When I say Tessera achieves 80-90% decision alignment in core domains, I want to be precise about what that means and what it does not mean.
What It Means
If you give Tessera an email chain that fully and accurately describes the problem, including context, constraints, stakeholders, and the signals I would normally key on, the likelihood that she responds the way I would, suggests what I would suggest, and requires only minor tweaks is approximately 80-90% in core domains.
Minor tweaks mean: adjusting tone for a specific audience. Tightening or loosening assertiveness. Adding one missing clarifying question. Reordering points for emphasis. Injecting a political or relational consideration that was never written down historically.
What I rarely need to do: change the core recommendation, reverse the risk assessment, or reframe the problem entirely. When that happens, it is usually traceable to missing input, not model failure.
The Domain Breakdown
In core operational and leadership domains, MSP operations, security posture decisions, escalation handling, vendor risk, technical leadership: 85-90%. In sales strategy and executive communication, where nuance and audience adaptation matter more: 75-85%. In novel or sparsely represented domains: 50-70%, with Tessera appropriately cautious.
The Key Condition
The phrase “completely describes the problem” is doing significant work. When the email chain captures the problem, the constraints, the stakeholders, and the decision boundary, Tessera performs at her best. She is not a mind reader. She is a judgment amplifier. Feed her the same inputs I rely on, and she usually lands where I land.
That distinction between mind reader and judgment amplifier is the entire design philosophy in one sentence.