AI Assistant
Talk to your IoT fleet. Your AI Assistant does the work.
OpenAI-powered agent built into the Datacake platform. Ask, analyse time-series data, build dashboards, manage rules and fix decoders, all in natural language with deep function-calling.
- 7 specialised skills: devices, analytics, rules, dashboards, decoders, workspace, general
- Page-aware: knows the device, dashboard or rule you are viewing
- Confirmation prompt before every change
- 1M tokens free per account, no credit card
Your IoT copilot, built into Datacake
The Datacake AI Assistant is an OpenAI-powered AI agent built into the Datacake IoT platform. It answers natural-language questions about your devices, analyses time-series sensor data, builds dashboards, creates automation rules and troubleshoots payload decoders. The Assistant lives in a slide-out panel inside the workspace. Open it, type a question, and the agent translates your prompt into the right Datacake API calls. It is not a generic chat bot bolted on top: every skill is wired into the same data and operations you use in the UI, so you can move from “what does my fleet look like?” to “save this dashboard” in a single conversation.
See pricing and the free tierAI Assistant for IoT, built for trust
1M tokens free per account, seven domain-specific skills, and zero conversation data ever touches Datacake servers. Here is the trust model in numbers.
1M
Free tokens per account
No credit card required. Tracked in the panel’s usage bar.
7
Specialised skills
Devices, Analytics, Rules, Dashboards, Workspace, Decoder & Debug, General.
0
Conversations on Datacake servers
Threads live in your browser’s IndexedDB. We never log or persist them.
Why a built-in agent, not a generic chat bot
Direct function-calling into Datacake APIs
Every skill calls the same GraphQL APIs the UI calls, not a brittle plugin layer. Tool results are real records, not best-effort summaries.
Built for time-series, not just text
The Analytics skill computes statistics over your sensor history (averages, rankings, time-above-threshold) instead of dumping raw data into the prompt.
Page-aware out of the box
No prompt engineering to teach it which device, dashboard or rule you mean. It picks up the context from the page you are viewing.
Production-grade write safety
Every change waits for an explicit Confirm with a plain-language summary. The Assistant cannot mutate your workspace silently.
Explore devices
Find and inspect any device by asking
Stop hunting through filters. Describe what you are looking for and the Assistant runs the right queries against your workspace: single devices, groups by tag, fleet-wide lookups. It can also fall back to custom GraphQL queries when the built-in tools are not enough.
- “Show me all devices in Building A”
- “What fields does this device have?”
- “Find all devices tagged with cold-chain”
- “What is the current battery level of my temperature sensors?”
Analyse your fleet
Time-series analytics in plain English
The Analytics skill is built for averages, rankings, outliers, trends and historical analysis across your sensors. Natural-language sensor mapping included: “how warm is it” maps to temperature, “air quality” maps to CO₂/VOC, “noise level” maps to loudness. The Assistant computes statistics over your time-series data instead of dumping raw history into the chat.
- “What is the average temperature across all my sensors?”
- “Which 5 devices have the lowest battery?”
- “How often was CO₂ above 1000 ppm last week in the office?”
- “Compare humidity levels between Building A and Building B”
Build & edit dashboards
Describe a dashboard, watch it appear
Tell the Assistant what you want to see and it assembles the dashboard: adding widgets, choosing chart types, configuring thresholds and rearranging the layout. All changes land in a draft first; nothing is saved until you confirm. Works for workspace dashboards and per-device dashboards alike.
- “Build a dashboard with temperature and humidity for all my sensors”
- “Add a line chart of the last 24h of power consumption”
- “Add a value widget for battery with a green/red gauge”
- “Rearrange so the charts are full-width and remove the old humidity widget”
Create & manage rules
Spin up automation rules from a sentence
Describe the alert you want and the Assistant proposes the rule, including whether it should run as an efficient product-level rule or a targeted device-level rule. Inspect, update or remove existing rules the same way, and confirm before any change is written.
- “Create a rule that emails me when temperature exceeds 30°C”
- “Show me all my active rules”
- “Update this rule to also trigger on humidity above 80%”
- “Delete the offline-notification rule”
Troubleshoot decoders & debug logs
Fix data ingestion without leaving the chat
The Decoder & Debug skill reads recent debug logs, inspects the payload decoder for a product, runs the decoder against the last received payload and helps you fix parsing errors. Because decoders are shared across all devices of the same product, the Assistant always reminds you of the impact before saving.
- “Why isn’t my device sending data?”
- “Show me the recent debug log for this device”
- “Test this decoder with the last received payload”
- “Help me write a LoRaWAN decoder for a 4-byte temperature/humidity payload”
Seven specialised AI skills for IoT operations
General
Broad IoT and Datacake questions, basic device lookup, getting-started orientation.
Devices
Deep device exploration: fields, current measurements, tags, custom GraphQL queries when needed.
Analytics
Fleet statistics, rankings, outliers, historical analysis and trend detection on time-series data.
Rules
Create, inspect, update and delete automation rules at the product or device level.
Dashboards
Build and edit workspace and device dashboards iteratively, widget by widget, with a draft step.
Workspace
Organise folders and tags, manage products and claim devices into the right workspace.
Decoder & Debug
Read debug logs, inspect and fix payload decoders, test changes against real uplinks before saving.
Page-aware context
It already knows what you are looking at
The Assistant is page-aware. Open it on a device, dashboard, rule or the devices list and it picks up the context automatically, with no need to repeat IDs, names or filters. The current page context is shown in a small bar below the panel header so you always see what the Assistant sees.
- On a device page: “Show me the current temperature” works without naming the device
- On a dashboard: “Add a humidity chart” edits the dashboard you are viewing
- On a rule page: “What does this rule do?” inspects exactly that rule
- On the devices list: search, filter and explore without repeating context
Production controls: confirmation, reasoning effort, code interpreter
Confirmation before every change
Write operations (creating rules, saving dashboards, updating decoders, modifying folders) always show a plain-language summary and wait for an explicit Confirm or Deny. Read operations execute automatically.
Auto skill selection & switching
The right skill is picked from your first message and switched on the fly when the topic changes. You can also pick a skill manually from the input pills.
Reasoning effort control
Switch between Low, Medium and High reasoning effort. Use Low for quick lookups, High for multi-step analyses where you need the model to think things through.
Text verbosity control
Tune response detail from brief, to-the-point answers up to thorough explanations with extra context, per thread, on demand.
Code interpreter
Optionally let the Assistant write and execute code to process data, run calculations or produce formatted outputs that go beyond the built-in tools.
Threads & token usage
Every conversation is a thread you can switch between. Token usage for the current thread is shown in the panel so you always know where you stand.
Free to try, predictable to scale
Every Datacake account starts with a generous free token allowance, enough for most users to explore the Assistant extensively. When you are ready to make it part of daily workflows, the paid plan is a flat per-user fee with a virtually unlimited allowance.
1,000,000 tokens, free on every account
No credit card required. The free allowance is tracked automatically across all conversations and shown in the panel’s usage bar.
Paid plan: 20 EUR per user / month
Activate the paid plan to keep using the Assistant beyond the free tier. Includes a virtually unlimited token allowance for regular, productive use.
Fair use, no surprise throttling
The paid plan is built for everyday productive use. Excessive or abusive consumption is rare under normal use, and we always communicate with you before taking any action.
1M free tokens are included on every plan, including the free tier. Talk to us to activate the paid plan whenever you are ready.
Privacy & GDPR: how the AI Assistant handles your data
Routed through the Datacake AI proxy
Your browser never talks to OpenAI directly. All AI requests are authenticated and relayed by the Datacake proxy on your behalf.
Conversations stay in your browser
All threads (messages, titles, tags and usage stats) are stored exclusively in your browser’s IndexedDB. Datacake servers do not log, store or persist your conversations.
Not used to train OpenAI’s models
Requests use the OpenAI API tier, which means data sent through the API is not used to train OpenAI’s models per their published policy.
EU infrastructure for workspace data
Your IoT and workspace data lives on Datacake’s EU-hosted infrastructure. AI requests are processed on OpenAI endpoints in the United States, covered by a Data Processing Agreement.
Secrets are never sent
Passwords, personal API keys and authentication tokens are never included in the AI conversation context, only the messages and tool results needed to answer you.
Per-browser, per-device threads
Conversations are private to the browser and device you are using. They do not sync across devices, and clearing your browser data deletes them permanently.
AI Assistant: frequently asked questions
Last updated: May 2026
Ready to get started?
Create your free account and start monitoring temperature, humidity and air quality in minutes, or book a demo to see how Datacake fits your use case.
Get in touch
We typically reply within one business day.