Яндекс.Метрика

Which AI tools should you choose for work?

One AI tool is better at texts, another at code, a third makes images, so a team usually needs different tools for different tasks. We've gathered popular foreign and Russian AI tools in one place: pick a task in the demo below and you'll see right away which ones suit you.

What you need to do

As of September 2026

Conditions

AI tools · 11 tools fit

Click a row to open its card

How to choose an AI tool for work

There's no single best AI tool for everything. One writes texts well, another writes code, a third makes images. So the choice comes down to four questions:

  • What's the task. Texts and emails, translation, going through documents, code, images, video, voice-over and music or finding information with links to sources.
  • Does it work from Russia. Some foreign services aren't officially available in Russia, and their paid plans take foreign cards only. For a team that's unreliable, since access can disappear any day.
  • How much it costs. Almost every tool has a free plan, but with limits. Daily work usually needs a subscription.
  • How it writes in Russian. Russian models are trained on Russian texts and handle the language better, especially in business writing.

In the demo above, you can pick a task and conditions and see which AI tools suit you. Service terms change often, so check plans and availability on the service's website before you buy.

Foreign AI tools for texts, code and search

  • ChatGPT (OpenAI, USA) is an all-rounder: it writes and edits texts, works with files and spreadsheets, writes code, makes images and searches the web. It doesn't officially work from Russia.
  • Claude (Anthropic, USA) is strong at long texts, large documents and code, and follows instructions closely. Many AI agents run on Claude, including Claude Code. It doesn't generate images. It doesn't work from Russia.
  • Gemini (Google, USA) handles very long documents, makes images, searches the web and is built into Google's mail and docs. It doesn't work from Russia.
  • Grok (xAI, USA) writes texts and code, finds fresh news and discussions on X, and makes images and short videos. The paid plan takes foreign cards only.
  • DeepSeek (China) is strong at code, math and reasoning, free in the chat and available from Russia. The model is open: you can run it on your own servers.
  • Qwen (Alibaba, China) is a family of open models for texts, code and working with images. The free chat opens from Russia. Companies often use Qwen as the base for an in-house AI assistant.
  • Mistral (Mistral AI, France) is a European AI: the Le Chat assistant writes, translates, works with documents and code. Some of the models are open. The paid plan takes foreign cards only.
  • Perplexity (USA) is an AI search engine: it answers the question and gives links to sources right away. The paid plan takes foreign cards only.
  • NotebookLM (Google, USA) answers questions using only your files: contracts, reports, articles, and links every answer to the exact place in the document. It makes summaries and podcast-style audio overviews. It doesn't work from Russia.
  • DeepL (Germany) is a translator that takes context into account. It translates whole documents and keeps the formatting, with glossaries for company terms. The paid plan takes foreign cards only.

AI tools for images, video and music

  • Midjourney (USA) makes images for illustrations, covers and concepts, and can animate them into short videos. There's no free plan, and it takes foreign cards only.
  • Nano Banana (Google, USA) is the image model inside Gemini. It edits an existing image precisely from a description: removes the background, swaps an object, keeps a face and style consistent across a series. It's available wherever Gemini is, so it doesn't work from Russia.
  • Veo (Google DeepMind, USA) makes short realistic videos with sound built in: speech, effects and music. It's available in Gemini with a subscription and doesn't work from Russia.
  • Kling (Kuaishou, China) makes videos from a description and animates images, keeping motion and faces consistent. The free credits are enough to try it, and the paid plan takes foreign cards only.
  • Seedance (ByteDance, China) makes videos from a description or from images, with several shots in one clip. It's available in ByteDance services such as Dreamina and CapCut.
  • MiniMax (China) offers Hailuo video, voice-over and voice cloning, and music generation. The company also has open text models.
  • Suno (USA) writes songs and music from a description, with vocals in Russian or English or instrumental. Good for background music for videos and podcasts. Commercial use of tracks needs a subscription.
  • ElevenLabs (USA) voices text with natural-sounding voices in dozens of languages, clones voices and dubs video. The paid plan takes foreign cards only.

The paid plans of these services take foreign cards only. Among Russian services, SaluteSpeech by Sber and Yandex SpeechKit handle voice-over and speech recognition, and Kandinsky makes videos from prompts in Russian.

Russian AI tools

  • GigaChat (Sber) writes well in Russian, is available in Russia and is paid in rubles. It generates images via Kandinsky and music inside the chat. For companies there's GigaChat API and Sber's business offering.
  • Alice AI and YandexGPT (Yandex). Alice AI writes texts, searches the web using Yandex Search, summarizes articles and videos and makes images. Companies connect YandexGPT models via Yandex Cloud, with a contract and payment in rubles.
  • Kandinsky (Sber) generates images and video from prompts in Russian. It's available in GigaChat and via API.
  • Shedevrum (Yandex) is an app for generating images and short videos, good for illustrations for posts.
  • GigaCode and SourceCraft are coding assistants from Sber and Yandex. They plug into your development environment and suggest code.

On complex code and large documents, Russian models are still behind the leaders. On the other hand, they're officially available and paid in rubles.

How to roll out AI tools in your team

According to Pachca's research, 58% of employees who use AI handle work tasks through personal accounts. This is called shadow AI. The sooner a company picks official tools, the less employees will work in personal accounts.

  1. List the tasks employees already handle with AI or want to: emails to customers, reports, code, images for social media.
  2. Pick one main AI tool for texts and documents and one or two more for code, images or search. Make sure they work from your country.
  3. Get access in the company's name: a business plan or API instead of personal subscriptions.
  4. Write a short policy: which services are allowed and who manages access. Pin it in the team chat.
  5. Connect AI to your work chat. That way the AI sees the context of the conversation, and nobody has to copy messages into a separate window.

In Pachca, calls in group chats and channels are transcribed automatically, and the AI summary lands in the thread after the call, as in the use case "How do you record a meeting and get the minutes automatically?". A personal AI agent running on Claude, GPT or another model can be connected to Pachca: it reads the same chats you do and drafts reports and messages. Which tasks to hand it is covered in the use case "How can a manager use AI for reports and planning?". And an agent for the whole team can be built without code.

Frequently asked questions

GigaChat or Alice AI for texts and search, Kandinsky or Shedevrum for images. Among foreign tools, DeepSeek and Qwen open in Russia.

Among Russian tools, GigaChat and Alice AI: they're trained on Russian texts. Among foreign ones, ChatGPT, Claude and Gemini write well in Russian.

Usually one main tool for texts and documents is enough, plus one or two for specific tasks: code, images or search with sources.