Skip to content

Who you are. What AI should say. How AI behaves.

Brand Definition tells SI who you are. Reusable Knowledge tells SI what to say. Mandates tell SI how to behave — when to reach out, in what tone, with what boundaries, and what lines to never cross. Set the rule once, in plain language. SI applies it at runtime — in every email, every call, every chat, every SMS — adjusted to the specific contact, channel, and moment.


Layers 3 governance tiers
Channels Email · Voice · SMS · Chat
Application Runtime, per-contact
02

You built your reputation on judgment. Now you need a system that honours it.

You know what to say, when to say it, and — critically — what should never be said. The tension isn't whether to automate. It's whether any system can carry the weight of your professional standards.

Every intelligent system you've considered forces the same tradeoff. And the fear is not irrational — it's the correct response to systems that don't let you encode your standards into their behaviour.

  • An automated email goes out omitting a required disclosure. You don't catch it until a client does.

  • Your system sends a "thinking of selling?" message to someone who just lost a family member. The algorithm saw a life event. It didn't understand what kind.

  • A nurture sequence crosses the line from stewardship to solicitation. The distinction matters in your profession. The system didn't know it existed.

  • A follow-up to your most valued client sounds nothing like you. It's polished, efficient, and completely wrong — because it was written by a system that never learned how you think.

  • Your advisor network sends non-compliant language you didn't catch until a regulator did. The system scaled beautifully. It also scaled the risk.

  • An automated check-in references a client's condition in a way that feels invasive. The intent was care. The impact was the opposite.

These aren't edge cases. They're the predictable result of systems that treat intelligence and governance as separate concerns. You shouldn't have to choose between a system that's smart but unpredictable and one that's predictable but rigid.

THE TRADEOFF YOU'RE FORCED TO MAKE ? Smart but unpredictable You spend your time reviewing its output Predictable but rigid Every message sounds like a template NEITHER OPTION REFLECTS YOUR JUDGMENT

The real question isn't which tradeoff to accept — it's why the tradeoff exists at all.

Smart systems

Fill in blanks with assumptions that don't reflect your standards

Rigid systems

Flatten your voice into something no one recognises as you

Why the old approaches don't hold

Each of these methods worked in a simpler context. None were designed for a system that thinks for itself.

Control method

Manual review

Works at five contacts. At five hundred, you spend more time reviewing output than the automation saves you. The bottleneck moves — it doesn't disappear.

"I'll just review everything before it goes out" — that works until it doesn't.

Personalisation method

Templates

Same instruction applied to every contact regardless of who they are, what stage they're in, or which channel they're on. The result is communication that feels generic to every recipient.

A template doesn't know your contact just complained. It doesn't know their role changed. It treats everyone the same.

Automation method

Workflows

Workflows fire the same sequence regardless of context. They can't adapt to a contact who just complained, a stakeholder who changed roles, or a relationship that shifted from warm to cold.

Sequences don't listen. They execute. And when the context changes, they keep executing anyway.


The consequence of all three: professionals either over-control — and lose the scale benefit — or under-control — and risk their reputation. Neither is sustainable as the contact base grows and AI capabilities expand.

"AI is smart enough to figure it out" — AI is capable, but it doesn't know your professional ethics, your industry's unwritten rules, or the specific sensitivities of your relationships.

The old approaches don't hold because they weren't designed for a system that thinks for itself. They were designed for tools that do what they're told.

04

The tradeoff between intelligence and control was never inevitable

It only exists when governance is embedded inside the engine. Separate the layers, and the false choice dissolves.

Every platform that tries to make AI "safer" does it the same way: they limit what the engine can think about. Restrict the topics. Narrow the vocabulary. Reduce the reasoning surface. The result is rigid, generic output that sounds like it was written by committee — because it was. The committee just happens to be a set of hard-coded constraints.

Let the engine think freely, and you get the opposite problem: unpredictable output that might say anything to anyone. Every automated decision your platform makes carries your name — and without precise directives governing what it says, when it says it, and what it never does, you're trusting your reputation to a system that doesn't know your boundaries.

Mandates resolve this by operating as a separate architectural layer. SI keeps its full reasoning capacity. It still learns every contact, still adapts to context, still makes intelligent decisions about timing, channel, and message. But every decision it makes runs through a layer of directives the professional has set.

You write a rule in plain language. The platform converts it into machine-readable instructions. SI reads those instructions at runtime — at the moment of every decision — and respects them without exception. No prompt engineering. No code. No configuration screens. No tradeoff between capability and control.

"Your role isn't to manage every gear shift. Your role is to set the destination. Mandates are how you set the destination."

From the Driver's Seat to the Command Center
Layer 01 — Professional Judgment

Your intent, expressed in plain language

You define what SI should prioritise, what it should avoid, and the boundaries it must never cross. No technical knowledge required — write it the way you'd brief a trusted colleague.

Layer 02 — Mandate Layer

Directives, compiled and enforced at runtime

Plain-language rules are converted into machine-readable instructions. SI reads them at the moment of every decision — before any action is taken, any message composed, any channel selected.

Layer 03 — Segmentation Intelligence

Full reasoning capacity, fully governed

SI retains its complete ability to learn contacts, adapt to context, and make intelligent decisions about timing, channel, and message. Nothing is restricted — everything is directed.

Layer 04 — Execution

Autonomous action within your boundaries

Every email, message, call, and content decision reflects both SI's intelligence and your professional judgment — simultaneously, without compromise.

5

Six Dimensions of Governance

Every mandate falls into one of six categories. Together, they describe every governance decision a professional would ever want to encode — covering not just what SI does, but how, when, and what it will never do.

Cadence

When & How Often

Governs → timing, frequency, channel sequencing

Fiduciary advisor: No past client receives more than two touchpoints per week. Email goes first — if the second touch goes unanswered, SI switches to voice.

Flow

Tone & Structure

Governs → message structure, opening patterns, style

Consultant: Emails to CFOs always lead with data and ROI. Emails to founders lead with story and vision. Same intelligence, different voice.

Hard Fact

What Is True

Governs → non-negotiable truths, credentials, scope

Lawyer: The firm only practices in two jurisdictions and never accepts contingency cases. Every email, call, and chat respects those boundaries without exception.

Prohibition

What Is Forbidden

Governs → absolute boundaries, compliance walls

Insurance principal: No reference to specific return percentages or guaranteed outcomes across the entire advisor network — the prohibition holds even if a client asks directly.

Objection Handler

How to Respond

Governs → pushback strategies, friction responses

Mortgage broker: When a past client raises rate concerns, SI acknowledges first, shares market context next, then offers a conversation — never leads with a refinance pitch.

Comparative

How to Position

Governs → competitive positioning, differentiation

B2B principal: Competitor names are never mentioned in nurture sequences. In proposal contexts, specific pre-approved differentiators can be referenced — nothing else.

6

Write it like you'd say it

You write the rule the way you'd explain it to a colleague. The platform converts your plain-language directive into machine-readable instructions — silently, underneath. You never see the technical layer. You see the rule you wrote, enforced everywhere SI operates.


New Mandate
Parsed to structured logic
Mandate active — enforced across all channels
Realtor

"Never use the phrase 'great investment opportunity' in any communication — say 'a property worth considering' instead."

SI enforces the substitution across every email, SMS, and call script.
Coach

"When a contact has gone silent for 90 days, the next outreach must reference something specific from our most recent conversation — never a generic check-in."

SI scans the conversation history and references it accordingly.
Lawyer

"Every email to past estate planning clients must include a sentence acknowledging that the law in their state may have changed since their plan was finalised."

SI inserts that framing in every relevant communication.
Accountant

"Between May and December, never lead with tax-related language — lead with advisory or planning language."

SI shifts its opening framework seasonally.

If you can articulate the rule, the system can enforce it. The gap between professional judgment and platform behaviour disappears.

Explore Mandates →

The Governance Hierarchy

Mandates are not a flat list of rules. They are a layered governance architecture with clear precedence — from brand-level defaults to contact-level overrides. The system resolves conflicts automatically. You never think about hierarchy.

L1

Brand

Global behaviour across all wares, all audiences, all channels. The foundation everything else inherits from.

Always sign off emails with my first name, never the company name.
L2

Ware

Behaviour specific to one offering. Overrides brand defaults when the context narrows to a particular product or service.

When discussing the Enterprise plan, emphasise dedicated support and SLA guarantees.
L3

Audience Segment

Adapts messaging for a specific persona. The system speaks differently to different people — because you told it to.

When engaging realtors, use real estate terminology and reference MLS data.
L4

Pipeline Phase

Governs behaviour during specific journey moments. Different stages demand different cadences, different tones, different urgency.

In the Proposal Sent stage, follow up within 48 hours if no response.
L5

Channel

Channel-specific formatting and constraints. What works in email doesn't work in SMS. The system knows the difference.

SMS messages must be under 160 characters and never include links.
L6

Contact Status

Safety boundaries that override everything. No exception, no override, no edge case. When this layer says stop, the system stops.

DNC contacts receive no calls and no SMS — email only, with explicit opt-in.

The more specific rule wins.

When two mandates conflict, the system applies the more specific one. A ware-level mandate overrides a brand-level default. An audience-segment mandate overrides the ware. A contact-status rule overrides everything else.

You write the rule for the context you care about. The system applies the hierarchy automatically. You never have to think about precedence — the architecture thinks about it for you.

08

Every Surface. Every Decision. Every Time.

Mandates are not applied to the output. They are part of the input. SI never generates a word, selects a channel, or times a message without the active mandate framework loaded first.

TONE COMPLIANCE BOUNDARIES POSITIONING FREQUENCY CONTENT PROHIBITED CADENCE Phona Email SMS Live Chat Reply Proc. SI CORE
Phona — AI Voice

Mandates load into the call context before the first word is spoken. Phona adjusts cadence, tone, prohibited topics, and competitive positioning in real time — even when the conversation goes off-script.

Governed by active mandates
Email Sequences

Active mandates are loaded before the message is generated. The email respects directives from word one — tone, content boundaries, and positioning are baked in, not filtered after.

Governed by active mandates
SMS Messaging

Mandates control tone, content, length, and frequency. Every text message SI sends is shaped by the same governance framework that governs a voice call or a long-form email.

Governed by active mandates
Live Chat

Every response in a website chat session is governed by the mandate framework. The AI doesn't drift — it operates within the same boundaries whether it's the first message or the fiftieth.

Governed by active mandates
Reply Processing

When a contact replies to any communication, mandates shape how SI interprets the response and formulates the next action. The governance loop is continuous — inbound and outbound.

Governed by active mandates
Scenario

A health practitioner writes a mandate prohibiting diagnostic language in automated communications. The next morning, SI generates a wellness check-in email — gentle, supportive, devoid of medical assertions. That afternoon, Phona conducts a follow-up call — the AI voice stays within the same boundary, even when the client asks a direct medical question. The evening's SMS reminder respects the same constraint. Same mandate. Three surfaces. One enforcement.

Mandates govern SI. SI governs Phona, Email, SMS, Live Chat, and every execution surface. One rule cascades through every capability.

09

Built for Your World

The architecture is universal. The governance patterns are yours — designed for the compliance boundaries, relationship rhythms, and communication norms your profession demands.

Every profession carries ethical red lines. Mandates encode them as hard boundaries — not guidelines that drift, but rules that SI enforces across every touchpoint, every time.

Declare the boundary. SI enforces it everywhere.

Fiduciary Advisors

"Never reference specific fund performance or imply guaranteed returns in any client communication."

Private Practice Lawyers

"Prohibit direct solicitation of represented parties. All outreach must be informational, never advisory."

Insurance Networks

"Enforce network-wide compliance: no comparative policy claims, no competitor disparagement across all advisor communications."

Health Practitioners

"Never reference diagnoses, treatment outcomes, or clinical details in automated outreach. Sensitivity-first language only."

Advisory Accountants

"Prohibit tax advice in nurture sequences. Advisory content must carry appropriate disclaimer language."

Real Estate Professionals

"All market communications must include brokerage attribution and comply with local fair housing language requirements."

Same architecture, different rules. Every profession uses the same Prohibition and Constraint mandate types. The content of those mandates — the actual boundaries — comes from your world, not ours.

Relationships don't move at the same speed in every profession. A mortgage renewal cycle looks nothing like a coaching engagement arc. Mandates define the cadence — SI maintains it without drift.

Define the rhythm. SI maintains it.

Real Estate Professionals

"Sphere cadence: quarterly market updates, home anniversary touchpoints, seasonal check-ins timed to listing cycles."

Mortgage Brokers

"Renewal window cadence: re-engagement begins 6 months before 3-5 year terms expire. Referral partner rhythm: monthly."

Advisory Accountants

"Year-round advisory cadence: bridge the tax-season gap with quarterly planning touchpoints and mid-year financial check-ins."

Coaches & Consultants

"Long-cycle nurture: 6–24 month engagement arcs. Thought-leadership touchpoints every 2–3 weeks. Re-activation after programme completion."

Fiduciary Advisors

"AUM retention cadence: quarterly portfolio context, annual review scheduling, life-event triggered outreach."

Health Practitioners

"Continuity-of-care rhythm: post-visit follow-ups, seasonal wellness reminders, re-engagement for lapsed patients at 90-day intervals."

Your calendar, encoded. Cadence mandates use the same Directive and Constraint types. The timing, triggers, and intervals reflect the natural rhythm of your professional relationships.

How you sound to a past client is different from how you sound to a referral partner. Mandates govern the voice for each audience — so SI adapts its tone without losing your identity.

Define how you sound. SI adapts.

Real Estate Professionals

"Market-first communication: lead with neighbourhood data and local insight. Personal warmth, never salesy. Sphere gets casual; leads get professional."

Coaches & Consultants

"Thought-leadership voice: challenge assumptions, share frameworks. Prospects get authority; existing clients get partnership tone."

Advisory Accountants

"Advisory voice: proactive, strategic, never reactive. Tax-season comms are operational; off-season comms are consultative and forward-looking."

Private Practice Lawyers

"Client stewardship voice: measured, authoritative, never presumptive. Existing clients receive proactive updates; prospects receive educational content only."

Health Practitioners

"Continuity-of-care voice: warm, empathetic, clinically appropriate. Active patients get supportive; lapsed patients get gentle re-engagement."

B2B Key Accounts

"Stakeholder-specific voice: executive sponsors get strategic; operational contacts get tactical. Never mix registers across the org chart."

One brand, many registers. Voice mandates use Persona and Directive types to govern tone by audience segment. Your identity stays consistent — the register shifts to match the relationship.

"My industry is too unique for a generic governance system."

Mandates aren't generic templates. They're your rules, stated in your language, for your context. A fiduciary advisor's mandates look nothing like a realtor's, and both look nothing like a health practitioner's. The six mandate types and the attachment hierarchy are the architecture. The industry-specific patterns are the application. You don't start from scratch — you start from governance patterns designed for your world and refine from there.

10

Trust the system. It follows your rules.

Every objection you're forming right now — we've heard it. Here's why the architecture answers each one before you finish the thought.

"I'll just review everything before it goes out."
That works at five contacts, not five hundred. Mandates let you define the boundaries once — then every interaction is governed without your manual review. You set the standard. SI enforces it at scale.
"AI is smart enough without boundaries."
AI is capable — but it doesn't know your professional ethics, your industry's unwritten rules, or the promises you made to specific clients. Mandates give intelligence the context it can't learn on its own.
"Mandates will make the system too rigid."
Mandates don't limit intelligence — they focus it. SI retains full flexibility within the boundaries you set. Think of it as a riverbank: the water still flows freely, but it goes where it should.
"Once I set mandates, I'm stuck with rigid rules."
Mandates are living directives — updated in seconds, governing every future interaction from that moment. Your governance evolves as fast as your business does. Nothing is permanent except your control.
Runtime enforcement Mandates aren't a feature bolted on — they're an architectural layer that SI reads at runtime. Every interaction is governed. No exceptions.
No sampling. No gaps. Every contact, every channel, every message. There is no after-the-fact validation because governance happens before execution — not after.
Instant propagation Update a mandate and it takes effect immediately across your entire network. No migration. No retraining. The system simply follows the new directive.

Governance you can prove.

For regulated professionals — fiduciary advisors, lawyers, insurance networks — mandates create more than control. They create a documented audit trail of governance decisions.

Every active mandate, every enforcement, every boundary respected is logged. This isn't just governance — it's demonstrable proof that you took proactive steps to ensure compliant, ethical communication at scale.

Explore Mandates
  • Mandate enforced Compliance disclaimer appended to all outbound financial communications — 2,847 interactions governed this month.
  • Boundary respected SI blocked performance guarantees in prospect outreach — 14 instances caught and corrected before delivery.
  • Directive updated New regulatory language mandate activated — propagated across all channels in under 3 seconds.
  • Audit trail logged Complete governance history exportable for compliance review — every mandate, every enforcement, timestamped.

Every mandate you write makes the system more precisely yours

Brand Definition says who you are. Reusable Knowledge says what to say. Mandates say how to behave. Together, they make Segmentation Intelligence an extension of your judgment — operating at scale, never crossing the lines you've drawn. The more precisely you define your standards, the more confidently you can let the system run.

Layer 1
Brand Definition
Layer 2
Reusable Knowledge
Layer 3
Mandates
Segmentation Intelligence — governed actions
See Mandates in Action

No commitment. See the system with your own context.

06

Straight answers

No. Write a directive the way you'd explain a rule to a colleague. "Never mention competitor pricing." "Always open with the client's first name." The platform converts your plain-language instruction into machine-readable governance. No configuration screens. No workflow builders. No technical background required.

Mandates don't limit intelligence — they focus it. A prohibition against mentioning competitor pricing doesn't make Segmentation Intelligence less capable. It makes it more precise. SI retains full flexibility within the boundaries you set, finding the optimal path through every interaction while respecting your non-negotiables.

Mandates are living directives. Write a new rule in seconds and it governs every future interaction from that moment. Retire an old one just as fast. Your governance grows with your practice — no migration, no re-training, no waiting for changes to propagate.

The attachment hierarchy resolves conflicts automatically. More specific mandates override more general ones. Contact-status mandates always win. You don't have to manage precedence — the system applies it. A brand-wide tone directive yields to a channel-specific instruction, which yields to a contact-status rule. The architecture handles the logic so you never have to.

No. Every decision Segmentation Intelligence makes — every word, every channel selection, every timing decision — is made within the mandate framework. There is no ungoverned state. Even before you write your first custom mandate, brand-level defaults ensure SI operates within your established voice and values.

Yes. Brand-level mandates govern everyone on the platform. Ware-level, audience-level, and channel-level mandates can be scoped as needed for specific teams or functions. Contact-status mandates protect individual relationships regardless of who's using the platform. The entire hierarchy scales from a solo practitioner to a distributed network.