The Obliq Platform
Obliq connects knowledge, models, agents, and tools into one observable runtime. Your architecture stays yours as models change.
One environment to build, deploy, and operate AI systems. No lock-in.
Who the platform is for
Obliq is for teams shipping AI that touches private data and live systems. If the work is a one-off completion, a model API is enough. If the work must retrieve the right source, call a ticket system, and explain itself later, you need a runtime that holds those pieces in one place.
How the system fits together
A request enters an agent. The agent retrieves from connected knowledge, calls a model endpoint you choose, and invokes only the tools it was granted. Obliq records each step. You inspect the trace, replay it with a different model or instruction set, and score groundedness and tool reliability before the next change ships.
System architecture
Production AI is a system: context, reasoning, tools, execution, and improvement. Obliq is the layer that holds those pieces together so a provider change does not force a rewrite of the workflow around it.
Connect the sources agents depend on and keep them versioned, current, and attributable. Retrieval then cites what was used, not a black-box context blob.
Assemble a runtime unit from instructions, a model endpoint, scoped knowledge, tools, memory, and policy. The definition stays stable when you swap a provider.
Let agents call APIs, MCP servers, webhooks, and functions under explicit grants. Every invocation is written into the execution trail.
Inspect each step of a run—retrieval, inference, and tool calls—with latency and source attribution. Replay the same input against another model or instruction set.
Measure groundedness, instruction adherence, tool reliability, and cost on real executions. Use those signals to change a component instead of guessing from a chat log.
Questions
What is the Obliq platform?
Obliq is infrastructure for AI systems. It connects knowledge sources, model providers, agents, and tools into one observable runtime so application logic does not have to be rewritten when a model or protocol changes.
Who is Obliq for?
Teams that already operate more than a single model call: they ingest private knowledge, invoke business systems, and need a record of what the agent actually did. Individuals can start on the same runtime and expand into evaluation and routing.
Does Obliq lock you into one model provider?
No. The runtime is architecturally model-agnostic. You can bring your own keys for leading providers or point agents at private and open-weight endpoints. Agent definitions reference capabilities, not a vendor SDK.