Firedog reads your provider usage and your gateway logs. It never sits in the request path, so it cannot slow down or break a workflow in production. What comes out is not a dashboard. It is data a leadership team can act on.
Read-only keys for OpenAI, Anthropic, Azure OpenAI, Bedrock, and your internal services. Fifteen minutes, no code in the request path.
Every prompt, model, user, and workflow is tracked automatically, and tagged with a team and an owner. Including the calls nobody wrote on purpose.
Cost per run, per user, per team, per workflow. The unit economics of every AI decision, before the invoice arrives.
Thirteen workflows companies already run, across sales, operations, HR, engineering, research and compliance. Each one is decomposed to the cent, and each one hides most of its cost.
Ingest 50-200 transcripts per cycle, run analyst Q&A against them with context stitched from a vector index.
Multi-source research agent that autonomously browses, retrieves, and synthesizes a structured sector brief.
10-K, prospectus, and fund-doc review using a long-context model to extract risk factors, covenants, and material changes.
Recurring agentic run across multiple tickers - news, filings, macro data - producing structured trade signals with compounding history.
Structured investment memo from 20-50 source documents - CIM, financials, management decks, diligence notes - with citation-backed sections.
High-frequency classification and summarization of news and SEC filings into structured alerts routed by materiality tier.
Bulk ingestion of counterparty packs, ownership structures, and AML docs with redundant retrieval patterns typical of compliance workflows.
Analyst chat interface over a RAG index of internal research notes, models, and meeting transcripts with multi-turn conversation history.
Draft a branded, on-message commercial proposal from Salesforce opportunity data and a Box library of past decks, with citation-backed sections.
Turn Zoom, Google Meet, and Teams transcripts into structured summaries and route action items into Slack and Salesforce automatically.
Employee-facing chat over Workday data and policy docs, answering PTO, benefits, payroll, and policy questions inside Slack or Teams with multi-turn memory.
A scheduled agent that reads meeting notes, emails, and market data, then updates Salesforce opportunities - filling fields, advancing stages, and flagging stale deals.
An in-IDE and CI coding agent that reads the repo, writes and edits code, runs tests, and opens PRs - iterating until the task is green.
Change one line in your OpenAI, Anthropic or Azure client. The collector becomes a transparent proxy.
Every call is priced, recorded, tied to a workflow. Prompts and responses stay in your VPC.
Cost, performance, ROI per workflow, per team, per organisation. Drill down to the CallRecord.
Compare model costs across providers on the same workload.
Understand how prompt size contributes to cost, run after run.
Trace every API call from start to finish. Routers, retries, reflection steps: real billed calls hidden inside one logical invocation.
Invisible on your invoiceMeasure embeddings, reranking and vector search costs, usually billed by two other vendors.
Invisible on your invoiceSpot conversations that quietly become more expensive over time. An agent pays for its own transcript again on every turn.
Invisible on your invoiceBox, Salesforce, Slack, Zoom. Every sync is a billed model call, logged as a pipeline step, never as workflow cost.
Invisible on your invoiceIn your VPC. Prompts, RAG context and responses are written to a local Postgres you control. Firedog receives only metadata — never content.
Control plane hosted in Europe. Any personal data stays in your data plane. DPA available on request.
Type I in preparation. The architecture (metadata-only on the Firedog side) sharply reduces the scope.
OpenAI, Anthropic, Azure OpenAI, Mistral, Bedrock. Any OpenAI-compatible API via base-URL swap.
One line of code to change, one line to revert. Zero lock-in.
Monthly. Cancelable anytime.