Research & conversation
Ask Lois, retrieve scientific records, continue conversations and generate session reports. Keep the evidence available to the next step in your application.
Connect a source, computation or instrument once. The research floor, REPL and compatible agents can all use it without losing provenance, limits or failure behavior.
Connect questions, sources, evidence, runs and reports through one research system. Build around records your team can inspect at every step.
Ask Lois, retrieve scientific records, continue conversations and generate session reports. Keep the evidence available to the next step in your application.
Organize projects into Minds. Read graphs, inspect individual records and sources, follow the ledger, and import or export research data.
Import PDF, DOCX, CSV, TSV, text, Markdown and JSON. Connect supported archives, collect source excerpts and review aggregation runs. Uploaded excerpts remain source material until a claim is supported.
Give a Mind an objective, select sources and follow a stored plan. Inspect progress, approve a review step, retry blocked work or cancel a run. Scheduled workers advance queued work after you leave.
Retain notes and observations, inspect the research journal and prepare sourced manuscript drafts. Unsupported statements are withheld; hypotheses and researcher notes remain working material.
Track hypotheses, experiments, verdicts and scientist runs. Inspect health, diagnostics, available libraries and session reports to understand what ran and what remains unresolved.
The dashboard uses authenticated account operations. Each request is checked against the signed-in user before research data is accessed. These application routes are not a general API-key service.
Visitors can exercise the research workflow with state carried by their browser tab. Sign in to retain work in an account. Demo access does not grant persistent account operations.
The local runtime exposes scientific instruments to the terminal and compatible MCP clients. Configure the runtime and its integrations for your environment; hosted browser sessions are a separate access surface.
The REPL + API plan describes per-key usage visibility and credit-based limits. It is currently on the waitlist. Arrange access and confirm supported operations before building a production dependency.
Scientific records retain database identity, accession or identifier, source URL and retrieval details. Inspect these alongside the result. Missing evidence, unavailable sources and failed runs remain explicit outcomes. A source citation makes a claim traceable; it does not make it independently validated.
Hosted floor operations require an account session. Device bearer tokens are restricted to pushing observations and checking health; they cannot read a Mind or run general research operations. Science API-key access belongs to the separate API offering.
The current upload interface accepts files up to 1.25 MB. Research sources have their own availability and request limits. Durable runs expose progress, review and cancellation. Confirm API credits, quotas and integration scope for your pilot.
Drag to pan. Zoom in. Select a node to see what it does. Keyboard: Tab between nodes, Enter to inspect; arrow keys pan when the diagram is focused.
Explore the live demo, then create a free account to retain your research. Minds, conversations, evidence and notes give each project a place to continue.
Use the local terminal workflow and connected scientific tools. The advertised paid REPL and API plans remain on the waitlist; confirm access and deployment requirements with the team.
Research credits are product usage units. They cover retrieval, computation and model-assisted work; they are not language-model tokens. The billing implementation records actual model token usage separately. Buying credits adds research capacity, not evidence or guaranteed discoveries.
These rates come from our billing implementation, which has not shipped on this website. They are a preview, not an active checkout or a promise of current account allowances.
| Operation | Credits per unit |
|---|---|
| Read your saved graph, ledger and conversations / write notes | 0 |
| REPL recall or trace | 1 |
| Extract / research / import one accession | 5 |
| Propose or revise a hypothesis | 5 |
| Register an experiment | 10 |
| Run a scientist experiment | 25 |
| Research chat turn | 25 |
| Compose an answer | 5 |
Free: 200 · Pro: 5,000 · REPL: 30,000 · REPL + API: 75,000
The published plan prices are $0, $20, $100 and $200 per month respectively. Paid plans are not open yet.
$10 / 2,000 · $50 / 12,000 · $200 / 60,000
At the draft rate of 25 credits per research chat, 2,000 credits covers 80 such turns if used only for that operation. Other operations consume the same balance at their own rates.
Explore the product guides and supported research workflows. Contact the team for the integration documentation relevant to your project.
Explore DocsScientific adapters return structured, source-pinned results. The Science API offering is described on the plans page; availability and access should be confirmed before building a production dependency.
Explore APIConnect compatible clients to the scientific research server through Model Context Protocol. Tool results retain the same provenance requirements across clients.
Explore MCPPlan how Perslis Science fits into your environment: your sources, connected tools, access requirements and deployment preference. Discuss project-specific details with the team.
Explore ArchitectureUse the released artifacts and public papers as the starting point. Contact the team for SDK or repository access; this page does not imply that a private codebase is publicly available.
Explore SDK / GitHubA recorded tour of the scientific tools and research workspace.
Recorded on a research prototype. The recording shows the scope demonstrated at that time.
Explore the research floor yourself →Start with a target, a dataset or a research question. Inspect what the evidence supports, what it rejects and what remains unknown.