— decision science for energy capital

Brent moved this morning. Your plan already knows.

EPPortfolio holds your opportunities, your sanctioned projects, the rigs and facilities they compete for, and their measured performance — on one live model. Change a price deck, slip a crew, cut capex fifteen percent: the cash flows and the constraints resettle together. Ask omni in plain language; the engine does the arithmetic.

live economics

Every input is connected. Change one and the portfolio recomputes — no re-run, no rebuild, no waiting on the asset teams.

schedule & constraints

Pads, crews, facilities and dependencies resolved as one program. The economics run over the schedule you can actually execute.

omni, the ai copilot

Ask in plain language. The agent finds the question; the sealed engine computes the answer.

— the cost of a slow answer

The question changes faster than the answer arrives.

A reprice is not hard. It is long. Pull the models, agree the assumptions, chase the asset teams, rebuild the rollup, reconcile the variance, put it in slides. Weeks of elapsed time and a handful of people, every time — and the market has moved again before the deck is printed.

So you ration your questions. You run the two scenarios you can defend and skip the eleven you'd like to see. The plan you present isn't the best one you could find. It's the best one you had time to price.

the reprice

A price-deck change means starting the cycle over.

the rollup

Assembled by hand, stale the day it lands.

the plan that can't be built

Priced, approved, and infeasible. You find out in execution.

— what instant actually means

One model, always solved.

EPPortfolio

EPPortfolio's core is a portfolio calculator and scheduler — candidates, rigs, facilities and dependencies resolved as one program — with economics computing over it continuously. Nothing is a batch job. Move a sanction date and the cash flows, the constraints downstream and the portfolio total all settle together.

The economics engine is yours to shape, not ours to gate. Today we implement your fiscal terms with you, inside the engine, alongside a sealed calibrated core — results from your models and results from the core stay distinguishable, and the reference math is always recoverable. The no-code editor that lets your team do it unaided is in build, shaped by the operators we build alongside. Either way, your fiscal terms are not a vendor ticket.

engine-versioned · inputs hashed · every figure reproducibleno-code editor · on roadmap
portfolio · live● recomputed 0.4s ago
portfolio NPV, $mm420
projects above hurdle4 / 6
capex called, $mm357
KESTRELAUGERMARLINTALUSVERDECORIANDER
illustrative portfolio · six assets
— the agent, and its limits

omni proposes. The engine computes. You decide.

Ask in the language you'd use with a colleague — what does 2028 free cash flow do if we defer the Phase 2 tieback and Brent sits at 58? — and get the answer with its workings attached. omni's job is to find the question inside your sentence and route it to the right tool. It never computes a number. Arithmetic belongs to the versioned engine, where it can be checked.

Every write is gated: propose → confirm → audit. MCP-native, so you point omni at your own frontier model, in your own tenant, under your own terms.

omni · governed copilot● connected
defer phase 2 tieback to Q3 2028 — what does it do to the portfolio?
Portfolio NPV moves −$18.4mm at P50. The rig frees in Q1 and two workover candidates pull forward, recovering $11.2mm —
−$7.2mmnet · P50
engine-versioned · inputs #a41f0c · P10 · P50 · P90 · 0.9s
— one number, all the way down

The rollup is a view, not another calculation result.

Opportunities, sanctioned projects and their measured performance live on the same data model. The portfolio total isn't assembled from twelve spreadsheets each Thursday — it is the model, read at a different altitude. Walk down through it: business unit, asset, project, well. Nobody spends the first twenty minutes of the meeting arguing whose version is current.

And because outcomes sit beside forecasts on that same model, learning from what happened is a property of the data, not an exercise someone runs once a year.

The same method prices a barrel and a kilowatt — oil and gas today, the rest of the mix on the same model.

rollup · NPV $mmP50
portfolio1,284
north business unit742
Kestrel field418
Kestrel-14 workover37
south business unit542
illustrative · same model, four altitudes
— where plans actually fail

The economics assume a schedule. The schedule assumes nobody is in the way.

A model that prices projects one at a time is answering the wrong question. In a development program the projects are not independent: pads share a frac spread, wells share a gathering system and a plant, and a well completed next door changes the curve on the one already producing.

So the plan clears the hurdle rate and still cannot be executed. Two pads want the same crew in March. The plant is full in July, and the volume you booked is curtailed — not deferred, gone. The offset completion lands and the parent never returns to the type curve you sanctioned it on.

None of this is in the spreadsheet, because a spreadsheet has no way to represent it. EPPortfolio resolves candidates, crews, facilities and dependencies as one program and computes economics over the resolved schedule, not the intended one.

program · constrained● resolved
M1M4M7M10M13M16Pad APad BPad CPad DPad EPad Frate, mboe/dplant capacity
curtailed volume, mboe283
NPV impact, $mm−7.9
crew conflicts1
first infeasible monthM9
illustrative program · six pads · one facility constraint · deterministic

We are starting where these constraints bind hardest and public production data is good enough to show the method before anything of yours changes hands.

Ranges and probabilistic evaluation are on the roadmap. Today the engine is deterministic over an explicit schedule.

Your data, and only yours.

never trains models. yours is never used to train any model — ours or anyone's.

isolated & encrypted. separated per client, encrypted in transit and at rest, stored in a region you choose.

yours to leave with. full export any time, in open formats. source-code escrow available.

reproducible. every figure engine-versioned, every action in an audit trail you can read.