Expand

Budget: 27000 UAH Deadline: 35 days

Pilot evaluation - 220,000 UAH and 35 days. This is not the final price for the entire system, but a stage with architecture, VPS, integration with QlikView via browser, Telegram management for several employees, task creation in CRM via API, and training on 2-3 suppliers. Overall, it's fine =)

Look, there's a nuance - it's better not to touch QlikView as the sole source of actions without a log and a layer of checks. I would conduct the pilot through a separate layer of rules and an agent, where the algorithm suggests an order, movement, or task, and the employee confirms critical actions before collecting statistics.

You will need test access to QlikView, 2-3 typical suppliers with purchasing rules, examples of old orders, descriptions of warehouses, and methods of movements. After that, we can keep it simple - we build the pilot, measure accuracy, and then expand to other suppliers and the sales analysis agent.

> Please clarify if QlikView has direct access to the database or only Excel export from the browser
> What actions can the agent perform independently, and which require employee confirmation

> https://business.ingello.com/fractal - close in agent architecture and solution automation
> https://business.ingello.com/forma-wms - close in warehouse logic, tasks, and roles
> https://systems-fl.ingello.com - main overview of Ingello Systems for such projects Expand

Budget: 15000 UAH Deadline: 14 days

Hello, Vitaly! The task is clear. Regarding the architecture — I do not recommend building pure browser automation on top of QlikView (it breaks with any layout changes). Better: a Python core for calculating balances + LLM specifically for adapting orders to supplier conditions + confirmation of critical decisions by the manager in Telegram. Experience: production Telegram bots with a role model (aiogram 3.x), integrations with Claude API, data parsing, deployment on VPS 24/7. Please clarify: does QlikView provide data via API/schedule or only manual export? Is there documentation for the CRM API? Are the supplier conditions already in the table? I suggest starting with a pilot on 2-3 suppliers.

Budget: 27000 UAH Deadline: 20 days

Let's do it. I have a lot of experience, feel free to reach out, I will be happy to collaborate.

Budget: 15000 UAH Deadline: 10 days

Hello! I created an AI agent that pulls data from exports (Excel/BI export) and distributes purchases and transfers between warehouses according to individual rules. I have experience with QlikView exports, so we can start directly with your ready-made exports every 30 minutes. The main question is: are the conditions for 40+ suppliers already described in a table somewhere, or do we need to gather them with you along the way? This will affect the training time of the agent. The first working prototype is estimated to be ready in about a week.

Budget: 3500 UAH Deadline: 14 days

Updating data every 30 minutes from QlikView means that the agent should work with a current snapshot rather than a real-time stream, simplifying the architecture. Deploy an AI agent based on the Claude API with an n8n workflow for orchestration: n8n will poll QlikView through browser parsing or export, pass the data to the agent, process its decisions regarding orders and mixes, and then send tasks to the CRM API and Telegram through separate nodes. A pilot with 2–3 suppliers will allow for the setup of prompt templates for each of their conditions, after which scaling to the rest will be a matter of copying the logic. Free support for 2 weeks after delivery—I will assist with configuring the agent for real data and clarifications in the logic. Ready to start.

Budget: 2000 UAH Deadline: 7 days

Hello, I worked on the Warehouse Accounting Automation System — we integrated an AI agent for inventory management between 3 warehouses, reduced order processing time by 40%, and connected a Telegram bot for 8 employees.

I’m curious to know: do you plan to transfer data from QlikView to the agent via manual Excel export, or is there a possibility to set up automatic API access to the report every 30 minutes?

I suggest we get in touch; I will provide you with free technical consultation and we can create a development plan + I will tell you about my team!

Budget: 999 UAH Deadline: 1 day

Hello! The task is right up my alley: AI agent + procurement/movement automation, reading analytics from the browser, management via Telegram and CRM through API. I work on Claude Code (the same "Cowork or analog"), CRM automation I did on Monday+Make (board 683 items, action history, migrations).

I see the pilot like this:
• Connection to the QlikView report via browser (Playwright): parsing residues/strings, exporting to a structured format
• Agent on Claude API: orders under individual conditions from 2–3 suppliers, movement between warehouses, tasks by product groups (cutting / insufficient lengths)
• Telegram: managing several employees, personal tasks; creating tasks in your CRM via API
• Deployment on your VPS, logs, instructions

I suggest starting with a pilot on 2–3 suppliers — I will quickly show a working result.
I will evaluate after a brief technical specification regarding supplier conditions and your CRM API. I will send my portfolio.
— Mykola Expand

Budget: 20000 UAH Deadline: 10 days

Good day. I can take the first stage as a practical pilot, rather than promising immediate autonomous procurement management. I have experience with AI products, browser automation, Telegram bots, API integrations, and monitoring systems. I would start with an audit of access to QlikView data, rules from 2–3 suppliers, and the approval model for actions. Next, I would implement an agent that generates proposals for orders and transfers, creates tasks via the CRM API, and sends them to Telegram. Critical actions, including the actual order to the supplier, should initially require human confirmation. After the pilot, the stack, scaling volume, and risks will be clearer.

Budget: 6000 UAH Deadline: 20 days

I will set up an agent on OpenClaw on your VPS: data from QlikView snapshots (export every 30 minutes as the main channel, browser access as a backup), management through Telegram by several employees with different rights, creating tasks in your CRM. Starting with 2-3 suppliers, as you plan.

Regarding the mechanics: the agent retrieves stock balances and movements from the export, sees shortages by items and deadlines of specific suppliers, and compiles a draft order or transfer from this. The conditions of each supplier live in a separate rules file: processing times, logistics, minimum batches. To add a new supplier after the pilot, it is enough to create another such file, without reworking the system.

The mathematics of balances and splitting by lengths is calculated by regular Python code; the agent makes decisions based on exact figures. This way, it does not make mistakes in arithmetic and spends a minimum of tokens, which makes a noticeable difference in the monthly API bill for a system that operates around the clock every 30 minutes.

The agent prepares orders and transfers as drafts, confirmed by an employee in Telegram, with each action logged.

I will be honest: such automation is not assembled overnight. For the agent to order like an experienced purchaser, it is necessary to analyze real orders for each supplier, embed the conditions in the rules, and run the agent on live data, fine-tuning the decisions. This is work, and I am ready to do it, rather than sell a magic button.

I work daily with OpenClaw and Claude Code in my own projects, with parallel agents and browser automation; I know the platform as a user, not from documentation.

We will discuss the details in the chat. Expand

Budget: 15000 UAH Deadline: 10 days

I understood the task: an AI agent for automating procurement and movements between warehouses, trained under the conditions of each supplier, managed through a Telegram chat by several employees, and creating tasks in CRM via API. Plus, connecting to a QlikView report (browser-based BI analytics) for up-to-date procurement data.

I work closely with parallel Claude Code/agent platforms — I lead a project where several agents simultaneously solve different tasks on one server, and I know the nuances of deployment on VPS and configuration for specific training scenarios.

Regarding architecture: I will deploy the agent on VPS, connecting to QlikView via scheduled Excel file exports (synchronizing with the report every 30 minutes) or a headless browser if there is no export via API or direct link. Training for 2-3 suppliers in the pilot — rules for processing times and logistics for each supplier as separate configs, not hardcoded logic, to allow for further expansion without rewriting.

Telegram part: the bot receives commands from employees, the agent forms orders and movement tasks and creates them in CRM via API, notifying a specific employee in the chat.

For the pilot with 2-3 suppliers, realistically 1-2 weeks, and then expansion to other suppliers goes faster because the architecture is already fine-tuned.

Budget: 5000 UAH Deadline: 30 days

Hello!

I have reviewed your task regarding the implementation of an AI agent for procurement and logistics. Your task is a classic example of how AI can save hundreds of hours of routine work, and I am ready to help you realize this.

My approach to the project:

Integration: I have experience working with the automation of complex BI systems. The priority will be to create a reliable data transmission channel with QlikView so that the agent operates stably.

Agent intelligence: We will use an architecture that allows adding new suppliers as separate modules (prompts/context), ensuring easy scalability to all 40+ suppliers in the future.

Control: Working through a Telegram bot with order confirmations from managers guarantees security and eliminates errors.

Estimation:
According to my calculations, the development of the MVP will cost $5,000.

What we will deliver:

A fully operational agent on your VPS.

Automation of procurement and internal transfers.

A ready infrastructure for further sales analysis, as you planned.

Are you ready to discuss the technical details of connecting to QlikView? This is a key stage that will determine the speed of the project's launch. Expand

Budget: 16000 UAH Deadline: 17 days

Good day, Vitaly!
I looked at your projects - integration with Bitrix, warehouse of rolled products. The specifics - the logic of cutting and task setting is clear.
My background with agents: ~2 years testing AI models at Scale AI (Outlier) and Meta AI - I know/predict agent failure points before they make mistakes in procurement.

Basic pilot (MVP) - what is stated:
Instead of OpenClaw - compact options, much cheaper and more accurate, with ready settings (skills, rules).
Agent on your VPS - (deployment with Docker or without, depending on what is already on the host).
Data from QlikView: snapshot every 30 minutes via export (browser option - backup).
Training under the conditions of 2-3 suppliers. Order formation, movements between warehouses.
Tasks by product groups (cutting by lengths), mathematics in Python core, LLM only adapts to the rules - zero hallucinations.
Telegram management - several employees, roles and rights, task to a specific performer.
Creating tasks in CRM via API.

Basic pilot, as MVP - 16k UAH / up to 17 days (depends on the depth and scope of testing), payment by stages (escrow):
1. 5k - data collection from QlikView + calculation core;
2. 6k - agents, supplier rules, Telegram with roles;
3. 5k - integration with CRM, log, acceptance on live data.

Beyond the request - bonuses and improvements (everything is available and works), we can discuss in a personal chat if you see feasibility:
- production-level security: protection against prompt injections, OAuth2, SSH keys, audit log of all agent actions.
- memory layer for agents during communication (corporate license - can be a gift).
- agent-sysadmin, pipeline control 24/7.
- MCP bridge in TG: a full-fledged agent (with all capabilities and context) in chat, not a model bot (for management/admin).
Scaling to 40+ suppliers + sales analysis agent with hybrid RAG (semantics + text query + document database) - the core of all business knowledge, as a separate stage.
Questions: Does QlikView provide export on schedule or only through the browser? Are supplier rules formalized? Expand

Budget: 1000 UAH Deadline: 1 day

Hello! The task is interesting: an AI agent that independently creates orders based on the conditions of each supplier and manages movements between warehouses — just what we love to assemble. I will handle management through a Telegram chat for several employees and task setting in the CRM via API, starting with a pilot for 2-3 suppliers and connecting to data from QlikView. I'm ready to discuss the details in private messages.

Budget: 27000 UAH Deadline: 14 days

Good day. I am ready to perform. I can start immediately after clarifying the details. 📞 Always available and open to communication.

Budget: 27000 UAH Deadline: 12 days

Good day!

A Telegram agent with integration into CRM via API and logic tailored to individual supplier conditions is the core task. Management from the chat by several employees, tasks assigned to specific individuals, creation of tasks in CRM.

Implementation: agent on Claude API on your VPS, data from QlikView via export, pilot with 2–3 suppliers (orders, transfers between warehouses, tasks by product groups), then scaling. For 40 suppliers, a custom build is more reliable than OpenClaw/Cowork - full control over the rules.

Minimum pilot - 30,000 UAH, 12 days: 2 suppliers, Telegram management, tasks in CRM. Check the logic live.
Full pilot - 45,000 UAH, 18 days: 3 suppliers + transfers between warehouses + tasks by product groups (cutting/lengths).

Scaling for others: 3,000 UAH/supplier.

Portfolio: in the profile and on karasov.co Expand

Budget: 15000 UAH Deadline: 8 days

Good day, Vitaliy.

Your warehouse management will receive an AI agent that will independently generate orders to suppliers, taking into account their individual conditions, and manage internal movements of goods to maintain optimal stock levels. The agent will learn from historical data from your QlikView report, which it will read through a headless browser, emulating user actions — this will avoid complex integrations with the closed BI system. In Telegram, you will have a single interface for the team: employees will be able to assign tasks to the agent, and it will send notifications about generated orders and create tasks in your CRM via API. After a pilot with 2-3 suppliers, the system will scale to the entire network without changing the architecture. For reliability and speed, the agent will operate on Go using Docker containers on your VPS.

Pilot:
10,000–15,000 UAH; ≈8 days
Creation of a basic prototype of the agent to validate the concept on a limited data set.
- Integration with QlikView to read data from one supplier
- Basic Telegram bot for receiving notifications
- Automatic order generation for 1-2 key items
- Deployment on VPS in a Docker container

Full launch:
25,000–35,000 UAH; ≈22 days
Fully functional AI agent for managing purchases and movements with integration into business processes.
- Full training of the agent based on the conditions of 2-3 pilot suppliers
- Automation of internal movements between warehouses
- Two-way communication in Telegram for multiple employees
- Automatic task creation in your CRM via API
- Module for generating specific tasks (e.g., material cutting)

Scalable:
45,000–60,000 UAH; ≈35 days
Expanded system supporting all suppliers and preparing for sales analysis.
- Scaling logic to all 40+ suppliers
- Enhanced control and logging system for all actions of the AI agent
- Optimization of data reading speed from QlikView
- Architectural preparation for the implementation of the sales analysis module
- Priority technical support and maintenance during the first month Expand

Budget: 15000 UAH Deadline: 14 days

Good day! I specialize in building AI agents based on Claude (Agent SDK / Cowork) - the very tools you are considering. I propose a pilot: I will deploy the agent on a VPS, set up data collection from QlikView (Excel export every 30 minutes), and train the agent on the individual conditions of 2-3 suppliers (processing times, logistics, minimum batches). I will add management through Telegram for several employees, create tasks in CRM via API, and auto-assign tasks to performers. The agent will generate orders for suppliers, suggest movements between warehouses, and set tasks by product groups (for example, cutting by lengths). After the pilot, we will scale to the remaining 40+ suppliers and an agent for sales analysis.

Budget: 18000 UAH Deadline: 14 days

Hello! I have extensive experience working with Python, databases, API integrations, and developing Telegram systems for businesses. I have significant experience in implementing similar solutions for process automation and analytics for enterprises.

The direct method you describe (a pure standalone AI agent clicking in the browser via Claude/OpenClaw) is impractical and critically dangerous for production procurement:

- Any change in layout or delay in QlikView will break UI automation.
- AI hallucinations in procurement/logistics pose a risk of miscalculating figures and freezing millions in incorrect goods from China.
- AI simply cannot calculate mathematical problems (length cutting optimization) accurately.

Instead, I propose implementing a hybrid approach:

- A reliable core (Python + DB): automatically (without a browser) retrieves Excel reports from QlikView and performs 100% accurate mathematical calculations of balances and cutting.
- AI (LLM): used solely as a communicator — adapts orders to the specific text conditions of each of the 40+ suppliers.
- Control (Human-in-the-Loop): AI does not send anything by itself. It generates a solution card in Telegram, the manager checks it with one click, and the system automatically creates a task in the CRM via API.

Since procurement and logistics have many internal nuances (for example, the API format of your CRM and specifics of working with China), this project requires individual calculation. I suggest moving to private messages to discuss these technical details further. After that, I can offer you several options for implementing a pilot with flexible timelines and budget that will fully fit within market frameworks. Please write. Expand

Budget: 25000 UAH Deadline: 20 days

We have experience in developing AI agents for warehouse logistics automation. We implement the solution through integration with your accounting systems for supply chain analysis and order processing time forecasting. This will allow for inventory optimization and minimize delays when working with suppliers. We are ready to discuss the architecture details.

Budget: 12000 UAH Deadline: 5 days

Ready to deploy an AI agent for inventory management in Telegram — it will read data from QlikView, respond to queries about stock levels, and generate reports on command.

The key here is the correct linkage QlikView → intermediate script on VPS → Telegram bot. QlikView delivers data in batches (snapshot every 30 minutes), so the agent will work with the current snapshot rather than a real-time stream — this simplifies the architecture and makes the bot stable even with large volumes.

- Data collection script from QlikView (export to JSON/SQLite on VPS)
- Telegram bot with commands: stock levels by product, movement over a period, low stock
- Connection of the AI agent for understanding text queries (“what is left in warehouse #2?”)
- Deployment on VPS, systemd service for uninterrupted operation

I created a Telegram bot for inventory management in a small warehouse — the bot responded to stock queries and sent alerts at critical minimum levels. It has been working reliably ever since.

I am currently completely free — I will take your project immediately and start today. Please provide the data format that QlikView currently delivers (API, file export, direct connection) — and I will draft a clear plan for the first stage. Expand

Budget: 27000 UAH Deadline: 60 days

Hello. I have worked with documentation and warehouses at a machine-building enterprise and for a printing house; the cases are described in my profile.

Regarding your task:
- Does QlikView provide data only through the browser/export to Excel, or is there access to the database/scheduled exports? This significantly affects reliability.
- What CRM do you have, and is there documentation for its API?
- Are the ordering rules with suppliers formalized somewhere (timelines, minimum batches, logistics), or do they need to be gathered from scratch?

Budget: 1000 UAH Deadline: 1 day

Good day, I am ready to complete your task quickly and efficiently. I have extensive experience in creating various bots. Please write to me in private messages to discuss the details. I will be happy to help :)