v1.0 · production Sid · live Felix · live more in the botshop

Small bots.
Specific moments.
Real behavioural depth.

BehaviorBots is Syntacy's family of single-purpose conversational AI agents. Each one occupies a single commercial decision moment — first purchase, redemption, renewal, complaint — and brings behavioural-science depth to it. Not a chatbot trying to handle everything. A specialised agent designed to do one thing properly.

Meet Sid → Meet Felix →
behaviorbots.system
v1.0
acq
Sid
ret
Felix
thesis

Most AI in commerce is generic. We don't think it should be.

The dominant approach is one large bot trying to handle every customer interaction across the funnel. Tempting model — single deployment, single dashboard — but it sacrifices depth at every decision moment. We've taken the opposite bet.

Status quo
"One big bot for everything."
Why it fails
Generic bots optimise for breadth, not depth. They handle support, sales, retention and acquisition with the same conversational shape — and miss the behavioural mechanics specific to each. Polite chatbots that don't actually move metrics.
Status quo
"AI is the answer to every funnel question."
Why it fails
Most commercial moments are governed by long-studied behavioural patterns — reciprocity, procedural justice, intermittent reinforcement. Bots that ignore these are text-prediction over a dashboard. They don't change behaviour because they don't engage with what behaviour responds to.
Status quo
"More personalisation = better outcomes."
Why it fails
Personalisation alone optimises for what a customer prefers, not what a brand needs. A discount-hunter "personalised" with bigger discounts becomes a more loyal discount-hunter. Without behavioural framing, personalisation entrenches the wrong patterns. The bot needs an opinion.
definition

A BehaviorBot is narrow on purpose.

Each bot in the family shares four characteristics. Together, they're what makes a BehaviorBot different from a generic conversational layer.

trait_01
One decision moment
Each bot occupies a single commercial moment — first purchase, redemption, renewal, refund. Never two. The discipline of narrowness is what allows depth.
trait_02
Behavioural-science calibrated
Each bot built on a specific theoretical foundation — Life History theory, procedural justice, paradox strategy, variable reinforcement. Not vibes. Citations.
trait_03
Dialogic, not interrogative
Acknowledges before asking, responds with specificity, walks away gracefully when the customer signals it's not the moment. Hosts, not interrogators.
trait_04
Score hidden, data captured
Five-axis behavioural read runs silently. Customer experiences a warm conversation. Brand receives a structured behavioural fingerprint plus preference data.
bots / live

Two bots running. Both whitelabel. Both calibrated.

Sid runs at acquisition. Felix runs at retention. Both whitelabel, both calibrated per partner deployment, both built on the same Analyzer Engine.

@sid
live
acquisition
Sid
The discount engine that refuses to reward extraction.
Sits at first purchase or signup. Probes for alignment with the product through specific, calibrated questions. Discount tier earned, not granted. Bargain-hunters held at list price, warmly. Aligned customers welcomed at the deepest rate.
Moment First purchase
Strategy Paradox
Replaces Vouchers
See how Sid works →
@felix
live
retention
Felix
The loyalty layer that listens, then rewards generously.
Sits at redemption inside loyalty programmes. Reads occasion, mood, taste profile through warm dialogue. Members who share more, generously, end up with more. Score hidden from member; structured preference data flows to brand.
Moment Redemption
Strategy Variable reward
Replaces Static menus
See how Felix works →
engine

One engine. Many bots, all calibrated against it.

Underneath every bot
The Syntacy Analyzer Engine
Every BehaviorBot runs on the same five-axis behavioural fingerprint, the same anti-hack verification layer, and the same dialogic conversation architecture. What changes per bot is the moment it's pointed at and the weights applied to each axis for that specific job.
i
Five-axis Life History scoring
Long-term orientation, price focus, politeness, urgency, manipulation risk. Scored live by an LLM prompted as a behavioural psychologist. Hidden from the customer, surfaced as structured data to the brand.
long_term price_focus politeness urgency manip_risk
ii
Anti-hack verification
Detects copy-paste manipulation, performative politeness, and incongruence between language and metadata. Customers who try to game the bot end up with the same outcome as customers who arrive with no signal — a polite hold.
iii
Dialogic conversation architecture
Acknowledgement before next question. Adaptive follow-ups based on the prior reply. Graceful exit when signal goes hostile. Same shape across every bot, calibrated to the moment each occupies.
iv
Structured ZPD output
Every conversation produces a record of zero-party preference data — fields the customer knowingly volunteered, with provenance and consent flags. Flows to the partner's CRM. Owned by the partner.
botshop

More moments. More bots in development.

Sid and Felix are the first two BehaviorBots in production. More are being built in the botshop, each calibrated against a different commercial moment, sharing the same dialogic architecture and Analyzer Engine. Design partners considered for each.

In design · Q3 2026
Iris
The renewal conversation that asks honest questions.
Sits at subscription renewal. Reads whether the customer is genuinely engaged or auto-renewing through inertia. Surfaces churn risk early; surfaces upgrade signal where it exists.
Moment · subscription renewal
In design · Q4 2026
Otis
The complaint bot that turns escalation into insight.
Sits at the moment a customer signals friction. Reads the difference between a frustrated advocate and an extractive complainer. Routes accordingly. Captures product-defect signal as structured data.
Moment · service recovery
Concept · 2027
Rae
The referral bot that knows who's worth asking.
Sits at the post-purchase moment, after a customer has shown alignment. Reads whether they're likely to refer well, and if so, scaffolds the ask. No spray-and-pray referral programmes.
Moment · advocacy
partners / fit

Sectors where behaviour matters more than volume.

BehaviorBots are calibrated for sectors where the difference between a returning customer and a one-time discount-hunter materially affects margin. Less suited to high-volume commodity retail. Built specifically for premium hospitality, cultural sector, membership programmes, and considered-purchase categories.

Sector
Why BehaviorBots fit
Bots most relevant
Cultural sector
Audience cultivation matters more than ticket volume. Wrong-discounting an unaligned visitor leaks margin and produces no return.
Sid · audience qualification at booking. Felix · season-membership cultivation.
Premium hospitality
Brand promise depends on knowing the guest. Generic discounting erodes positioning. Insight at the booking moment compounds.
Sid · stay qualification. Felix · loyalty redemption cultivation.
Membership & rewards
Curation at scale needs preference data. Members redeeming silently produce no signal. Conversation is the unlock.
Felix · core retention layer. Iris (Q3) · renewal conversations.
Considered-purchase D2C
High average order values, high return rate sensitivity. Aligned-customer signal at first purchase is a leading retention indicator.
Sid · acquisition qualification. Rae (2027) · referral cultivation.
Wellness & education
Long customer journeys, repeat behaviour required. Routine-commitment signal at the front door predicts retention months out.
Sid · trial qualification. Felix · streak cultivation.
Status
Two bots live. More being built in the botshop.
BehaviorBots is what we're building at Syntacy: small, specialised conversational AI agents that bring behavioural-science depth to specific commercial moments. Sid and Felix are the first two. Walk through how each works, or get in touch if you want to design-partner the next one.
Meet Sid → Meet Felix → Talk to us →