When artificial intelligence (AI) enters the conversation in a company, the scenario often follows a familiar pattern. A top executive might suggest: "Let’s build our own AI assistant. It’ll be smart, help employees, and save time."
The project kicks off with vendors, technical specifications, endless discussions, and flashy demos. Ambitions grow, costs skyrocket, and soon the budget hits tens of millions of rubles — yet the impact remains unclear.
Why does this happen? Saving 30 minutes a day for one employee is valuable, but recovering the cost of a massive project requires efficiency gains across thousands of people — a scale rarely achievable in reality.
A more practical approach exists. Instead of complex "hyperagents," companies can adopt a simple, scalable strategy: empower employees to use AI themselves. Think of it like Excel—no one outsources spreadsheet creation. Each employee builds their own table tailored to their needs. AI works the same way.
Level 1: Chat Access
The first step is giving employees access to AI chat tools.
Even without advanced agents, AI can assist with:
- Drafting emails or reports,
- Creating presentations,
- Brainstorming event ideas,
- Processing data.
A good model in a chat interface is all it takes for employees to boost daily productivity.
Level 2: DIY Simple Agents
Next, employees can create their own mini-assistants using intuitive tools.
They can:
- Set a system prompt,
- Upload a few documents,
- Provide a knowledge base.
For example, HR could build an AI-Buddy for new hires, explaining leaves, sick days, and business trips. Lawyers might craft an assistant to review contracts, offering templates and flagging common errors. Platforms like OpenWebUI, which we often use, make it easy for anyone to customize such helpers.
Level 3: Low-Code AI Agents
For slightly complex tasks, low-code platforms like n8n or Make come into play, enabling:
- CRM or task-tracker integration,
- Scheduled triggers,
- API usage,
- Dialog state management.
An example is an assistant that compiles HR reports on sick leaves from multiple systems every morning.
Level 4: Custom Development
For advanced needs, professional developers step in, handling:
- Complex calculations,
- Schedule optimizations,
- Integration with internal databases and intricate logic.
Take an agent optimizing delivery routes across 50 cities, factoring in warehouses and traffic—a task requiring custom expertise.
Also read:
- What Fuels 'Dad TV'? A Peek into Fathers' Viewing Habits
- Facebook Isn’t Just for Grandpas: The Social Network Still Rules, Despite TikTok’s Rise
- Apple Puts All Its Chips on F1 Movie: Potential Sensation or Costly Flop?
Real-World Distribution
Our experience shows:
- 50-60% of initiatives involve chats and simple agents employees can build themselves.
- 20-30% need no-code/low-code solutions.
- Only about 10% demand professional development.
We’re essentially fostering a new literacy within companies. AI isn’t about megaprojects — it’s about equipping people with tools, knowledge, and the freedom to use them effectively.
