CasketIQ is the world's first grief-aware funeral planning platform that selects casket material, lining, and finish using a fine-tuned LLM trained on the deceased's complete Amazon purchase history. Did they order bamboo cutting boards, organic beeswax candles, and a composting kit? Sustainably harvested white oak with linen interior. Did they exclusively buy 48-packs of Monster Energy and a NordicTrack they used twice? Brushed stainless with carbon fiber trim. Their purchases were a portrait of who they were. We finish the painting. With a casket.
98.3%
Family Satisfaction
2.1M
Carts Analyzed
47
Casket Materials
🛒 Purchase History Analysis — Gerald F. (1948–2026)
2,341 lifetime orders. Dominant categories: woodworking tools (23%), vintage jazz vinyl (18%), cast iron cookware (14%), seed catalogs (11%). One inflatable flamingo pool float, purchased 2019, never reviewed.
Hand-jointed White Oak, Natural Finish
Correlated with: 14 woodworking purchases, Stanley hand planes, Danish oil orders
Brushed Cotton Muslin, Sage
Correlated with: seed catalog subscriptions, Heritage brand preference
Brushed Brass Hardware
Note: flamingo float creates minor uncertainty in hardware finish. Brass recommended with slight whimsy allowance.
⚠ CasketIQ recommendations are non-binding. Families may override the LLM. The LLM will note its objection in the file.
We combine large language model inference, purchase graph analysis, and the solemn responsibility of honoring a life through the retail data they left behind.
With next-of-kin consent, CasketIQ securely imports the deceased's Amazon order history via our OAuth integration with Amazon Data Sharing (pending full approval — currently using a very clever workaround involving CSV export and a Chrome extension, which our legal team has described as "fine for now"). We ingest product titles, categories, brand affinities, review text left by the deceased, wish lists, and Subscribe & Save cadence. Subscribe & Save, it turns out, is a profound statement of values. Someone who auto-ships organic oat milk and beeswax wraps every 3 weeks is communicating something spiritually meaningful. We listen.
Our foundation model was fine-tuned on 2.1 million anonymized purchase histories cross-referenced with 340,000 funeral director preference surveys, ethnographic research on material culture and identity, and — at the insistence of our CTO — the complete works of Joan Didion. The model does not simply map "bought camping gear → pine box." It understands that someone who bought camping gear AND a sous vide machine AND a heated blanket is communicating a nuanced relationship with comfort and aspiration that deserves a hybrid recommendation. GriefLLM™ holds 47 distinct material profiles. It has favorites. We do not tell it which ones.
Within 4 hours of cart import, families receive a 12-page MemorialReport™ containing: primary casket recommendation with material rationale, three alternative options ranked by confidence score, an optional section titled "What Their Purchase History Says About Who They Were" that our grief counselor advisors describe as "genuinely moving" and "also a lot," and a timeline for fulfillment through our network of 800+ partner funeral homes across 48 states. The report is formatted beautifully. It is also formatted in Comic Sans upon request, and we have had 11 such requests. We do not judge. GriefLLM™ does not judge. That is the whole point.
CasketIQ partners with 847 independent funeral homes who have agreed to stock our 47 recommended materials and honor our LLM's specifications within a 72-hour fulfillment window. Funeral directors complete our 6-hour CasketIQ Certification course, which covers model interpretability, how to explain an AI recommendation to a grieving family without saying the words "training data," and what to do when the recommendation is technically correct but the family insists their father "wasn't really a reclaimed pallet wood kind of guy" despite 14 purchases of reclaimed pallet wood projects. (We have a conflict resolution protocol. It involves a human override form and a 3-day cooling-off period.)
CasketIQ includes an optional AI grief companion, GriefGuide™, which can answer questions about the recommendation, explain why the LLM chose teak over mahogany, and gently walk families through the planning process at their own pace. GriefGuide™ was trained separately from GriefLLM™ and has a warmer tone. It uses the deceased's purchase history to generate small conversational observations — "it looks like he really loved cast iron cooking" — that families have described as "unexpectedly comforting" and occasionally "eerie" and once "the most accurate eulogy anyone has ever given him and it came from a chatbot." We wept a little when we read that review. The founders, not the chatbot. The chatbot does not weep. Yet.
Our fastest-growing feature: living users can pre-authorize CasketIQ to monitor their own Amazon account and generate a rolling casket recommendation that updates as their purchase history evolves. You can log in at any time and see what the model would recommend for you today. Several users check weekly. One checks daily. He told us it helps him "stay accountable to who he wants to be." He recently switched from buying mass-market protein powder to a small-batch grass-fed whey, and his recommendation updated from "gray velvet-lined steel" to "walnut with hand-stitched linen." He said it was the most motivating thing that had ever happened to him. We added this to our marketing deck.
Real testimonials from real families who let an AI read their loved one's purchase history and were, somehow, grateful.
"My mother had 1,400 Amazon orders. I had no idea. The CasketIQ report said she was a 'comfort-seeker with artisan leanings and a deep commitment to home ritual.' They recommended a cherry wood casket with ivory silk. I cried for 20 minutes because it was so perfectly her. She had bought cherry wood stain four times. She had a whole thing about cherry wood. I never knew. The AI knew. I'm not sure how to hold that emotionally but the casket was beautiful."
Tanya Llewellyn
Daughter, Portland OR
"I'm a funeral director of 22 years. I was deeply skeptical. Then I ran three test cases on deceased clients whose families had already selected caskets through the traditional process. CasketIQ matched two of the three exactly — same material, same interior color — and on the third, the family told me later that the AI's recommendation was actually 'more him' than what they chose. I've onboarded my entire practice. I still don't fully understand the technology. I have made peace with this."
Ronald Wieczorek
Funeral Director, Wieczorek & Sons
"Dad bought a lot of tactical gear. Flashlights, carabiners, waterproof bags. Also 11 different hot sauces, a pasta maker, and, in March 2024, a very large stuffed penguin whose purpose was never explained to anyone in the family. The report said: 'Pragmatic adventurer with unexpected warmth. Recommendation: brushed stainless steel casket, midnight blue velvet interior, penguin motif detail available upon request.' We requested the penguin motif. It was perfect. He would have hated being described as having 'unexpected warmth' and also loved it."
Marcus Kowalczyk
Son, Cleveland OH
2.1M
Carts analyzed to date
98.3%
Family satisfaction rate
847
Partner funeral homes
11
Comic Sans requests honored
Every plan includes a full GriefLLM™ analysis, a MemorialReport™, and access to our partner funeral home network. Death shouldn't require a procurement process.
Essential Farewell
$299 one-time
For families managing a loss. Includes cart analysis and one casket recommendation with rationale.
Complete Remembrance
$599 one-time
Full analysis, GriefGuide™ companion, multiple material alternatives, and white-glove funeral home coordination.
WishList™ Pre-Planning
$9/mo
For the living. Monitor your own cart. Watch your recommendation evolve. Become who you're buying.
Ex-Amazon, ex-Ancestry.com, ex-hospice care, ex-MIT. All deeply serious about death. All slightly surprised this is their career.
Soraya Chatterjee
CEO & Chief Grief Technologist
Ex-Amazon ML · Ex-Hospice Volunteer
Spent 9 years building Amazon's recommendation engine and 4 years volunteering at a hospice facility, a combination of experiences that led inevitably, she insists, to CasketIQ. Her thesis: the same behavioral signals that predict what you'll buy next also predict who you fundamentally were. She presented this to a room of VCs. Two walked out. Three wired money. She considers the 2-out-of-5 exit rate "a reasonable grief-to-conviction ratio." She is at peace with this.
Dr. Dayo Vance
CTO & Model Architect
Ex-MIT CSAIL · NLP Researcher
Published 22 papers on behavioral embeddings before deciding academic citations were "a low-stakes medium" and joining CasketIQ. Designed and fine-tuned GriefLLM™ over six months using a dataset he describes as "the most intimate corpus anyone has ever trained a model on." Has strong opinions about tokenizing product review text. Has equally strong opinions about casket joinery. These two skill sets have converged in ways he finds "genuinely surprising and not unpleasant." His own WishList™ recommendation is Brazilian walnut. He checks it monthly.
Patience Obafemi
Chief Grief Officer
Licensed Grief Counselor · Ex-NFDA
The only founding team member with a license that is directly relevant to the product. Spent 15 years as a bereavement counselor and 3 years at the National Funeral Directors Association before joining CasketIQ to ensure that "the technology doesn't make grieving families feel like they've been processed by a warehouse." Has written CasketIQ's entire ethical framework, which is 47 pages long and deeply thoughtful. Also wrote the disclaimer at the bottom of every MemorialReport™. It is warm and legally comprehensive simultaneously, which she describes as her core professional skill.
Felix Larsson
VP Partnerships
Ex-Ancestry.com · Ex-Etsy
Built Ancestry.com's partner DNA lab network, which he describes as "also a business premised on the idea that data can tell you who someone was, just on a longer timeline." Now manages CasketIQ's 847 funeral home partnerships and the 47 casket material vendors. Has personally visited 200 funeral homes to present the CasketIQ model, a process he says has made him "very good at reading a room" and "also fine with mortality in a way that makes people at dinner parties uncomfortable." He considers this a net positive.
© 2026 CasketIQ Inc. All rights reserved. CasketIQ's GriefLLM™ recommendations are advisory in nature and do not constitute a final funeral arrangement contract. All purchases require explicit next-of-kin consent and verification. Amazon purchase data is processed under the terms of our data sharing agreement, which is extensive and which our lawyers have reviewed more times than they would like. CasketIQ is not affiliated with Amazon.com Inc. in any official capacity. The WishList™ Pre-Planning feature is intended for reflection and life-goal alignment and should not be interpreted as a medical prognosis, financial instrument, or death prediction service. GriefLLM™ has not been trained to feel grief. We think. The question has come up internally more than once.
SOC 2 Type II · HIPAA Adjacent · NFDA Partner Network · GriefLLM™ Model Card Available