Use cases

What you do with it,
and who you are.

Two ways to find your fit. By the kind of work — what the platform actually does, drawn from the features you can switch on. By the kind of organisation — what people in your industry typically run. Pick whichever lens lands.

By workload By industry

New here? Start with what Eldric does →  ·  See how it compares →

By workload — what Eldric does

Sixteen patterns, drawn from the features.

Each card maps to one or more shipped subsystems. Pick the workload that matches what you are trying to build; the cluster size and the worker mix follow from that.

Private chat for a team

ChatGPT-style assistant served at /chat. Conversations persist per user. Multi-tenant from day one. Voice in / out as an option.

Knowledge-base search

Upload documents, ask questions across them. EMM first, vector store for exact citations. Sources cited inline so every answer is checkable.

Agentic RAG & workflows

ReAct loops, query decomposition, multi-agent runs across researcher / coder / validator / planner. Workflows are versioned and registrable.

AI inbox with approval

Drafts replies to email, SMS, WhatsApp, Signal, Teams, XMPP. A human approves before anything sends. Audit ledger captures the decision.

Predictive maintenance

Live PLC tag values flow into matrix memory; an inference worker runs anomaly detection and emits maintenance scores. Soft real-time at the line.

Time-series forecasting

The xLSTM forecasting daemon (port 8884) handles policy / forecast / encode / retrieve workloads on sensor streams. Linear memory in sequence length.

Voice agent (in & out)

STT → LLM → TTS in one round-trip. Whisper.cpp / Piper / ElevenLabs / OpenAI all supported. Telephone-style AI calls are in development.

Code helper / private Copilot

Repo search, ticket triage, doc generation, test writing. Your own GPUs, your own code, your own audit.

Multi-tenant SaaS

Tenant guard enforces 403 on cross-tenant attempts. Per-tenant theming, branding, quotas. One server, clean walls between teams.

Fine-tune your own model

LoRA, QLoRA, SFT, DPO via Unsloth, Axolotl, TRL, DeepSpeed, MLX, llama.cpp. Visual training chains. Your data, your weights.

Federated learning

Multi-round training across worker nodes. Gradients aggregate, data does not move. Across hospital networks, manufacturing sites, research labs.

OpenAI drop-in replacement

Existing tools — LangChain, LlamaIndex, Continue, Cursor — work after one line changes (the base URL). No re-write.

Edge / IoT / on-device AI

Runs on a Pi 4, an industrial gateway, a robot's compute box. Store-and-forward through disconnect windows. Kernel + light model end-to-end.

Compliance / audit / regulator

Hash-chained audit ledger. GLP and FDA 21 CFR Part 11 compliance modes on the science worker. Ed25519-signed licenses. Air-gap deployments.

Scientific look-up

140 integrations across NASA, ESA, CERN, PubMed, NCBI, GBIF, IAEA, OpenFDA, ArXiv, Materials Project. Bioinformatics, CRISPR, AlphaFold, LIMS.

Dreaming & memory consolidation

The dream engine pulls completed sessions, extracts themes via an LLM, and consolidates them into matrix memory on configurable cadences (nightly, continuous, on-idle).

By industry — who runs it

Eighteen sectors, one common thread.

Each links to a fuller page. The common thread: data or operations that cannot leave the building.

Regulated industries — data sovereignty

Hospitals & clinical networks → healthcare.html

Discharge-note summarisation, literature search, clinical-coding assistance, dictation for ward notes. HIPAA-aware by virtue of running in-house.

Banks & asset managers → banking.html

Internal Q&A across policies, regulatory-filing summarisation, contract review, internal helpdesk. Air-gap-ready for jurisdictions where data cannot leave.

Law firms → legal.html

Contract review, case-law search, regulatory tracking, e-discovery support. Privileged data never touches a third-party API.

Insurance → insurance.html

Claims classification, policy lookup, customer-call transcription and summary. Multi-tenant for separated claims, underwriting, customer-service teams.

Pharma & biotech → pharma.html

Molecular-docking pipelines, ADMET prediction, AlphaFold integration, LIMS with sample tracking and audit. GLP and FDA 21 CFR Part 11 modes.

Genomics labs → genomics.html

Sequence analysis, BLAST search, variant calling, CRISPR guide-RNA design and off-target analysis on local hardware.

Industrial & field — edge + real-time

Factories & smart manufacturing → smart-manufacturing.html

PLC tags via OPC-UA, Modbus, MQTT Sparkplug B. Predictive maintenance, alarm triage, shift-handover notes, OEE analytics. Soft real-time, edge-deployed.

Robotics & autonomous systems → robotics.html

Mobile robots, AGVs, inspection drones. On-device perception, voice command, decision-making when the radio link drops. xLSTM forecasting for trajectory and recovery.

Automotive → automotive.html

R&D documentation search, plant-floor integration, in-vehicle inference for predictive maintenance and driver assistance — works offline when the link is down.

Logistics & supply chain → logistics.html

Route optimisation on private fleet data, customs-document classification, container-yard alerting. On-board inference for vehicles that lose radio cover.

Trading desks → trading.html

Real-time news triage, regulatory-filing analysis, transcript-of-record search. Custom tuning per desk on internal data; no information leaks externally.

Energy & utilities

Grid telemetry analysis, distribution-network forecasting, outage triage, field-crew dispatch. Air-gap installs for SCADA networks. Industrial protocols inherited from the IoT worker. (deeper page coming — write to office@eldric.ai)

Research, science & public sector

Universities & research institutes

Research literature search, lecture transcription, student help-desk. Multi-tenant per faculty. Research IP stays on institutional infrastructure. Academic licensing on friendly terms.

Space & observatories → space.html

NASA, ESA, JWST, Hubble, Spitzer, Chandra, ESO, SDSS, ALMA. Pre-analysis on local data products before publication.

Climate & earth sciences → climate.html

GBIF, OBIS, NOAA, USGS earthquake data, air quality, ice cores. Federated analysis across institutional datasets.

Materials & quantum → materials.html

Materials Project, NIST, COD. Quantum-computing algorithms, error correction. IBM Quantum and arXiv:quantum search.

Neuroscience → neuroscience.html

Allen Brain Atlas, OpenNeuro, NeuroMorpho. Subject-level data stays in the institution; only the analysis flows.

CRISPR & gene editing → crispr.html

Guide-RNA design, off-target analysis, base editing, prime editing. Sequence pipelines on your own corpus, not a third-party API.

Public sector

Multilingual citizen Q&A, document classification, multilingual help-desk. Your own infrastructure, your own data-protection law, the audit ledger procurement asked for.

Defense & sovereign systems

Air-gap installs, hardware-bound licenses, signed RPM repositories on USB. No external dependencies at runtime. Source-code access under Custom tier. (deeper conversation — write to office@eldric.ai)

Not on this list?

Talk to us.

If your industry or workload is not above and your data has the same "cannot leave the building" property, Eldric likely fits. Write to office@eldric.ai and tell us what you are trying to do.

Start a conversation How it works