Frequently asked questions

Your business runs on a Business Brain (your own SOPs, in plain text) plus AI agents that read it before answering. Goal: 60–80% of recurring workflows handled by AI; humans focus on judgment, exceptions, and customer relationships.

ChatGPT-the-chatbot is generic. The Be AI-First Framework gives the AI YOUR rules, YOUR pricing, YOUR tone, YOUR org chart — and wires it into the tools you already use. The AI stops making things up.

No. It replaces middle-management reporting (status emails, weekly summaries, daily standups). Your people get those hours back to do the high-judgment work software can't do.

Week 0–2: Champions report 30+ minutes saved per task. Week 6: First production agent live. Week 12: Daily intelligence brief replaces the morning standup.

The framework assumes that. Champions are 'curious, not necessarily technical'. Bettroi engineers do the wiring; your team writes notes in Markdown (which is just typing in plain text).

Bettroi leadership includes a practising surgeon. We've shipped clinical-grade software. We never feed PHI/PII into external AI APIs without a Data Processing Agreement. Per-record consent gates are default-OFF.

A folder of plain-text Markdown files on your laptop, synced to GitHub privately, indexed into a database your agents read. No 'AI black box' — you can open every file in Notepad.

1–2 'AI Champions' per department (curious people, not necessarily technical). Leadership for ~2 hours/week of review meetings. One designated AI Ops Lead by Week 6.

No. We connect to what you already use. The Business Brain is additive. We may flag tools as redundant during Discovery, but the call is yours.

Your SOPs and rules live in a private GitHub repo + Supabase database (your tenant; we never co-mingle). Customer/patient data stays in its existing system. Combined only at agent runtime.

Every agent output passes through a scoring checklist before delivery. Below threshold = loops back for revision automatically. Customer-facing outputs go through a human reviewer until trust ≥9/10. Two independent kill switches let you stop AI features instantly.

Yes — without engineering. You edit a plain-text file in your vault. ~10–15 minutes later the agent reads the new version. No deploy, no PR, no waiting.

Bettroi's open-source skill harvester. Every time a Claude Code session in your tenant produces a useful prompt or workflow, ironbark captures it as a reusable skill. Your AI Skills Library grows automatically.

Discovery Sprint is the entry point (~one week, fixed scope, $2k–5k). After that you commit one sprint at a time. Full first-year typical: $50k–130k.

You do. Code is in your GitHub repo, data in your Supabase project, agents in your hosting account. Bettroi consults; we don't lock you in.

You keep everything: the vault, the agents, the Skills Library, the integrations. We hand over docs and stop billing. Migration risk is near-zero because the stack is open.

Yes — Claude API, embedding models, etc. Typical 10-person team: $200–$2000/month depending on agent volume. Optional migration to claude-code-deepseek-backend cuts this 60–80%.

The AI Leverage Ratio dashboard (live by Week 11): tasks completed per week ÷ hours of human input. Target trajectory: 2× → 5× → 10× over four quarters.

Scale Retainer: monthly cadence. We run the Learn → Wire → Automate loop on the next 10 processes. Quarterly retro audits which agents are pulling weight and retires the ones that aren't.

Yes — see github.com/chatgptnotes for 20+ live deployments and public repos: zeroriskagent.com, bni121.vercel.app, plcautopilot.com, drmhope.com, Linkist NFC, and more.