Teaching Doti in Real-time

Doti is not a black box. It’s a living, evolving knowledge engine - and your team plays a direct role in keeping it accurate, up to date, and deeply aligned with how your company works.

We’ve designed two powerful mechanisms to help you teach, correct, and improve Doti continuously, without tickets, bottlenecks, or long feedback loops.

---

1. Modify Answers & Use Doti Memo

Whenever Doti gives an outdated, incomplete, or even incorrect answer - you can correct it on the spot.

How it works:

• Every Doti answer includes a "Modify" button

• When clicked, you can rewrite or enhance the answer

• The new response is saved as a Dot (memo)

📌 From that moment on, Doti will use your updated answer for similar queries - across Slack, the browser extension, and even the API if agent-scoped.

This is a game changer.

✅ No more updating Confluence pages ✅ No waiting for ops to "clean up the data" ✅ No hunting down the right owner - just fix it and move on

🧠 You’ve just closed the knowledge loop in seconds - directly in Slack.

---

2. Structured Thumbs Down Feedback

Most AI tools offer a “thumbs up/down” icon that does... something.

But here’s the problem: they usually don’t tell you why an answer failed. And worse - they can’t tell whether the knowledge even exists in your company.

So we built a better system.

Doti Thumbs Down = 3-Option Feedback

When you click 👎, you’re given three simple, high-signal options:

1. Incorrect resources → Doti cited the wrong docs

2. Correct resources, bad answer → Good retrieval, poor reasoning or synthesis

3. Outdated answer → The content used is no longer relevant or correct

⚡ No typing required. Just one click - real signal.

---

What Happens with This Feedback?

1. Actual Optimization

Each type of feedback helps Doti improve differently:

Signal	Optimization Focus
Incorrect resources	Tune retrieval logic, improve filters, debug relevance scores
Correct resources + bad answer	Refine GPT instructions, agent steps, formatting logic
Outdated answer	Flag stale sources or prioritize Dots and fresher material

This is not just model tuning - it’s full-stack optimization: retrieval → reasoning → formatting → outcome.

---

2. Knowledge Posture Insights for Your Org

At a higher level, we aggregate this feedback to generate a “Knowledge Posture” report for your leadership.

It answers questions like:

• What % of answers are relying on outdated content?

• Where are your biggest knowledge gaps?

• Which teams give the most corrections?

• What are the common types of failure in reasoning vs retrieval?

🧩 This becomes your company’s AI readiness scorecard, showing where to invest in enablement or documentation.

---

Summary

Method	What It Does
Modify Answer	Rewrite an incorrect or outdated answer - instantly creates a persistent Dot
Doti Memo	Log new knowledge manually to be used in future answers
Thumbs Down (Structured)	Provide 1-click, meaningful feedback tied to specific failure modes
Optimization Outcomes	Improves retrieval, GPT reasoning, and org-level knowledge hygiene
Strategic Impact	Surfaces actionable gaps for ops, enablement, and leadership

---

💡 Doti isn’t just getting smarter - it’s getting smarter because of you. And unlike generic AI tools, every correction, memo, and signal leads to action.