How the work goes
Six steps — and only the first and the fifth are yours.
- Task
The task
You describe the task in your own words.
- Clarify
Clarification
The system asks questions: which period, which breakdowns, what to do with returns, whether group totals are needed.
- Approve
Approval
You see the final specification before any code generation begins.
- Build
The build
Code is written and compiled by the Designer on your base.
- Check
Your check
You review the result.
- Done
Done
You take the file.
If the code does not compile, you never hear about it: the compiler error goes back into the loop and the code is rewritten.
Not happy with the result? Say what is wrong — the task repeats with your remarks.
The system asks first, then acts
Clarifying questions
The questions are asked against your configuration — the system already knows which objects and attributes it has.
Specification check
Requirements are re-read in a separate pass: no contradictions, no gaps.
Approval with you
You see the final specification before generation starts.
Why the result works instead of “looking right”
- 01
The code is compiled by the 1C Designer on your base — the same tool a programmer uses.
- 02
If it does not compile, it is not delivered. You only receive what passed compilation.
- 03
The system works with the exact contents of your configuration, including your modifications and installed extensions.
- 04
Compile errors automatically return into the loop: the code is rewritten based on the Designer’s message.
Your data stays with you
What exactly leaves your perimeter — and what never does.
| What | Leaves your perimeter |
|---|---|
| Configuration structure: object and attribute names | Yeswithout it there is nothing to write code against |
| Module texts, your business logic | Nothey stay on your machine |
| Base data: stock, documents, contractors, amounts | Nothe service never connects to the base |
| The task description and clarification thread | Yesthis is the task itself |
| Generated code and the compiled file | Yesthey are the result of the work |
Compilation runs on your machine with your own Designer — the service has no access to the base.
Strict separation: the cloud thinks, the on-site agent executes
Service cloud
- Chat, requirement clarification, code generation
- Routing between AI providers
- Task state, billing, history
- Account and admin panel
Your perimeter (Windows machine)
- The agent — a Windows service on the machine running the 1C platform
- Launches the Designer: structure export, build, check
- Polls the service itself — no inbound ports to open
- The access key is stored using Windows protection mechanisms
Channel separation: the backend never talks to 1C directly, and the agent never talks to the AI.
Setup takes one working day
No separate server needed: any machine already running 1C will do. No changes to your configuration — everything works through external files.
If data must not go to external AI
Russian models
YandexGPT and GigaChat — data never leaves the country.
Local models
Ollama and LM Studio on your own hardware — no internet needed for generation at all.
Your choice
The provider is switched in settings; the allowed list can be restricted.
A local model generates worse code than a top cloud one — a conscious trade-off. Modes can be combined: consultations locally, hard generation on a cloud model.
Worth knowing in advance
- The product is at the MVP stage: fully built and working, now being validated on real bases.
- Complex tasks with non-standard logic still require a programmer — the system targets routine.
- A standard report, a selection, a simple data processor — yes. Reworking accounting methodology — no.
- Working with the Designer requires the base not to be exclusively locked at that moment.
- The first pilot runs on a test copy of the base — a condition, not a suggestion.
Frequently asked questions
- 01
Can AI damage the books?
Everything written to the base goes through approval: the system prepares — a human confirms. Before serious operations — a backup and a trial run on a copy of the base, and the first pilot always runs on a test copy.
- 02
How is AI Dev Assistant different from 1C:Naparnik?
1C:Naparnik is built into 1C:EDT and helps a developer write code. AI Dev Assistant is for a specialist without a programmer: you describe the task in words and receive a ready file (.erf, .epf, extension), compiled by the Designer on your base.
- 03
Does our data leave the perimeter?
Only the configuration structure goes out — object and attribute names. Module texts, business logic and base data stay on your machine. Russian or fully local models are available.
- 04
What happens if the code does not compile?
You never hear about it: the compiler error returns into the loop and the code is rewritten automatically. Only what passed compilation is delivered.
- 05
How much does it cost?
Quota-based plans; the company materials give a 2–15k ₽/month range depending on work volume. Exact names and prices are shown on the demo.
A pilot on a single task
Demo — one-day agent setup — a real report on a test copy of your base — your decision.