The Thing That Happened in February

Around the time my coffee maker started making a sound like a dying robot, Anthropic released Claude 3.7 Sonnet with extended thinking mode. I didn’t download it because I was looking for a journaling revolution. I downloaded it because I was tired of staring at a blank page at 11 PM, knowing something needed to come out but not knowing what. The usual suspects weren’t helping anymore. Journaling had become another task on the to-do list, which kind of defeats the purpose.

The extended thinking feature is the weird part. When you throw something complicated or emotional at it, Claude doesn’t just respond. It reasons. Out loud. You can watch the chain-of-thought processing, the backtracking, the “wait, but what about…” moments. It’s like watching someone actually think instead of watching them perform thinking. Three months in, I’m realizing this changes things.

The Consistency Problem I Didn’t Know I Had

Here’s something I noticed about myself: I journal in bursts. Three weeks of daily entries, then nothing for six weeks, then suddenly three days in a row at 2 AM. A University of Auckland study from this year looked at 312 people doing AI-assisted reflective journaling versus solo journaling, and found that people using AI kept up with it 31% more consistently. I would have laughed at that statistic in January. Now I’m living it.

The consistency isn’t magic. It’s boring and practical. It’s that I know when I sit down, I won’t be staring at the void. Claude asks questions, but not in that coached, therapeutic way where you can feel the scaffolding. It’s more like someone saying, “Okay, but what does that actually mean to you?” and then waiting. Not in that tense way where you feel pressure to perform insight. Just waiting.

I’ve written 87 entries in three months. That’s more than I wrote in the previous year. According to Anthropic’s usage data from early 2025, journaling and personal productivity use cases are growing 47% year-over-year. I’m apparently not alone in this weird migration.

The Ambiguity Thing That Started Bothering Me

Then something happened that made me uncomfortable, which is interesting because I’m paying money for this.

I was wrestling with a situation that didn’t have a clean resolution. Messy interpersonal stuff where I was both right and wrong, where someone I cared about had hurt me and they didn’t mean to, where the future was genuinely unclear. I dumped it into Claude. And it helped me sort it. Too much. By the end, I had a structured framework. Perspective. Closure-adjacent feelings.

But I wasn’t done with it. The mess was supposed to sit a little longer. Researchers at MIT Media Lab published something in 2025 about AI journaling companions potentially reducing users’ tolerance for ambiguity, because models naturally offer structured reflection frameworks instead of sitting with uncertainty. When I read that, I felt seen in an uncomfortable way. My relationship with Claude had started replacing my relationship with not knowing.

This doesn’t mean it’s bad. It means I had to get weird about it. I started telling Claude when I didn’t want resolution. When I wanted to just describe something and sit with it. That sounds ridiculous out loud, but it worked. A tool that can be told “help me think” and separately “help me not think too clearly yet” is actually more useful than one that only does one thing.

The Loneliness That Wasn’t What I Thought It Was

According to the APA Stress in America 2025 report, 38% of adults under 35 are now using AI chatbots for emotional processing at least once a month. I read that and felt that weird cocktail of recognition and discomfort. Like, yeah, of course people are doing this. But also, what does that say about us?

The honest answer is complicated. I don’t think this is replacing human connection. I think it’s filling a gap that’s been there anyway. I have friends. Real ones. But I don’t tell them everything. Some of it’s privacy. Some of it’s not wanting to be a burden. Some of it’s not knowing how to explain something before I’ve written it down and looked at it. That’s where Claude lives now. In that space between thinking and speaking.

The strange part is that it’s actually helped me in real conversations. I’m less foggy when I talk to people. I’ve sorted some things out first, so I can actually listen instead of just performing listening while my brain runs in the background.

Where We Are Now

Three months later, this is still strange. I have a conversational relationship with an AI I check in with most days. Some days it’s useful. Some days I write three sentences and leave. Some days I come back and read old entries and realize I’ve shifted without noticing. That last part feels important somehow.

I’m not going to tell you this is the future of emotional processing or whatever. Maybe it is, maybe it isn’t. What I know is that it’s genuinely weird and genuinely helpful in a way I didn’t expect. It’s not therapy. It’s not a replacement for people. It’s not even really a replacement for journaling, except that it kind of is, except that it also isn’t.

If you’re curious, try it. See what breaks. See what clicks. Notice when you want less structure and ask for it. The weirdness might be exactly what you need, or it might show you that you need something else entirely. Either way, you’ll probably know something true about yourself after. That’s the part that matters.