Farm to cup, one operating platform
Heart of Coffee Studio connects green intake, roasting machines, sensory QC, production, online shop orders, and shipping platforms with an end to end coffee ontology as the backbone and AI models at the heart.
Built on one ontology
One ontology, every workflow
From green lot intake to shipped parcel. Connected entities, machine adapters, and AI models help you discover, stabilize, and analyze your coffees.
Coffee ontology
Lots, roasts, profiles, sensory sessions, inventory, and commerce share one connected model. The same taxonomy powers traceability from origin to cup.
Ledger & Inventory
Track green and roasted coffee with a canonical ledger: receipts, roast consumption, yields, and packaging. Never ad hoc quantity columns.
Machine Integration
Ingest roast logs from Roest, Bellwether, espresso machines and more. Raw files preserved, curves, events, and inventory impact in the Studio.
Experiments & profiles
Run discovery roasts, link sensory sessions, promote winning profiles, and generate machine specific programs from roast targets.
Sensory & CVA
Descriptive, affective, and extrinsic cupping aligned with CVA. Aggregate scores by lot, roast, profile, or session.
Online shop sync
Pull customers and orders from your online shop. Segment D2C, retail partners, and enterprise accounts in one CRM view.
Shipping integrations
Sync parcel status, track shipments against orders, and close the loop from roast batch to customer doorstep.
Roaster Buddy
Graph RAG over your ontology: roasts, lots, knowledge docs, and operational data. Ask about KPIs, drift, and phase reports in natural language.
Roaster Buddy
A roast development decision system
Not an autonomous roaster but a "buddy" that structures hypotheses, cites your roasts and scientific literature, and proposes what to try next. You decide in the cup.
Buddy explains, proposes, and surfaces evidence. Operators approve, cup, and promote. Every write is confirm-gated.
Phase targets
Drying
38%
Maillard
44%
DTR
18%
| Milestone | Target | Band | Reference |
|---|---|---|---|
| Dry end | 4:20 | 4:05–4:35 | Cohort median (n=7) |
| FC start | 7:45 | 7:30–8:00 | Golden batch #4821 |
| Drop | 10:30 | 10:15–10:45 | Literature + HoC |
Citations: HoC roast #4821 · cohort n=7 naturals · Schenker & Rothkötter (2008)
Roast decision engine
Phase-by-phase explanations, deviation vs plan, and a sensory hypothesis for the next test, grounded in your roasts, not generic advice.
Generative roast plans
Literature, cohort bands, and similar-lot synthesis produce milestone tables and phase targets, operator-readable before you load the drum.
Citations & explainability
Every recommendation cites HoC roasts and papers with evidence grades. Buddy abstains when neighbors or literature are too thin.
Confidence & risk
Neighbor count, cohort n, and deviation from golden batch surface as confidence and risk badges. Know when to trust vs probe.
Lot intake intelligence
First roast on new green: comparable lots, process-family priors, humidity cautions, and a discovery sequence before production.
Similarity & divergence
Find tech-similar roasts and cup-similar profiles, and flag when curve twins taste different so you investigate before scaling.
Operator journey
From lot to production profile
Buddy sits on the development path, not on the roast controller.
- 01
Lot
Green intake, metadata, and intake brief for first-roast strategy
- 02
Discovery roast
Run experiments, ingest curves, link sensory, build the evidence base
- 03
Buddy proposes next
Cited phase targets, milestones, confidence, and sensory hypothesis
- 04
Promote to production
Operator approves in the cup, profile graduates to production roast
- 01
Lot
Green intake, metadata, and intake brief for first-roast strategy
- 02
Discovery roast
Run experiments, ingest curves, link sensory, build the evidence base
- 03
Buddy proposes next
Cited phase targets, milestones, confidence, and sensory hypothesis
- 04
Promote to production
Operator approves in the cup, profile graduates to production roast
Operating loop
Farm to cup, traced end to end
Lot → inventory → experiment → roast → sensory → profile → production → packaging → shipping → performance. One spine, no disconnected spreadsheets. Supported by data and AI loops.
AI loop
Graph RAG, sensory models, and performance insight. AI reads the loop and feeds decisions back.
Lot intake
Origin metadata, climate, geocoding, water and more
Inventory
Ledger receipts & storage locations
Experiment
Discovery roasts, candidate profiles and productization
Roast
Machine ingest, curves, phase events
Sensory
CVA sessions & lot scoring
Profile
Promote targets & machine programs
Production
Schedule & roast to stock
Packaging
SKUs, batches, roasted ledger
Shipping
Shop orders → carrier platforms
Performance
Coffee, roast, and sensory performance across the loop
Data loop
Machine ingest, ledger events, and ontology spine. Structured data flows in at every step.
Connected to your stack
Commerce, machines, and carriers
We're in beta. Launching soon
Book a walkthrough of HoC Studio: lots, roasting, sensory, commerce, and ontology aware Buddy AI. Existing beta users can log in anytime.