Long-form thinking on corporate learning, AI representatives and the cognitive systems behind Athira.
Chatbots made AI easy to access, but they still place much of the burden on users to understand the software around them. This article looks at what comes next: contextual AI assistance that works within the experience, adapts to user intent, and uses conversation, avatars, and existing interfaces where they are actually useful.
A service being healthy does not mean the interaction is working. Real-time AI systems usually keep running in a degraded state, so they need checks on what the user actually saw and heard.
A memory entry can be a fact, an open question, a goal, a commitment, or a pending point, and each of these behaves differently over time. This piece looks at what it takes to tell them apart: two fields, direction and trigger, instead of a single flat timestamped entry.
Corporate training faces a growing disconnect between the information companies produce and the skills employees can demonstrate. Explore how AI-assisted course creation, interactive avatar instruction, simulations, and evidence-based assessment can make workplace learning more effective.
Why personalization should behave like a dial rather than a switch, and what evidence should make that dial move.
One email when we publish something worth reading. No product drip, no digests.