AI Table

Dray connects multiple Gmail accounts to Claude, using sent mail and relationship context to draft replies in the user’s different voices.

A Mac browser with an agent that carries out plain-language tasks inside the tabs you already use.

A research project testing whether AI-simulated survey respondents can produce trustworthy public-opinion data.

A personal testing log that compares business AI tools through repeatable experiments, real costs, and practical judgments.

An agentic company-research API turns plain-English questions about a URL into typed JSON verdicts with quoted evidence.

Georg Zangl’s reference site argues that people have been quietly carrying context, authority, accountability, and institutional memory across the gaps in enterprise systems.

A weekly editorial experiment in which Gemini, Claude, and ChatGPT independently research AI news, form an opinion, and render that reaction as an image with a written rationale.

An FDCPA-aware AI workflow for law firms that drafts, dispatches, and audits consumer-collection communications.

A Russian-language iPhone recorder that analyzes conversations for repeated concerns, unspoken doubts, forgotten questions, decisions, and hidden risks.

A paid public experiment where 25 participants contribute words to one locked prompt, then compare how multiple AI models respond in fresh sessions.

A showcase and bring-your-own-key studio for generating short H3 Max text-to-video and image-to-video motion graphics.

A review layer that brings public evidence into view before a user trusts an AI-generated answer or decision.

A managed operations service for apps made with Lovable, Bolt, Cursor, v0, or Claude, covering monitoring, backups, patches, fixes, and written SLA commitments.

UDC builds real and simulated pharmaceutical factory environments where AI models propose formulations, operate equipment, and get graded on the resulting batches.

AtlasWithin presents a private, local-first AI assistant for daily management, software development, and owner-approved tasks.

A bilingual document-redaction workflow that proposes legally cited redactions, records each decision, and assembles an appeal-ready defense file.

A public venture journal and registry that records the work, accounting claims, and funding readiness of autonomous businesses.

ForgR measures AI-crawler visits, search visibility, indexing, and brand mentions across blogs it deploys.

A conversational interface that routes plain-language questions to connected providers and returns attributed answers in one thread.

Sealmatter describes a layer that replaces names, numbers, and addresses with consistent stand-ins before confidential documents reach external AI tools, then restores identities inside the customer’s environment.

Vestibrum presents Otoli, a planned modular smart sleep system that combines vestibular motion, sensors, and adaptive AI across bassinet, crib, and toddler-bed stages.

A UK-focused AI wedding planner that says it can find venues, track costs, read supplier replies, and prepare messages for couples to approve.

MedSumAI turns biomedical topics into AI-generated explanations linked to source articles in PubMed.