For the past several months, we've been having a different kind of conversation. Not with vendors or investors first, but with the people doing the work: retreat leaders, integration therapists, ketamine clinic directors, facilitators, and the experiencers moving through their own arcs of care. We asked them one direct question: where, if anywhere, does AI belong in psychedelic-assisted therapy?
The answers were more consistent than we expected and more honest than most vendor conversations tend to be. Almost no one we spoke with was anti-technology. What surfaced again and again wasn't fear of AI itself. It was fear of substitution.
The fear isn't AI — it's substitution
Facilitators described watching administrative software promise to support their work, only to begin creeping into some sacred parts of the job. One retreat director put it simply. She didn't want a tool that tried to become a wiser version of her. She wanted a tool that gave her back the hours she was losing to scheduling and follow-up so she could be fully present when it mattered most.
Beneath the specifics, the same concern surfaced in nearly every conversation: that the industry would mistake automation for care, much as the broader healthcare system already has. Several facilitators pointed to what's happening across mental health today: overwhelming caseloads, clinician burnout, and the gradual replacement of human contact with chat interfaces. Psychedelic work asks people to loosen their grip on identity, memory, and meaning at the exact moment they are most open to suggestion. Anyone inserting themselves into that space, including a piece of software, carries real responsibility for what they say and how much authority they're given to say it.
That's the part we wanted to sit with honestly rather than argue away. The instinct to keep AI out of the room is a good one. Where we differ is in believing that AI never has to be in the room to be useful.
Upstream of care, never inside it
So we built Asendar around a single, non-negotiable principle: the technology sits upstream of care, never inside the therapeutic encounter itself. It doesn't sit across from anyone during a session. It doesn't interpret what surfaced during someone's ceremony. It never decides what an experience meant.
Instead, it supports the structure around the entire arc of care. It coordinates the people involved in intention, immersion, and integration. It surfaces when a check-in is overdue. It carries the practical weight of preparation and follow-up so facilitators can spend less time managing logistics and more time doing the one thing technology cannot do: being fully present with another human being.
You don't need a shaman in your pocket
We say this to every team we work with, and we'll keep saying it here: you don't need a shaman in your pocket. You don't need a companion app auditioning to become your therapist, your guide, or your interpreter between sessions. What you need is something closer to scaffolding: invisible when it's working, present enough to catch what would otherwise fall through the cracks, and never claiming authority over someone's inner experience.
The therapist stays the therapist. The integration coach stays the integration coach. The witness stays the witness. Asendar's job is to make sure each of them has the time, the signal, and the continuity to do that work well, not to do it for them.
Who owns the record
The other thing practitioners told us, almost without exception, was that they don't trust where their data goes once it leaves the room. That distrust is earned. Mental healthcare has a long history of treating deeply personal information as an asset to be owned rather than a story to be stewarded.
We built Asendar the other way around. Every action is consent-gated and auditable. The people whose healing is at stake retain sovereignty over who sees what. We don't own that record. We safeguard it on their terms.
None of this is a finished argument. The field is still being built, and the norms surrounding AI in psychedelic care are being written in real time. That's exactly why it's worth getting them right now rather than trying to fix them later.
If there's one thing this round of conversations confirmed, it's that the practitioners closest to this work already know what they want. They want fewer things competing for their attention and more time to give it fully. They aren't asking AI to become wiser, warmer, or more present than they are. They're asking it to get out of the way long enough for them to be exactly that.