Skip to content

TALK_OUTLINE.md

Latest commit

 

History

History
92 lines (64 loc) · 2.97 KB

TALK_OUTLINE.md

File metadata and controls

92 lines (64 loc) · 2.97 KB
Raw

Part 1 Talk Outline

Title

Morning espresso with Nimble

Thesis

Nimble turns live web chaos into structured, source-grounded, agent-ready intelligence — before the LLM reasons.

Deck

Premium keynote deck (rebuilt from first principles):

  • Present: with the app running, open http://localhost:3001/slides/index.html
    • Or open http://localhost:3001/ and click Open keynote deck →
    • /deck redirects to the same file
  • Speaker notes: presentation/SPEAKER_NOTES.md
  • Controls: arrow keys, on-screen buttons, F for fullscreen
  • Demo video: place the 4-minute recording at public/slides/demo.mp4

Runtime (20 minutes)

Minutes Beat
0–4 Title + embedded Morning Signal demo
4–7 What you saw + live-web problem
7–10 Agent architecture + Nimble in the critical path
10–14 Search, Extract, Agents, Crawl/Map, Drivers
14–17 2×2 positioning + why Nimble wins this use case
17–20 Build path, generalization, close

Hook

Most GTM research is stale before it reaches the meeting. Competitor pages change, docs launch, pricing shifts, and buyers start asking new questions. A generic LLM can summarize what it already knows — it cannot know what changed today unless it can reach the live web through a structured intelligence layer.

Problem

GTM teams do not suffer from a lack of information. They suffer from scattered, constantly changing public signals:

  • Competitor launches
  • Pricing changes
  • Docs updates
  • Community threads
  • Job postings
  • Analyst / market reports

CRM, battlecards, and LLM memory do not know what changed today.

What I Built

Morning Signal is a real-time GTM intelligence agent.

Flow:

Company URL
  → company profile
  → approved GTM context
  → LangGraph planner
  → Nimble Search
  → Nimble Extract
  → evidence packet
  → OpenAI synthesis
  → Morning Signal brief

The LLM is only one node. The agent is the workflow around the model.

Why Nimble

Before synthesis:

  1. Search discovers fresh sources
  2. Extract turns pages into usable evidence
  3. Metadata and source links make output auditable
  4. Structured evidence prevents generic filler

Nimble’s wedge vs. retrieval-only search, crawl-to-markdown tools, automation marketplaces, and heavy web infra: agent-ready structured web intelligence — live search + extraction + agents + crawl/map + pricing/control.

Demo Beats

  1. Enter nimbleway.com
  2. Show researched profile, competitors, ICP, markets
  3. Human approval checkpoint
  4. Live agent-work map: Plan → Search → Extract → Verify → Synthesize
  5. Newsletter: what changed, why now, moves today
  6. Point at sources and LangSmith traces

Close

The most persuasive developer content does not just explain an API. It proves a workflow. Morning Signal shows live web discovery → structured extraction → grounded business action — a pattern developers can inspect, run, and adapt.