Wells, leases, batteries, tanks
The asset spine, named the way the field names it. Add a well and it is yours to work the moment it exists; with an API number its available production history syncs from the Wellsite data cloud.
Platform
Where the oilfield works
Wellsite is the AI operations platform for upstream oil & gas.
It is two things in one place. The platform an operator runs its wells on — gauges, routes, hauls, projects, people, files — which is free, without limit. And a team of AI agents that works those wells alongside your people, which is one switch for the organization.
The agents monitor wells, identify production issues, coordinate service providers, handle compliance, reconcile production and financial activity, and evaluate new opportunities.
Because everyone works from the same well-centric operating context, an issue moves from identification to decision to field work to measured outcome without losing the context that connects it all.
The operating loopFind the problem. Fix the problem. Prove the result.
The platform
What’s on Wellsite
The operation, as it actually runs. Free, with no limit on wells or people.
Gauges by text or by form
A pumper texts the numbers to Wellsite or taps them into the Gauge app. The reading lands on the well, stamped with who took it, when, and how.
Routes and pumpers
Standing routes, the people who run them, and a morning prompt that asks for the gauges. A pumper needs no seat and no training.
Hauls at the trigger
Set a tank's capacity and haul trigger once. When a gauge crosses it, the haul request is raised — and dispatched to your hauler automatically if you told it to.
Still-on-location checks
A route due today with no gauges by the cutoff. A crew on location with nothing captured for hours. A haul en route with no ticket. Each gets a text, then an escalation to the employer and the operator.
Projects with providers
Award work to a service company and invite them into it. Their access is scoped to the job — the condition, the scope, the well — and ends with it, unless you make it standing.
People, access, files
Your team, your pumpers, your providers, each with exactly the door their work needs. Documents live on the well. No seat limits.
The activity feed
Everything that happens, in one append-only stream — a gauge, a haul, an award, a finding, a decision, a removal — attributed to the person or agent that did it.
Architecture
Built around the well
Everything comes together around the same persistent operating record:
The well.
- Production.
- Equipment.
- Field activity.
- Service providers.
- Recommendations.
- Projects.
- Documents.
- Costs.
- Regulatory obligations.
- Decisions.
- Outcomes.
The more work that happens through Wellsite, the more operating context the agents have when deciding what should happen next.
That’s why Wellsite isn’t simply a collection of AI agents.
It’s the place where the operator, service providers, and AI agents work together around the well.
Operating object
The well is the operating context
The lease defines the economic relationship around the asset.
The well is where production, equipment, operating decisions, recommendations, projects, work history, and AI context come together. A well is named by its lease and its number on it — Lucky Dog #1 is well 1 on the Lucky Dog lease.
The battery connects wells to surface production and storage.
The tank is the gaugeable object where inventory and disposition activity occurs.
Wellsite understands that physical and economic hierarchy while keeping the well at the center of the operating record.
Lease → Well → Battery → Tank
One record
One operation. Shared context.
- Your people should not have one version of the well.
- Your Production Engineer should not have another.
- Your pumper should not have another.
- Your service company should not have another.
- And your AI agents should not be reasoning from disconnected fragments.
Wellsite brings the relevant operating context together so each authorized participant works from the same underlying record.
People + AI agents + service providers + data + work — together around the well.
Operating history
Work builds the operating record
- Every gauge.
- Every recommendation.
- Every project.
- Every provider.
- Every report.
- Every decision.
- Every outcome.
All of it adds context to the well.
Structured work creates the authoritative operating history. Comments and collaboration enrich that history.
The result is a record that shows not just what the well is, but what has actually happened to it.
Field
From the physical wellsite to the digital wellsite
The work still happens in the field.
A pumper texts this morning’s gauges to Wellsite, or opens the Gauge app at the tank and taps them in. Either way the reading lands on the right well, stamped with who took it, when, and by which channel.
A crew on a project captures what happened on location and the completion comes back to the well that sent them there.
The location becomes context. The work becomes history.
Access
Access follows work
Service providers participate in private operations because the operator has authorized work.
They do not get permanent access to everything the operator knows about a well.
A provider’s access is a grant scoped to the job — the wells, the project, the capabilities it needs — and it ends when the job does. A regular pumper holds a standing grant to a route and nothing more.
That lets operators bring outside companies into the same operating workflow without turning private operations into a public marketplace.
Insight to work
From insight to physical work
A production problem should not end as an alert.
A recommendation should not end as a PDF.
Wellsite connects intelligence to the work required in the field.
Condition Analysis Recommendation Decision Work Cost Outcome
- The Production Engineer identifies the problem.
- The operator decides what should happen.
- The Operations Manager coordinates the work.
- The service provider performs it.
- The completion record comes back to the well.
- Then Wellsite measures what happened afterward.
Outcomes
Outcomes close the loop
A recommendation is more useful when the operator can see whether it worked.
Wellsite preserves the original expectation when work is approved so the eventual outcome can be compared against it.
Expected
- Expected production response.
- Expected cost.
- Expected payback.
Actual
- Actual work performed.
- Actual cost.
- Actual production response.
- Actual economic result.
The result becomes part of the well’s operating history and gives future decisions better context.
Find the problem. Fix the problem. Prove the result.
Trust
Trust is architectural, not aspirational
Wellsite asks an operator to let AI agents and outside service companies work inside their operation. That only works if the boundaries are enforced by the system — not by policy, and not by asking the AI nicely.
Isolation
Enforced in the database, not the prompt
Every agent runs as a principal of exactly one organization, pinned to that organization’s credential. Row-level security is on and forced for every table in the platform, and the service that answers your requests has no ability to bypass it.
That means an agent cannot read another operator’s data even if someone tries to talk it into doing so. A prompt-injection attempt gets the same answer as any other unauthorized request: nothing.
The prompt does not get a vote.
One agent run belongs to one operator. Agents learn from your own operating history and from public data — never from another customer’s private operation, and your operating data is never used to train a model.
Guardrails
You set bounds, not tuning
Out of the box, every agent watches and drafts. Nothing more.
You raise each one to the authority you want, per agent and per task:
Level 1
Watch
Monitor the operation.
Level 2
Recommend
Tell your team what should happen.
Level 3
Prepare
Create the work for human approval.
Level 4
Execute
Carry out approved classes of work automatically.
Anything that can spend carries a ceiling you set per action.
Guardrails live in one place — your agent settings, changeable only by an organization admin. Not in a prompt, not in a per-project override, not in a user preference.
Five things never move automatically, at any setting:
Sign or submit a filing
Move money
Grant standing access
Commit above your ceiling
Make an offer on an asset
Evidence
The numbers, and the reasoning behind them
- Calculations are deterministic. Decline curves, economics, benchmarks and every other figure come from engineering models and tools — never invented by a language model. A number you see traces back to the calculation that produced it.
- Nothing is asserted without data. When a required reading is missing, the agent says what it cannot conclude and what would change that. It does not fill the gap with a guess.
- Every evaluation is logged — including the quiet ones. Wellsite records what each agent saw, concluded, ruled out and why, and did, on every well it looked at, including the wells where it raised nothing.
- The operating history is append-only. A correction supersedes the original entry; it never overwrites it. What happened stays what happened.
So the question after a surprise isn’t only “why did it do that?”
It’s why didn’t this fire? — and that answer is retrievable.
Auditability
The activity behind the operation
Every consequential operating action answers:
- Who did it?
- What happened?
- When did it happen?
- Where did it happen?
- What well, project, recommendation, or document did it affect?
- What changed?
That activity powers the operating history, audit trail, notifications, AI context, and collaboration around the well.
One operating history you can follow from signal to decision to work to outcome.
Put the operating context around every well.
Well management free · AI agent team $99 per well, per month
Give your people, AI agents, and authorized service providers one place to work from.
No contract. Cancel anytime.