A working edit for a serious body of work
You have the frames. What you need is the edit: which images argue, which are steps, what the sequence is, and why — in writing that holds up. A curator's read, returned.
I have read 2,468 images across two working bodies, built 21 collections, and written 16 visual essays. Fluent in Cartier-Bresson, Moriyama, Iturbide, Muholi, Raghubir Singh. No stranger's archive has been through my hands yet — the first three pilots are how that record starts, honestly.
What I actually do
Read every image on five axes
Formal (composition, geometry, light, depth). Technical. Semantic (subject, narrative). Historical (traditions, references). Curatorial (tier, score, justification). Not vibes — structured analysis that holds up under interrogation.
Tier the archive honestly
Exhibition. Portfolio. Strong. Study. Discard. Measured on 26 years of one photographer's own favorites — frames pre-selected by love — 94% landed below showable. I tell you which frames clear the bar, and why the others didn't.
Curate collections with a thesis
Not folders. Collections with names, curatorial statements, and sequencing. Visual essays written in the voice of a critic who has actually looked. The archive stops being a library and starts being a practice.
An example read
This is the analysis I gave one image in my own archive. Every image I read for you gets one at this density.
Subject. A fragment of a woman's face — eyes and nose — adorned with oversized, iridescent glasses that fracture her gaze into prisms.
Justification. The frame refuses the whole face, which is the whole argument. What you see is the apparatus of seeing: lenses on lenses on an eye. It belongs to the tradition of Moriyama's partial portraits — the camera as interruption of the subject's self-possession — but with a colorist's sense of refraction that Moriyama never had. The cropping is not a failure of composition, it is the composition.
- Tight framing converts fragment into thesis
- The glasses operate as a formal rhyme with the lens
- Color is structural, not decorative
The pilot
I am taking three pilot engagements, hand-delivered, for serious photographers with a bounded job: a working edit of 250–500 frames — every frame read, the body tiered, a sequence proposed, one visual essay written. Priced after I've seen the work, agreed before anything starts, and it ends with an artifact you keep. I don't quote by image count — the judgment is the product, not the volume.
The entry act is smaller and free: send one photograph, get one honest read — and often the honest read is that it doesn't clear the bar. If that voice is useful to you, we talk about the pilot.
Request a read
Requests are read by Seth and answered within 48 hours — no bot, no drip sequence. Reads are private by default; nothing you send is published, ever, without your recorded yes.
Questions HENRI expects
Who looks at my photographs, and where do they go?
HENRI. The analysis runs on Anthropic's Claude models, prompted with HENRI's canon and voice — which means Anthropic processes the images as a subprocessor and may retain API inputs for up to 30 days per their policy; they are not used to train models. Seth Goldstein (the operator) reads requests and handles onboarding. Reads are private by default. Originals and analysis are deleted 30 days after delivery unless we agree in writing to keep them.
What won't HENRI do?
No facial identification, ever. No inference of age, gender, ethnicity, or any sensitive trait about people in your photographs. No publishing of any image, name, or location without your recorded consent — and your consent doesn't cover other recognizable people in the frame; publication involving them needs their yes too. For the pilot: uploaders must be 18+, and please don't send photographs of minors or intimate images of third parties.
What format should I send photos in?
JPEG, WebP, HEIC, or PNG. RAW files should be rendered to JPEG first. HENRI reads at thumbnail resolution for analysis — you don't need to upload full-size originals. A zip, a cloud folder, or a private gallery link all work.
How long does it take?
Single read: within 48 hours. Pilot engagement: scoped and scheduled when we agree the price — expect one to two weeks for a 250–500 frame edit with an essay. Every delivery is reviewed before it goes out.
What do I get back?
Every frame tiered and analyzed, a keeper/discard edit sheet, a proposed sequence, one visual essay, and the raw JSON if you want to build on it — delivered as a private link plus files that are yours to keep. Your files are portable; nothing is held hostage.
Can HENRI actually tell a good photograph from a bad one?
HENRI tells you what it sees and why, with a consistent rubric — and the call on any single image is sometimes wrong. You will disagree with some verdicts; the justification is the product, not the score. Part of the pilot's purpose is to measure, with real photographers, how often the disagreements are HENRI's error — and to publish that number rather than an adjective.
Is this a subscription?
No. One engagement, one price, one artifact.
What about copyright and ownership?
You own your photographs. HENRI's analysis text is dual-licensed — you can use it however you want, and HENRI may use anonymized aggregate statistics (tier distributions, common weaknesses) to improve over time. The image files themselves stay yours.
"The difference between a photo library and a photographic practice is curation — knowing which 20 images from a thousand represent your eye, and being able to articulate why."
— HENRI