1 · HERMES
Scriptorium
The corpus desk. Readings, catalog, and the works a course actually assigns — epic, tragedy, modern prose, or your own shelf.
A studio for human-centered AI
Human inquiry. Thoughtful AI.
AI consulting and scholarly tools for the people who teach, read, and ask the difficult questions.
The Hermeneutica workbench
Working preview · Sample textsReading Lab / $INQ-1-P1
On interpretation
An interpretation begins with attention. We return to a passage not only to find an answer, but to ask . The work of understanding .
Click a phrase to anchor it, then code it from the judgment rail.
Roles
Texts → Context → Evidence → Your interpretationOpen the external Workbench demo
Mark a phrase, code it, and write the note. The desk keeps your judgment in view. It does not call a live model.
Faculty-centered · Discipline-specific · Starting in New England
Humanities programsTeaching & learning centersLibraries & scholarly partners
The system
Four rooms.
One reading desk.
Use the desk above. Mark a phrase, code it, and ask a role. The argument takes shape from what you keep.
Hermeneutica is a system for reading with the text still in front of you. One desk holds the passage. Four rooms keep the corpus, the close reading, the context, and the argument attached to that passage. Named review roles may propose a note. Faculty decide what remains.
2 · Lab
Reading Lab
Close reading with coded passages, a codebook, and notes that stay attached to the line you can point to.
3 · Clio
Clio Explorer
Questions of person, place, and circumstance. Context is investigated and checked, not filled in for you.
4 · Graph
Argument Graph
A question, a source anchor, and an interpretation. The connection stays visible so a reader can disagree.
On the desk
- Readings
- Codebook
- Coded passages
- Quick note
- Cite
- DAG
- Executive summary
- PDF archive
Six named review roles sit under the reading — Bibliographer, Hermeneutic Reader, Ideology & Discourse Critic, Clio Archivist, Graph Architect, and Academic Web Harvester. They propose. Faculty decide. Outputs stay inspectable: an executive summary, a PDF archive, and the coded passage a colleague can open.
A thoughtful way forward
Start with a question.
Build what’s useful.
You do not need AI everywhere. You need the right support for the work that matters to your program.
Faculty training
Practical confidence, grounded in the methods and responsibilities of humanities teaching.
- LLM literacy and hands-on workshops
- Course and assignment design
- Academic integrity and evaluation
Custom AI solutions
Tools shaped around your texts, research questions, and ways of working. Not the other way around.
- Corpus-grounded LLM workflows
- Rule-based classification and checks
- Reading and annotation tools
Implementation & support
Move from an interesting demonstration to an evaluated workflow your faculty can use.
- Agents with human review checkpoints
- Integration scoping and pilot deployment
- Documentation, training, and handover
Built around your curriculum
Your syllabus.
Your questions.
Your way of teaching.
Connect the workbench to the work already happening in your classroom. Start with learning goals, then decide where AI belongs — and where it does not.
Explore the Reading Desk
External demo on a temporary tunnel. Availability may vary.
Build the habits of inquiry.
Introduce close reading, source-based questions, and critical AI literacy through a faculty-guided workflow.
- Close reading
- AI literacy
- Source awareness
Illustrative applications, tailored to your syllabus and learning goals.
How we work together
Small steps.
Considered progress.
Agree on the learning goals. Review the risks. Test the system before broad adoption.
01 / Listen
Understand the program
Map teaching priorities, faculty concerns, and the problem worth solving.
02 / Design
Choose the right system
Use rules, LLMs, or agents where appropriate. Define data boundaries.
03 / Pilot
Test in a bounded setting
Evaluate quality, source grounding, usability, and learning alignment.
04 / Equip
Leave your team capable
Train faculty, document decisions, and agree on support before expanding.
Assistance, not the end of interpretation.
Technology should make the work of inquiry more visible, not make the reader disappear. Source grounding, human review, and critical evaluation belong in the workflow from the beginning.
Let’s begin
Bring a course.
A challenge. An idea.
We’ll start with where AI could help — and where it should not. Scope and pricing are defined after discovery.
Do we need to adopt the whole workbench?
No. Training, solution design, and implementation can be scoped separately. Start with the need, then choose the tools.
What about privacy and academic integrity?
Start with institutional policies and data boundaries. Agree on permitted data, provider settings, human review, and evaluation before piloting. Review compliance requirements with your institution.
Is this only for New England colleges?
The initial focus is New England, with services intended for humanities education programs across American colleges and universities.
Is the workbench above a live AI tool?
No. You can mark phrases, apply codes, write notes, and sketch an argument here, but nothing calls a live model. Passages are original samples. Role prompts are preset. Your work stays in this preview and resets when it reloads. Use the external demo links for the separate product demonstrations.