Confidential · © 2026 Space-O Technologies Inc. · Prepared exclusively for DMA. Please do not copy, save or share.

Section 9 · Buyer Psychology & Sales Method

9.1Detect buyer type

Built in the POC
No.
9.1
DMA's reference
Decision Book p.20
Phase
Phase 1
Autonomy
L4 · Autonomous
DMA's decision
Yes

What DMA asked for · Decision Book p.20

Recognise an urgent buyer, a skeptic, a researcher, a price shopper or an executive who wants it short.

Why it matters to DMA: So the agent adapts its pace and tone instead of running one script for everyone.

In our words, from DMA's Product Decision Book v3, page 20.

How we do it

  • After every exchange, a smaller AI model (the note-taker) reads the recent chat and records the buyer type as one of six fixed values: urgent, skeptic, researcher, price shopper, brevity executive or not yet known.
  • The conversation service passes that buyer type to the strategist before its next reply, and the strategist's approved instructions say how to adapt to each.
  • The note-taker can only choose from the fixed list, and the visitor's words reach it marked as untrusted data, so a visitor can't talk it into anything else.

The flow

It plays on its own while it's on screen; hover or use the controls to pause or step through.

Visitor writestone and wordingNote-takerpicks one of sixBuyer typestored with the chatTurn contextgiven to the strategistAdapted replypace and length
Step 1 of 4

The visitor writes

"Every agency promised us the world. Why would you be different?"

  • Person
  • AI
  • Stored data
  • Rule in code
  • Screen

Where the data goes

The same six stages on every page. Nothing reaches Nutshell except through the write gate.

  1. 1Website chat

    Source

    What the visitor writes.

  2. 2Sync

    Copies Nutshell changes into the bridge database

    Not involved.

  3. 3Bridge database

    A copy of the CRM data, plus what the AI works out

    Keeps the running notes, including the buyer type.

  4. 4AI

    Claude models, only through DMA's own gateway

    A smaller model records the buyer type; the strategist adapts to it.

  5. 5Write gate

    The only way back into Nutshell: checked, approved when needed, sent once

    Writes the lead, its score and the summary if the visitor books or leaves an email.

  6. 6Nutshell

    The system of record

    The rep sees the summary and the score.

The tables behind it

Drawn from the POC's database catalogue: structure only, no data.

conv.conversationsPKiduuidchanneltextstagetextIDagent_version_iduuidstarted_attimestamp+ 17 more columnsconv.messagesPKiduuidFKconversation_iduuidseqintegerIDparticipant_iduuidsender_kindtextsent_attimestampmetajsonbIDdecision_iduuidIDagent_version_iduuid+ 12 more columnsctrl.decision_logPKiduuidPKoccurred_attimestampIDagent_version_iduuidIDrun_iduuidtooltextdecision_typetextIDconversation_iduuidpolicy_resulttextreason_codesjsonb+ 19 more columns

PK primary keyFK reference the database enforces (solid line)ID reference kept by id (dashed line)

TableWhat it holdsColumns
conv.conversationsOne row per chat: its stage and the AI version that ran it.22
conv.messagesEvery line of the chat, encrypted.21
ctrl.decision_logEvery reply and check, with the model, tokens and cost.28
How the tables connect (8)
ColumnPoints toKept by
conv.conversations.agent_version_idagent.agent_versionsThe application (by id)
conv.messages.conversation_idconv.conversationsThe database (foreign key)
conv.messages.participant_idconv.participantsThe application (by id)
conv.messages.decision_idctrl.decision_logThe application (by id)
conv.messages.agent_version_idagent.agent_versionsThe application (by id)
ctrl.decision_log.agent_version_idagent.agent_versionsThe application (by id)
ctrl.decision_log.run_idagent.runsThe application (by id)
ctrl.decision_log.conversation_idconv.conversationsThe application (by id)

Worked example

Synthetic demo data: every name, business and number is made up.

Adapting in a real chat

1 / 2

What you see

1.The visitor's message

The note-taker reads it after the reply and updates the notes, including the buyer type.

A law-firm owner describes the problem: the strategist answers like a consultant, not a brochure.
Callout 1

What happens in the system

  • LLM gatewaynote-taker

    Claude Haiku 4.5, structured output only: the buyer type is one of six fixed values.

Right-hand values come from a real run of the POC on September 25, 2026 (trace tr_muh5t7yd21f50nl0xe4kx, tr_muh5t8021x0f48ojhiv2d). Rows marked “sample” are illustrative.

The six types and how the strategist adapts (the adaptations are in its approved instructions).

Buyer types

Buyer typeSounds likeThe strategist
Urgent"We need more calls this month"A shorter path to the call.
Skeptic"Every agency promised us the world"Leads with the evidence from their own site.
Researcher"Just looking at options for now"Teaches first, no push to book.
Price shopper"What do you charge for SEO?"No price: a strategist covers it on the call, or a person takes over.
Brevity executive"Short answers please"Two sentences at most; the shortest qualification.

In the running POC

A law-firm owner describes the problem: the strategist answers like a consultant, not a brochure.

A law-firm owner describes the problem: the strategist answers like a consultant, not a brochure.

  1. 1The visitor's problem, in their own words.
  2. 2A short reply that reflects the problem and asks one question.

Website chat · Harbor & Finch Law demo site. Captured from the running POC on September 25, 2026; synthetic data.

See it live

Opens the exact screen in the running POC, in a new tab.

OpenWhereWhat to try
Harbor & Finch Law demo siteDemo prospect websiteAsk for short answers, or say you've been burned before, and watch it adapt.

Status

Built in the POC

Built in the POC. This capability runs in the POC today, on synthetic data, end to end.

9.1 Detect buyer type · DMA AI proof