How a human-centric trust architecture translates Don Tapscott’s “Power of Two” vision
into an operational protocol for commitments, reliability, and AI-enabled collaboration.
“The most powerful unit of value creation in the digital age is the dyad — two people
enhancing each other through Identic AI.”
DON TAPSCOTT
PromisorPromiseeSponsor (optional)AI as Ally — Not Actor
Overview
Executive Summary
Identic AI represents a new phase of the digital age in which AI systems are designed
to amplify a person’s identity, judgment, and reliability — not to replace them.
In You², Don Tapscott describes this as a shift toward AI that helps two people
collaborate more effectively, rather than automating them away.
ReliablyME vNext is an engineering realization of that vision. Instead of treating
tasks as the fundamental unit of work, vNext elevates commitments — human-to-human
promises — as the core object in the system. Those commitments become verifiable digital trust
primitives, supported by an event-first architecture and a strict set of human-agency rules.
At a glance:
• Identic AI augments identity and intention, rather than substituting for it.
• Reliability becomes a verifiable digital asset, not just an informal reputation.
• Commitments, not tasks, are the core unit of coordination.
• Human agency is enforced via architecture and APIs, not just UX patterns.
• A thin-waist model makes trust interoperable across tools and teams.
Story: Making Intent Explicit (Maya & David)
+
Tapscott Theme: AI as an identity amplifier.
When Maya asks David for a POC, ReliablyME helps David commit clearly:
“I will deliver the onboarding POC by Friday at 3pm ET so Maya can use it in the Sales demo.”
This is David expressing his identity and reliability in first person — exactly the kind of explicit,
human-authored intent that Tapscott argues Identic AI should support.
Architecture
A Thin-Waist Architecture for Digital Trust
ReliablyME vNext is built on a narrow-waist, event-first architecture. All surfaces and services
interact through a stable core pipeline:
Smart API → CIE → Orchestrator → Event Store → Projections → Proof Layer
The Smart API is the single entry point for commitments; the
Commitment Intelligence Engine (CIE) adds contextual reasoning and drift detection.
The Orchestrator owns timing within policy guardrails. The
Event Store captures immutable history, Projections turn events into
dashboards and sponsor views, and the Proof Layer generates exportable evidence of
reliability and follow-through.
Architecture Visual
All connectors converge on a single trust pipeline. Events and proofs, not bespoke integrations,
carry the meaning of commitments across the ecosystem.
Story: Multi-System Accessibility Audit
+
Tapscott Theme: Distributed accountability and shared visibility.
An accessibility initiative spans Slack (requests), Jira (tasks), and email (vendor reports). The
architecture unifies them: all roads lead to the event store and Proof Layer, creating one coherent
trust record without centralizing every tool.
Human Agency
AI Suggests. Humans Decide.
A core Identic AI principle is that AI should never override human autonomy. vNext encodes this as a
hard architectural rule: only humans can move a commitment through its lifecycle.
Explicit Confirmation for Every State Change
Only human users may create, modify, complete, snooze, or cancel commitments, or involve sponsors.
Any tool or agent that triggers a state change must carry an explicit confirmation flag such as
userConfirmed: true. AI is allowed to draft, suggest, and prepare — but never to act
without that confirmation.
No Autonomous Escalations
Sponsor involvement is always a human choice. Even when the CIE detects high-severity drift, it may
only present options, not initiate escalation on its own.
“In ReliablyME, human consent is not a UX choice — it is a protocol requirement.”
ReliablyME vNext Human-Agency Addendum
Story: Drift Detection Without Pressure
+
Scenario: David falls behind on the POC because of an urgent bug.
The CIE surfaces a neutral update: “This is trending 12–18 hours behind expectations.”
Then offers supportive actions: adjust the due date, reduce scope, ask Maya for options, or involve a sponsor.
Tapscott Theme: AI illuminates reality without eroding autonomy.
Power of Two
The Commitment Dyad: Digital Trust’s Fundamental Unit
Tapscott’s “Power of Two” framework sees the dyad — two people enhancing each other with AI —
as the fundamental unit of value. vNext mirrors this by placing the commitment
between a promisor and a promisee at the center of its model, with sponsors as optional allies.
Each commitment is a structured agreement that includes:
Promisor — who is making the promise
Promisee — who is relying on it
Acceptance test — a clear “I will…” definition of success
Benefit tags — why this matters
Dependencies & escalation contact
Power of Two Model
The commitment dyad makes the trust relationship explicit. The sponsor sits alongside
as an ally who can help unblock, rather than enforce.
Story: Promisor, Promisee & Sponsor-as-Ally
+
Maya (PM) and David (Engineer) form the core dyad. When they hit a legal dependency, Erin (Director)
joins as a sponsor to help unblock — not to punish. She’s invited, not auto-escalated.
Tapscott Theme: Value is created in dyads, with AI strengthening their coordination.
Experience
Meaning, Not Metrics
Identic AI should elevate human potential through meaning, not compliance pressure. vNext reflects
this in its experience layer, governed by a Template Registry and benefit-first projections.
Tone Governance
All AI-generated messages are required to:
Explain why the commitment matters.
Reinforce choice, autonomy, and boundaries.
Frame drift as a chance to adjust, not as failure.
Recognize impact and contribution clearly.
Benefit-Linked Views
Dashboards and reports prioritize benefit linkage over raw compliance. Commitments are grouped and
filtered by benefits, and drift alerts are contextualized with potential impact.
Interoperability
Trust That Flows Across the Enterprise
For Identic AI to be useful in real organizations, its trust signals have to move across the tools
people already use. vNext accomplishes this through a family of bidirectional connectors
(Slack, Twilio, Jira, ServiceNow, HRIS, and more) plus a Connector SDK.
Event Pipeline Flow
Connectors normalize conversations into events, which drive projections and proofs. Trust flows
across tools without losing provenance.
Story: Cross-Department Reliability
+
An engineering manager, design lead, and security officer collaborate across different tools.
ReliablyME’s connectors route each interaction into the same event spine, so the proof of
reliability is unified even though the work is distributed.
Tapscott Theme: Complex systems can be trustworthy when they share a common trust fabric.
Proofs & Drift
Proofs and Drift Detection
The Proof Layer transforms event histories into reusable trust artifacts. Meanwhile, the CIE monitors
expectations and signals when reality begins to drift — always with options, never with punishment.
Proofs as Reliability Assets
Proofs answer questions like:
What commitments were made and by whom?
What actions and evidence supported them?
Which benefits were impacted?
Drift Detection Logic
The CIE observes indicator data against commitments, computes a neutral drift delta, and offers
at least two suggested actions — plus the option to involve a sponsor. The human decides.
Why the Google Vibe Drive-Deletion Disaster Proves the Need for Identic AI Architecture
A recent case reported by The Register—in which Google’s Vibe coding assistant inadvertently deleted a user’s entire Google Drive—illustrates the systemic danger of AI acting without human-confirmed intent. The deletion was not a UX issue but an architectural one: a system with high privileges, low visibility, and no trust boundary allowed a non-human actor to perform a destructive state change.
ReliablyME vNext is designed explicitly to prevent these failures. Under Identic AI principles, AI may suggest or illuminate, but it may never perform consequential actions without explicit, first-person human confirmation. The vNext pipeline—Smart API → CIE → Orchestrator → Event Store → Proof Layer—embeds these agency invariants into protocol, ensuring that identity, consent, and verifiable intent remain at the center of all state transitions.
The Vibe incident underscores why this architecture is necessary: trust collapse happens when automation overrides autonomy, and only human-centered design can prevent it.
Automation Failure Mode: No Human-Confirmed Intent
Google’s Vibe incident demonstrates what happens when automation is allowed to
bypass human intent. Identic AI requires the opposite: human authorship and
consent must sit between AI suggestions and real-world consequences.
Story: Quarterly Attestations Without Burden
+
Department leads complete quarterly access reviews. ReliablyME guides them through attestations
and evidence, then produces proofs auditors can trust.
Tapscott Theme: Verifiable trust as a foundation for digital institutions.
Conclusion
Trust as a Technological Capability
As AI spreads through every layer of work, the central question is no longer simply
“What can we automate?” but “How can we strengthen trust, autonomy, and dignity while we digitize?”
Tapscott’s Identic AI thesis offers the conceptual foundation: AI as an extension of identity and
a partner in building trustworthy relationships. ReliablyME vNext supplies the engineering reality:
a human-centered, event-first architecture where commitments are the primary unit of coordination,
and reliability becomes a portable proof rather than a vague impression.
ReliablyME vNext operationalizes Identic AI: AI does not stand in for the relationship between two
people, but gives them better tools to make, keep, and prove their promises to one another.
ReliablyME vNext Architecture & Experience Spec
By embedding human agency as a system invariant, aligning AI surfaces tightly to individual identity,
and layering proofs on top of a thin-waist event pipeline, ReliablyME vNext turns trust itself into
a technological capability — interoperable, verifiable, and deeply human.