Omnismith

What Omnismith is

Omnismith is a managed database product for organizations that need records with reviewable change history, metrics that track over time, and automations triggered by thresholds and state changes. Omnismith lets the schema itself become data that agents and teams can modify at runtime without needing migrations or release windows. Its own use-case pages point at developers building ops tools and at infrastructure and asset tracking, from server monitoring and facility fleets to clinical trials and ESG carbon accounting.

The company emphasizes that in Omnismith, the schema is data. In a fixed-schema tool, the columns are part of the build; changing them is a migration, a review and a release window. In Omnismith, templates and their attributes are records, created and changed through the same API that creates rows. This design principle flows through everything the platform does: a new field is one API call, role-based access applies to an agent session exactly as it applies to a person, and change history is kept from the first write.

What you build and how

Omnismith organizes data into templates (like tables) with entities (records), attributes (fields), dimensions and metrics. The home page says what an agent builds there carries change history, charts, role-based access and a REST API from the first write.

The platform supports four typed kinds of fields. Dimension holds current state. Metric holds a time series, so each metric attribute carries its own time-series storage and charts. List holds a controlled option set. Reference points at another record. Eight storage types sit underneath: string, number, boolean, datetime, date, file, image and markdown.

Every dimension change lands in the append-only history log with the author and timestamp. When you query the history, you see the previous value, the new value, who made the change and when. You can filter the history log by attribute, date window, author or value. This audit trail is immutable and reviewable without depending on a separate log trail.

Metric attributes carry their own time series. Per-entity charts aggregate metrics with avg, sum, min, max, count, first or last over time buckets from one second to one month. Dashboards can assemble stats, charts, gauges and filtered lists across the whole fleet or workspace.

Automations evaluate multi-attribute conditions on entity and metric events, then dispatch to Telegram, webhooks or mobile push. If a refrigeration unit reports coolant temperature above a threshold for ten minutes, an automation can send a message to whoever owns the site. A sample prompt on the home page asks for refrigeration units across 40 retail sites with a coolant reading every five minutes, and the agent answers by creating a template with six attributes, where coolant temperature and compressor hours are metrics. Readings arrive as batched observations through POST /v1/entities/{id}/metrics, so gateways, scripts and scheduled jobs can write straight into the record an operator opens in the browser.

Agent sessions work inside the same role-based access controls that apply to people. Owner, Admin, Editor and Viewer roles gate every write. Dynamic entity scopes narrow what a member reaches. When a person authorizes an MCP session, the agent inherits that person's permissions and can read and write using the same APIs. The home page offers a starter prompt for a multi-currency agency CRM with commission tiers, and says it works with Claude Desktop, Cursor and any MCP client.

The REST API exposes the structure agents create. Programmatic access tokens read and write the same templates, attributes and entities. The OpenAPI specification describes all of it, making the tool catalog possible for AI agents.

Plans, limits and hosting

Pricing is a flat tier per workspace rather than per seat, and every plan includes unlimited users, the REST API, 8 languages and Google Sign-In. The FAQ defines the unit that matters: an entity is "a single record in your database", such as a server, a product, a client or a device, based on a template you define.

The Free plan costs "$0 forever" and is labeled "Perfect for trying things out". It includes 250 entities, 25 attributes, 7 days of history retention, 1,000 AI credits, 1 dashboard and 5 automations, with no credit card required, and the page stresses that it carries "the same history, metrics, and API as every other tier". The Team plan, marked "Recommended", costs $79 a month for "teams running live workflows" and includes 15,000 entities, 250 attributes, 30 days of history retention, 15,000 AI credits, 15 dashboards and 50 automations. The Business plan costs $199 a month "For larger datasets and longer review windows" and includes 50,000 entities, 600 attributes, 90 days of history retention, 50,000 AI credits, 30 dashboards and 100 automations. A link offers to "Compare all 5 plans", so the 3 cards are not the whole range.

The FAQ adds the plan terms. You can change plan at any time, and downgrades take effect at the end of the current billing period. If you hit a limit, "You'll see a notification and won't be able to create new resources beyond the limit", and the FAQ adds that existing data is always safe.

Above the plans sits an Enterprise and On-Premises offer headed "Data Sovereignty & Custom Self-Hosted Licensing". Under the line "Zero Vendor Lock-in" it says you run on the managed cloud today, and that "custom self-hosted licensing and full database export are available on request" when regulatory, air-gap or infrastructure policies require on-premise execution. The FAQ repeats this and names the export as a complete PostgreSQL database export, arranged through engineering via Book a Demo. The company's comparison pages describe the standard product as managed and list general self-hosting as planned, "not available today", so on-premise use is the enterprise route rather than a download.

On compliance the FAQ is careful. Asked about SOC 2 and ISO 27001, it describes an "Audit-Ready Architecture" with immutable change logs, cryptographic author attribution, role-based access control and continuous time-series telemetry, designed to satisfy ISO 27001 and SOC 2 evidence requirements "without claiming vendor certification". One of the starter prompts is a SOC 2 IT asset inventory, which fits that framing.

Search, rules and starting points

Beyond records and metrics, the home page lists structured filters across dynamic attributes, full-text search over string fields and semantic search "for the cases where the exact wording is unknown". Rules are enforced on every write as when/then logic ("A status change is either valid, or it never happened"), and named Actions bundle a precondition, the fields an operator fills in and the values set silently into one guided operation. Attributes can be addressed by slug in the API, so a payload written today survives an id change.

The MCP server uses OAuth 2.1. You add the server URL to Claude Desktop's custom connector dialog or to a config-file client such as Cursor, approve a browser consent screen, and pick the one project the session may reach. Every call then executes as you and lands in the audit log under your name. The tool catalog covers schema design, records, telemetry ingestion, search, dashboards and automations, with tools such as scaffold_data_model, search_entities, get_entity_chart and create_automation.

Starting points come two ways. Starter prompts pre-configure a workspace at sign-up, with options for an agency CRM, SOC 2 IT assets, a cold-chain fleet and support SLA tickets, so there is "no empty canvas". Blueprints install a working set of templates, attributes and sample records in a single call, across Agency Project Tracker, Content Editorial Pipeline, Event Attendee Logistics, CRM and Deal Pipeline, IT Asset Management, Product Information Management, Product Roadmap Planning, Server Monitoring, Smart Facility Fleet, AI Agent Auditing, Adaptive Clinical Trials, ESG Carbon Accounting, Crisis Management and Precision Agriculture. The schema stays runtime data afterwards, so the first customization is another sentence to the agent.

Who it suits and who it does not

Omnismith fits infrastructure and facilities teams tracking fleets where the same entity needs current state, metric trends and later review of changes: chillers, pumps, HVAC and sensor nodes spread across sites, which the home page calls "the clearest fit". It suits developers building ops tools who want schema flexibility and APIs together with history, metrics and alerting in one managed product, and teams that already work with an AI agent and want it to design and fill the backend inside the same role permissions as a person. Flat pricing with unlimited users makes it easy to bring in contractors, who can be scoped to their own sites.

The caveats are about scope. It is a managed cloud product first, with on-premise use reserved for custom enterprise licensing. The Free plan caps history retention at 7 days and allows 5 automations, so review windows and alert volume are the first limits a growing team meets. And it is built around records, metrics and agents rather than a spreadsheet grid, so a team that wants a familiar table front end over an existing SQL database is describing a different kind of tool.