docs(demo): rewrite the pitch as a SPOKEN speech (no terminal/slides)

Sam delivers this on a stage to 50-100 people — voice only, no screen. Rewrote
from a terminal-demo script into a pure ~60s spoken pitch built for the ear:
short sentences, hard stops, [pause] cues (silence = the loudest move on stage),
and one memorizable detonation line ("a root cause never looks like the bug it
creates"). Opens about the audience's pain, not the product; promises the
laptop demo afterward rather than fumbling a terminal live. Includes a 30s
hallway version and a 5s one-liner.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
Sam Valladares 2026-06-27 19:05:49 -05:00
parent a7d8105584
commit b049cd5d39

View file

@ -1,136 +1,108 @@
# Postdict — the 60-second "blow it wide open" script
# Postdict — the 60-second SPOKEN pitch
**Audience:** investors. **Goal:** they feel the pain, see the flawed axiom crack,
watch the impossible happen live, and understand the market is *everything*.
**Format:** live terminal, one take, you talking over it. ~60s.
**Rule:** ~20s of talk. The rest is the terminal doing the impossible.
**This is a SPEECH.** You, on stage, looking at 50-100 engineers / founders /
investors. No terminal. No slides. No screen. Just your voice and the room.
> Why NOT "X companies got hacked": it's a commodity security-FUD hook, it needs a
> stat you can't cite live (which torches your credibility — your whole moat is
> *honest + reproducible*), and it shrinks you from "every agent in production" to
> "security." Instead, open with the wound every engineer in the room has lived.
Built for the *ear*: short sentences, hard stops, one line they'll repeat in the
hallway. Read the **[beats]** as pauses — silence is your loudest move on stage.
---
## THE SCRIPT (what's on screen · what you say)
## THE PITCH (≈60 seconds, spoken)
### [0:000:12] — THE WOUND (make them feel it personally)
**On screen:** a clean dark terminal, blinking cursor. Nothing else yet.
**You say (calm, then sharper):**
> "Every engineer in this room has lost a day to this: production breaks, and the
> cause turns out to be a one-line change you made days ago and never thought
> about again. The bug looked nothing like that change — so you never connected
> them. You just *suffered* until you stumbled onto it.
> Every engineer in this room has lost a day to the same thing.
>
> Now we've handed that exact job to AI agents. And here's the problem—"
> Production breaks. You burn hours hunting. And the cause turns out to be one
> line you changed three days ago — and forgot.
>
> **[pause]**
>
> The bug looked nothing like that change. So you never connected them. You just
> suffered until you stumbled onto it.
>
> **[pause]**
>
> Now we've handed that exact job — debugging — to AI agents. And every AI memory
> system on Earth is about to fail at it. Mem0, Zep, all of them. Because they're
> built on one assumption: that *relevance equals resemblance.* They search your
> memory for whatever *looks like* the problem.
>
> **[slow down — this is the line]**
>
> But a root cause never looks like the bug it creates.
>
> **[full stop. let it hang.]**
>
> So the entire industry is searching in the one place the answer can never be.
>
> **[pause]**
>
> We built the opposite. When your agent hits a failure, our memory reaches
> *backward in time* — and finds the quiet change, days earlier, that actually
> caused it. The one no similarity search will ever surface, because it isn't
> *similar* — it's *upstream.*
>
> We didn't invent this. We ported it from your brain. There's a 2024 *Nature*
> paper: when something goes wrong, the brain reaches backward to find the cause
> — backward only, because a cause is always in the past. We turned that into
> software. It runs locally. Today.
>
> **[pause]**
>
> And the incumbents can't copy it — their whole architecture *is* the wrong
> assumption. To do this, you rebuild memory from the brain up. We already did.
>
> **[land it]**
>
> Everyone else built a memory that *remembers.* We built the first one that
> *realizes.* Every AI agent that touches production needs it — that's the whole
> market, and it's on fire.
>
> I'm Sam. We're Postdict. The first memory that finds the *cause*, not the
> lookalike. Come find me — I'll prove it on a laptop in thirty seconds.
---
### [0:120:22] — THE FLAWED AXIOM (the line they repeat to their partners)
## HOW TO DELIVER IT (this is 80% of the win on a stage)
**On screen:** type one line:
`relevance ≠ resemblance`
1. **Memorize three sentences. Improvise the rest.** If you only nail three,
nail these:
- **The wound:** *"The cause turns out to be one line you changed three days ago — and forgot."*
- **The detonation:** *"A root cause never looks like the bug it creates."*
- **The category:** *"Everyone else built a memory that remembers. We built the first one that realizes."*
**You say:**
> "—every AI memory framework on Earth, every VC-backed startup, every platform
> layer, is built on one flawed assumption: that **relevance equals resemblance.**
> They search memory for what *looks similar* to the problem. But **a root cause
> never looks like the bug it creates.** The entire industry is searching in the
> one place the answer can never be."
2. **The detonation line is the whole pitch.** Walk *toward* the audience as you
say it. Then STOP. Say nothing for two full seconds. That silence is you
letting 100 people independently realize you're right. Do not rush it.
*(Beat. Let it sit.)*
3. **Open about THEM, not you.** The first 15 seconds is their pain, not your
product. When heads nod because they've lived it, the room is yours — you
haven't even said what you do yet.
4. **Short sentences. Hard stops.** On a stage, a long sentence loses the back
row. Every sentence above is built to be said in one breath. Trust the
periods.
5. **Don't demo on stage — *promise* the demo.** "Come find me, I'll prove it on
a laptop in 30 seconds" is stronger than fumbling a terminal in front of 100
people. It pulls the serious ones to you afterward, one-on-one, where deals
actually start.
6. **End on your name + the one-liner.** "We're Postdict. The first memory that
finds the cause, not the lookalike." That's what they Google in the parking lot.
---
### [0:220:35] — MAKE IT CONCRETE (type it live)
## THE 30-SECOND VERSION (if you only get a hallway / elevator)
**On screen:**
```
$ postdict ingest "Set API_TIMEOUT=2 in the deploy env" --ago-days 3 # the quiet cause
$ postdict ingest "500 error in the billing service" --ago-days 20 # an old lookalike
$ postdict ingest "Service crashed: 500 on the auth endpoint" # today's crash
```
> "Every AI memory tool searches for what your bug *looks like.* But a root cause
> never looks like the bug it creates — it's a config change from three days ago
> that looks nothing like the crash. So they all miss it. We built the first
> memory that reaches *backward in time* to find the actual cause. Ported it
> straight from a *Nature* paper on how the brain does it. Every agent that
> touches production needs it. I'll show you on a laptop right now."
**You say:**
> "Watch it happen. A one-line config change three days ago. An old, unrelated
> 500 error weeks back. And today — the auth service crashes. Which past memory
> caused it? To a vector database, today's crash looks most like that old billing
> 500. The thing that *looks* similar is never the thing that *caused* it."
## THE ONE LINE (if you get five seconds)
---
### [0:350:50] — THE PROOF (the money shot · go silent on the reveal)
**On screen:** `$ postdict backfill --contrast`
```
── 1. SIMILARITY SEARCH · keyword (BM25) ──
1. 500 error in the billing service ← top match (WRONG)
→ ranked by RESEMBLANCE. its top hit is a lookalike, not the cause.
── 2. POSTDICT (reach backward for the CAUSE) ──
#1 Set API_TIMEOUT=2 in the deploy env
↩ reached back 3.0 days before the failure
🔗 causal join: api_timeout (RIGHT)
```
**You say (let `↩ reached back 3.0 days` hold in dead silence for a full second):**
> "Same database. Same question. Similarity search returns the lookalike —
> confident, and wrong. Postdict reaches **backward three days** and finds the
> actual cause. Not because it's similar — because it's **causally upstream.**
> That day you lost? Gone. It's instant now."
---
### [0:500:58] — THE MOAT (kill "can't they just add this?")
**You say:**
> "And they can't copy it. This is a faithful port of a 2024 *Nature* result —
> the brain reaches *backward* in time to find causes, backward-only, because a
> root cause is always in the past. The incumbents can't bolt this on; their
> whole architecture **is** the flawed axiom. You have to rebuild memory from the
> cognitive science up. We already did. It runs locally, today."
---
### [0:581:00] — THE ASK
**On screen:** `the first memory that finds the cause, not the lookalike.`
**You say:**
> "Every agent that touches production needs this. That's the whole market, and
> it's on fire. We're raising [X] to make every AI agent debug like a senior
> engineer. The seed's in the repo — run it yourself."
---
## DELIVERY — THE 6 RULES THAT MAKE IT LAND
1. **Open with the wound, not the product.** 0:000:12 is about *them*, not you.
When an investor nods because they've lived it, you've already won the room.
2. **The monologue is ~20 seconds total.** Investors fund what they *see* work.
Get to the terminal fast; the contrast carries the weight.
3. **Two sentences, memorized cold:** the axiom — *"They believe relevance equals
resemblance."* The detonation — *"A root cause never looks like the bug it
creates."* Everything else can be loose.
4. **Dead silence on the reveal (0:350:50).** When `↩ reached back 3.0 days`
appears, say nothing for one full second. The screen sells it.
5. **Answer the moat before they ask it.** "Can't Mem0 add this?" — no, their
architecture is the axiom. That sentence is what turns "neat" into "fundable."
6. **End on the market (TAM), not the feature.** "Every agent that touches
production" is the size of the prize. The demo earned you the right to say it.
## THE THREE LINES THAT DO THE WORK
- **The wound:** "Production breaks, and the cause is a change you made days ago and forgot."
- **The detonation:** "A root cause never looks like the bug it creates."
- **The category:** "We're not a better memory. We're the first memory that reasons backward."
---
## NOTE ON THE BINARY
The script types `postdict` (the new name). The shipping binary is still `vestige`
today — for recording, either (a) `alias postdict=./target/release/vestige`, or
(b) record now with `vestige` and re-record after the rename. The on-screen output
is identical; only the command word differs.
> "We're the first AI memory that finds the *cause* of a bug, not the lookalike —
> because a cause never looks like the bug it creates."