Conference affiliation: Isadora & Co | The Bloom House AI · 2026
On this page
Isadora Martin-Dye is the founder of The Bloom House and Isadora & Co, and the owner of Rixey Manor, the Virginia wedding venue that inspired her approach to customer-facing AI. Her products address a deceptively consequential failure: assistants that sound warm and authoritative while inventing facts, mishandling sensitive circumstances, or misrepresenting their identity.
From wedding venues to AI products
Born in England, Martin-Dye studied at the University of Exeter and worked in film, television, talent management, and production in Los Angeles. In 2014, she bought a foreclosed Virginia manor and restored it; Rixey Manor opened as a wedding venue the following year.
Running the venue exposed the limitations of fragmented customer information and generic automation. Couples appeared across inboxes, booking platforms, contracts, and scheduling tools, while conventional assistants struggled to preserve the venue’s voice or recognize when a personal crisis demanded a different response. Martin-Dye began developing The Bloom House and Ground in 2025 and established Isadora & Co in 2026.
Bloom combines unified couple records with venue-specific response drafting, marketing attribution, operational intelligence, and an AI concierge. Staff can review outgoing messages, and their edits help shape subsequent responses. Ground applies similar continuity to personal organization through an AI companion.
The safeguards behind her assistants
Four-layer assistant architecture. Martin-Dye structures behavior around immutable identity rules, situational context, example-anchored brand voice, and a post-generation veto. Consolidating previously scattered prompts into an ordered assembly process separates nonnegotiable constraints from contextual judgment and stylistic preferences.
Honest identity and contextual sensitivity. Bloom assistants identify themselves as AI immediately and cannot imply they can personally conduct venue tours or meet customers. When someone is grieving or caring for an ill relative, responses can adapt without repeating sensitive details; coordinators receive different framing from couples.
Deterministic output safeguards. Before an answer reaches a customer, checks can reject fabricated dates, prices, policies, or promises. Martin-Dye favors predictable pattern-based enforcement while acknowledging its narrower coverage, and identifies centralized validation and context resolution as architectural improvements. Her AI Engineer presentation also emphasizes strict tenant isolation: missing venue identity should halt a response instead of borrowing another brand’s configuration.
Threadline and high-stakes uncertainty. Martin-Dye is developing Threadline, a free public utility for researching missing-person and cold-case information. Its restrictions prohibit presenting speculative connections as proof that someone has been identified or found—a safeguard calibrated to the extraordinary emotional consequences of false certainty.
