Pailot
What stays in the contour versus what leaves?
| Inside the contour | Must call outside | |
|---|---|---|
| Language model | PailotGPT can run the agent step on your hardware | Every step calls an external API |
| Files and speech | RAG and speech stay in the same contour | Files leave for a third-party parser |
| Copilot | Cloud with redaction, or entirely inside | The assistant always lives outside |
| Browser shots | In the run log only when a human is needed or a run fails | A remote desktop stream to the vendor |
What do security and procurement ask?
Security and procurement want the stack on their hardware. The box below is the delivery list. This is not a certificate page.
What does not have to leave?
Own language model, search over company files, speech recognition, and in-place document parsing.
- The language model runs agent steps on customer hardware.
- Document search answers from company files and shows the source.
- Speech recognition stays in the same contour.
- Browser screenshots land in the run log only when a human is needed or a run fails.
Which two copilot modes exist?
Cloud copilot with redaction, or copilot entirely inside. Already published processes do not need the cloud. The choice is fixed when the platform is installed.
- Cloud copilot: personal and commercial fields are redacted before a request leaves.
- Copilot entirely inside: the assistant, the model, search and speech stay on your hardware.
What is in the box
Core of the delivery, not a connector list. Integrations sit under this.
- Copilot
- Drafts the process from a brief, checks links and publishes. At install time you choose a cloud copilot with redaction or a copilot that stays on your servers.
- Language model
- An LLM runs the agent step: reads text, decides, writes the reply. In the contour that model sits on the customer’s hardware.
- Document search
- RAG answers from your files and shows the source. The index stays in the contour.
- Speech and diarization
- Speech-to-text transcribes audio and video. Diarization splits speakers. A meeting can become a protocol or a reply in the same thread.
- Document parsing
- PDFs, spreadsheets, audio and video are parsed on-site. The file does not have to leave for a third-party parser.
- Browser step
- The process opens a live page and works like a clerk when no connector exists.
Domestic connectors in the box: Битрикс24, amoCRM, MAX, VK, Telegram. The rest come from the catalog (162 blocks and connections) when they already live in the company contour.
Where an agent can answer
| Channel | What it does |
|---|---|
| Web chat (fastest) | The fastest way out: the process becomes a chat page. A person types, the agent answers. Use it before wiring other systems. |
| API | Your system starts the process with a request and reads the result. |
| A2A | Agent-to-Agent: publish the process so other agents can call it over the A2A protocol. |
| MCP | Model Context Protocol: publish the process as a tool another assistant can call. |
Questions
- Do you publish on-premise license prices?
- No. Price is on request after we see the process and the hardware outline.
- Is PailotGPT required?
- PailotGPT is the in-contour default for agent steps. Other families can be opened for the copilot if the customer allows it.
- Where is the company based?
- ООО «Профитайзер», 127566, г. Москва, Высоковольтный проезд, INN 9715424371.
- Can the copilot stay fully inside?
- Yes. That mode is chosen at install. Published processes then run without an external call.
- How do we get access?
- Submit «Получить доступ». Hardware and scope come after the request.