Fleet Builder · Sign up on the web

Build the first agent today.

Start online with Torquay, Sparky, and optional frontier-model connectors. Test one real workflow, put approvals around it, and bring Apture in when it is ready for production integrations or a private deployment.

No install. No hardware. A magic link and a browser tab.
●app.apturelabs.com/fleet · live demo, click around
What you get

Everything in the demo above is the product.

01 / Playground
Draft, test, and version agents

Write the agent, pin its metadata to a team, basin, and cost center, then test against real prompts before anything ships.

draftsversioningv3.3
02 / Model choice
Frontier or local, per agent

Use live Apture models Torquay and Sparky, connect an approved frontier model, or bring a compatible model of your own.

TorquaySparkyfrontier connector
03 / Tools
Built-in capabilities plus your functions

Sandboxed code execution, grounded retrieval over well files, and web search. Add custom functions with strict schemas.

get_well_statusquery_historian
04 / Connectors
Your systems, pre-wired

Jira, Confluence, BigQuery, GitHub, Slack, and Drive out of the box. Register MCP servers for historians and land systems.

MCP serverswcsb-regulatory-library
05 / Deployment
Move from test to a governed endpoint

Keep the workflow in Fleet Builder, expose an approved endpoint, or work with Apture to deploy the agent inside your systems.

approval gateREST endpoint
06 / Spend
Seat pricing inside the harness

Use the included model access without forecasting a separate token bill. Export agent calls as Python, TypeScript, or cURL.

flat seatPythonTypeScript
The harness

Choose the model. Control the path.

Fleet Builder makes the model, tools, approval gates, and run history visible while you prove the workflow. The production boundary is a deployment choice, not a promise hidden in the fine print.

Sign up for Fleet Builder
01 · Choose
Pick the model and data path

Use Torquay or Sparky, connect an approved frontier model, or bring your own compatible endpoint for the workflow.

02 · Control
Set tools, limits, and approvals

Define what the agent can read, what it can draft, and which actions must stop for a person before anything runs.

03 · Deploy
Harden the workflow for production

Apture can add the integrations, evaluation, governance, custom model, internal deployment, and staff training the workflow needs.

Good first workflows

Examples to prove before production.

wellsite-constraint-triageProduction
Overnight alarms, ranked by deferred production

Ranks every overnight alarm by barrels at risk and drafts the intervention plan. The morning meeting starts with answers.

Examplehuman-reviewed triage
field-ticket-extractionAccounting
Field tickets coded before the coffee is poured

Reads scanned tickets and invoices, codes them to AFE and cost center, and flags rate mismatches against master service agreements.

Exampledraft before posting
aer-filing-assemblyRegulatory
AER submissions assembled from your own records

Pulls from the wcsb-regulatory-library and your well files to draft directive-compliant submissions with citations.

Examplecited drafting support
morning-report-drafterDrilling
Morning reports that draft themselves

Assembles the 6 a.m. report from WITSML feeds and the night tour sheet. The drilling engineer edits instead of writes.

Exampleengineer approval required

Start with one workflow.

Sign up for Fleet Builder now. If it proves valuable, Apture can build the production agent, deploy it privately, and train your team.