AI Procurement Agent: Automate Routine Purchasing

Every day, a procurement manager spends 3–4 hours on routine request processing: manually entering data from emails into 1C, searching for analogs, requesting quotes. Errors are inevitable — wrong SKU, incorrect price, missed deadlines. We automate these tasks with an AI procurement agent — a digital employee that processes requests, compares offers, checks documents, and creates orders. In practice, our AI agent reduces the procurement cycle for standard items from several days to hours. The core is a combination of GPT-4o, LangChain, and integration with 1C. The solution suits manufacturing companies, distributors, and retailers with a volume of 100+ requests per month.

What problems does the AI procurement agent solve?

The three main pains: manual request processing, long supplier search, and data entry errors. The agent automatically classifies the request, checks budget limits, and launches a tender among 3–5 suppliers from your registry. For most requests up to a set threshold, automatic purchasing is possible without human involvement. Data entry errors are reduced to virtually zero through parsing from messengers and email. According to APQC, procurement automation reduces operational costs by 20–40%.

How is security ensured during integration?

The model runs in your environment — on-premise or VPC. No data leaves the infrastructure. All connections are TLS-encrypted, with a role-based access model. For 1C, we use asynchronous REST calls through a dedicated gateway with auditing. The agent has no direct access to financial documents — only through approved scenarios.

How does the auto-approval mechanism work?

Based on set rules (amount, category, budget), the agent automatically approves up to 62% of requests. If a rule doesn't trigger, the request is sent to the manager. All decisions are logged for auditing.

How we implement: architecture and stack

We use a microservice architecture:

• AI core: GPT-4o + LangChain for natural language processing and decision-making.
• Integrations: REST API to 1C, Telegram Bot API, email via IMAP/SMTP.
• Data: PostgreSQL for purchase history, vector database (ChromaDB) for similar request search.
• CI/CD: Docker + Kubernetes on k3s, monitoring with Prometheus + Grafana.

async def handle_purchase_request(request_text, requester):
    # Parsing the request via GPT-4o structured output
    parsed = await client.chat.completions.parse(
        model="gpt-4o",
        response_format=PurchaseRequest,
        messages=[...]
    )
    # Auto-approval and tender launch logic
    ...

How much faster do purchases become? Practical data

For one of our clients — a manufacturing company with 200 requests per month — we deployed an AI agent integrated with 1C and Telegram.

The AI agent processes requests 4.6 times faster than traditional processing. The standard procurement cycle was reduced by a factor of 4.6. Managers shifted to strategic negotiations and non-standard items. Operational cost savings reach 40%.

Implementation stages of the AI procurement agent

Total: 6–8 weeks turnkey. Implementation cost ranges from $15,000 to $35,000 depending on integration complexity.

What is included in the work: deliverables turnkey

We deliver:

• Source code of the AI agent tailored to your architecture.
• Docker images for deployment in your environment.
• Operational manual and model retraining guide.
• Integration setup with 1C, Telegram, email.
• API documentation and Postman collection.
• Support for 14 days after launch.

Timeline and cost

Exact cost is calculated individually — depends on integration complexity and number of suppliers. Estimated timeline: analysis — 1 week, integration with 1C — 2–3 weeks, AI core — 2 weeks, testing — 2 weeks. Total 6–8 weeks. Want to evaluate your project? Request a consultation — we'll send a sample implementation and a detailed plan. Or contact us directly.

Why trust our experience?

We have over 5 years of experience in AI process automation and 10+ successful implementations for procurement processes. We guarantee NDA compliance and full customization for your ERP. Each project is supported by a certified ML engineer.

LLM Development: Fine-Tuning, RAG, Agents, and Production Deployment

Using GPT‑4 or Claude 3.5 Sonnet through a public API is not a solution — it's just a tool. When the requirement is to "make it like ChatGPT, but on our data," there is a real engineering challenge behind it: from prompt engineering to training a 70B model on your own infrastructure. End-to-end LLM solution development is a complex stack, and we have been doing it for over 5 years. During this time, we have completed over 20 projects in generative AI: from RAG systems for legal departments to custom support agents. Where exactly your task falls depends on data, latency requirements, budget, and how critical confidentiality is.

A typical situation: the client has already tried ChatGPT, but results are unstable — sometimes accurate, sometimes hallucinating. Or they need integration into a corporate portal while complying with security policies. Let's break down each layer of the stack in detail — from RAG to production deployment.

Why Do RAG Systems Break and How to Fix It?

RAG (Retrieval-Augmented Generation) looks simple: find relevant documents, put them in context, get an answer. In practice, it fails in several places.

Chunking without overlap. Classic mistake: chunk_size=512, overlap=0. If the answer lies across two chunks, retrieval won't find either with sufficient confidence. Solution: overlap 15–25% of chunk_size, or better yet, sentence-aware splitting with spaCy or NLTK instead of naive character splitting.

Poor embedder. text-embedding-ada-002 is good for general use, but on legal or medical texts, specialized models like E5-large-v2, BGE-M3, or fine-tuned sentence-transformers on domain data outperform it. Recall@5 differences can be 15–25%.

No re-ranking. Vector search optimizes for speed, not relevance. A cross-encoder re-ranker (ms-marco-MiniLM-L-6-v2, bge-reranker-large) after initial retrieval improves top-3 accuracy with acceptable latency (+50–150ms). This is often more impactful than improving the embedding model.

Hybrid search. Dense vectors alone work poorly on exact queries: names, SKUs, codes. BM25 (sparse) finds exact matches but misses semantics. Hybrid via RRF (Reciprocal Rank Fusion) is the optimal compromise. Qdrant, Weaviate, and pgvector 0.7+ support hybrid search natively.

When to Fine-Tune Instead of Prompt Engineering?

Prompt engineering solves ~70% of LLM adaptation tasks for a domain. The remaining 30% require fine-tuning. Three indicators: the model ignores a specific output format even with detailed prompting; the task requires deep knowledge of specialized vocabulary (medicine, law); you need to significantly reduce token costs by replacing a large model with a smaller specialized one.

LoRA and QLoRA are the standard for SFT. LoRA adds trainable low-rank matrices to attention layers. A typical configuration for Llama-3 8B: r=64, lora_alpha=128, target_modules=["q_proj","v_proj","k_proj","o_proj"] yields ~0.8% trainable parameters, training on one A100 40GB. QLoRA adds 4-bit quantization (NF4) and allows fine-tuning 70B models on two A100 40GB, though speed drops by half compared to bf16.

DPO instead of RLHF. Direct Preference Optimization requires only (chosen, rejected) pairs, not scalar reward signals. DPOTrainer from the trl library (Hugging Face) implements it in a few dozen lines.

Common mistake. A dataset of 500 examples, 5 epochs, validation loss 0.8 — seems fine. But on test, the model degrades on general instructions. Cause: catastrophic forgetting. Solution: add 10–20% general instruction-following examples (Alpaca, FLAN) to the training set to preserve original capabilities.

How to Choose a Base Model: 8B or 70B?

For most tasks, fine-tuning an 8B model is sufficient. 70B is needed when deep reasoning is required or the 8B baseline does not reach the required quality even after fine-tuning. Inference cost for Llama-3 8B via vLLM on A100 is efficient; the exact cost depends on volume.

What Does PagedAttention Bring to Production?

vLLM is the first choice for serving open-source models. PagedAttention is the key technical innovation: KV-cache is managed like virtual memory in an OS, without fragmentation. This yields 2–4x higher throughput compared to naive HuggingFace Transformers inference. The vLLM documentation confirms that continuous batching and PagedAttention are the standard for high-load LLM services.

Typical numbers on A100 80GB for Llama-3 8B (bf16): 400–600 req/s, P50 latency 200–400ms, P99 latency 600–900ms at concurrency 64. For 70B on two A100 with tensor parallelism: 80–120 req/s, P99 latency 1.5–2.5s. AWQ or GPTQ quantization reduces memory consumption by 2x with quality loss within 1–3%.

Multi-Agent Systems

Agents are LLMs with access to tools: search, code execution, API calls, database interaction. Common patterns:

• ReAct (Reason + Act): the model reasons → chooses a tool → observes the result → reasons again. LangChain and LlamaIndex implement it out of the box.
• Multi-agent orchestration: multiple specialized agents with a coordinator on top. Example: coordinator → researcher (search + summarization) → coder (code generation and execution) → critic (verification). Tools: AutoGen (Microsoft), CrewAI, custom implementation on LangGraph.

In production, agent systems are non-deterministic. Essential: guardrails, step limits, logging of each step, human-in-the-loop for critical actions.

How We Work: Stages, Timeline, Deliverables

What Is Included

We deliver:

• Technical documentation (model card, configs, deployment instructions)
• Access to infrastructure (code repository, trained weights)
• 1 month of post-deployment support (consultations, bug fixes)
• Customer team training (2–3 sessions on system operation)

Timeline: basic RAG prototype — 1–2 weeks. Fine-tuning with customer data — 3–6 weeks (including data preparation). Production system with monitoring and retraining — 2–4 months. Cost is calculated individually based on data volume, model complexity, and infrastructure requirements.

We guarantee the quality of the final model with performance benchmarks and ongoing monitoring. Our engineers have hands‑on experience with dozens of production LLM systems.

Want to evaluate your project? Leave a request — we will prepare a preliminary summary within 1–2 business days. Or get a consultation on choosing the approach: RAG, fine-tuning, or hybrid — we will tell you what works best for you. Contact us to discuss your LLM development needs. Schedule a free consultation today.