AI Assistant Inside Your App, Connected to Your Data and Tools by Ahmad MuhyidinAI Assistant Inside Your App, Connected to Your Data and Tools by Ahmad Muhyidin
AI Assistant Inside Your App, Connected to Your Data and ToolsAhmad Muhyidin
Goal: an AI assistant in your app that answers from your real data and shows what it used.
For a Swiss industrial client I built the AI assistant and MCP server of their data platform from the start, inside their engineering team.
Starter: a chat assistant, one model, one read-only API tool on one service account, on staging.
Standard: adds up to 50 text files (10 MB) with sources, 2 read-only tools (API or MCP server), your app's login and per-user access.
Advanced: up to 3 tools incl. one write the user confirms, an audit trail, answer tests, one deployment.
Not included: an MCP server for outside AI tools (my MCP Server service), background AI without a chat (my Workflow Automation service), model training, voice, autonomous agents.
Done when:
it passes the agreed test questions and shows an error if a tool fails;
from Standard, a test user sees only what they may see;
in Advanced, the write waits for confirmation and is logged;
you have the source, setup notes and known limits.
You pay for model and API use. NDA on request.
Packages: Starter $400, Standard $1,000, Advanced $1,900.
Goal: an AI assistant in your app that answers from your real data and shows what it used.
For a Swiss industrial client I built the AI assistant and MCP server of their data platform from the start, inside their engineering team.
Starter: a chat assistant, one model, one read-only API tool on one service account, on staging.
Standard: adds up to 50 text files (10 MB) with sources, 2 read-only tools (API or MCP server), your app's login and per-user access.
Advanced: up to 3 tools incl. one write the user confirms, an audit trail, answer tests, one deployment.
Not included: an MCP server for outside AI tools (my MCP Server service), background AI without a chat (my Workflow Automation service), model training, voice, autonomous agents.
Done when:
it passes the agreed test questions and shows an error if a tool fails;
from Standard, a test user sees only what they may see;
in Advanced, the write waits for confirmation and is logged;
you have the source, setup notes and known limits.
You pay for model and API use. NDA on request.
Packages: Starter $400, Standard $1,000, Advanced $1,900.