ComboChat + ComboCompute v1 Beta Overview
Purpose
This document is a high-level overview of what ComboCurve’s AI Agent can currently accomplish, as of the release of v46 on July 6, 2026.
Note: AI can make mistakes. Please double-check responses as well as share feedback on any prompts or workflows that behave unexpectedly — your input is valuable for improving the product and experience.
Temporary disclaimer: agent operations are capped at 1000 wells
Overview
ComboChat + ComboCompute can be accessed anywhere in the platform by selecting the logo in the bottom right corner. The icon can also be moved around the screen by hovering over and selecting the arrows to move the icon to different corners on the screen.
Once clicked, the Combo Agent panel will open where you can then type your instructions. At the top there are the following icons:
- Pin - pin panel to side
- Agent Rules - here you can add rules to customize the agent behavior
- Save Configuration/Workflow
- Copy chat history
- Reset Conversation
You can also report feedback to responses by selecting the broken dog bone icon -
Capabilities
General
- Add wells to a project from the company level, or create new wells in the project
- Filter wells to create a forecast, scenario, etc.
- Operate on the same permissions structure as other ComboCurve operations by user
Forecasting
- Create, copy, and alter Modified Arps auto-forecasts within accessible projects
- Configurable parameters: b factor, D-Eff, Peak Preference (Max or Last Peak), Date range (All or Last XXX Data Points), forecast resolution
- Utilize saved configs in auto-forecast form, or override them with above params
- Ability to run a ratio forecast
- Proximity forecasting: ability to set # of data points threshold, add criteria, select normalizations, and change fit type parameters
- Ability to add wells to the editing bucket via certain criterion and reiterate forecasts on those wells
- Ability to reference another forecast (or previous iteration) in EUR restrictions to change parameters but remain within a certain % threshold
Diagnostics
- Run diagnostics immediately after running or refreshing an auto-forecast
- Change the amount of forecast/production data used in diagnostics error calculations
- Agent can analyze raw diagnostics data — the only current capability not replicable in the platform UI
- Supports statistical analysis (P10, P50, P90, etc.) on all stock numeric diagnostics scalar metrics
- Example: flag wells outside a b-factor range and reforecast them with the P50 value
Type Curve
- Create a new type curve and allocate a forecast to it
- Re-run a type curve with a stock fit (Arps + M Arps Non-Smooth, 1 to 1 normalization)
- In existing type curves: change the attached forecast and re-run a fit after forecast updates
- Apply type curve to a forecast or group of wells in a forecast
Scheduling
- Create a new schedule (defaults to all wells without monthly production data in project)
- Assign time steps for drilling and/or completion operations (v1 default: 1 drilling rig, 1 completion crew)
- Set or override the schedule start date
- Re-run an existing schedule with a new start date, or copy a schedule to make changes
Economics
- Run economics on a pre-existing scenario
- Create a new scenario with a subset of wells in the current project. Automatically creates a General Options model with a current day Aggregation Date
- Ability to create and save econ models: Ownership & Reversion, Dates, CAPEX, Pricing, Differentials, Expenses, Stream Properties, Taxes
- Ability to create and edit multiple qualifiers for the same assumption type. Ability to assign different models to different qualifiers
- Ability to run combos of qualifiers through the agent
- Ability to assign forecasts and schedules
- Ownership: Add reversions, also can add data snaps references to WI & NRI headers
- Pricing: Ability to call and reference NYMEX/Henry Hub price decks
- Pricing/Diffs/Expenses/Stream Props: Ability to create a time series
- Taxes: Auto-assign taxes models by state when well header is populated upon scenario creation
- Ability to reference scenario templates for easy scenario creation
Rules & Workflows
- Store rules for the agent to recall in future prompts; ask the agent to suggest and save a preliminary rule for a given operation
- Store prompts and step-by-step workflows for repeatable agent execution
Sample Prompts
- Create a forecast in this project named Forecast123.
Run using all wells data and enable proximity forecasting for wells with less
than 6 months of data. Run diagnostics on the forecast.
- Retrieve diagnostics results from Forecast123 - what
are the P10, P50, and P90 for the oil b factor? What is the P50 Absolute
Relative Error Percentage?
- Retrieve diagnostics results, please flag all wells
with oil b factor > P90 oil b factor and oil b factor < P10 oil b factor.
Then reforecast the flagged wells with a set oil b factor = P50.
- Run the type curve ‘WCA TC’ with the updated forecast
‘Forecast123’.
- Run schedule ‘Infill Well Schedule’ with a start date
of 6/1/2026.
- Run economics in the scenario ‘Cashflow – PUD’.
Sample Workflow
Example 1:
Run a new forecast on the
current well set, called ‘One Click PDP Analysis’; then iteratively improve it
using diagnostics.
Steps:
1. Run the auto forecast with the last 60 months of data,
as well as enabling proximity forecast for wells with less than 6 months of
data.
2. Run diagnostics on the forecast.
3. Retrieve diagnostics results - identify wells with
absolute relative error > 40%.
If problematic wells are
found:
4. For the flagged wells, re-run the forecast with a
shorter lookback period by 12 months from the last forecast step, and the P50
b-factor from Step 3.
5. Run diagnostics again on the re-forecasted wells.
6. Check results again - repeat from step 4 until either:
all wells have absolute relative error of less than 30% or you have done 4
passes, in which case please stop and report remaining outlier wells
When done with the forecast iterative process:
7. Create a scenario called ‘Cashflow – PDP’. Reference the WI & NRI ownership headers, use an As of Date of 6/1/2026, use the price decks from 6/10/2026, insert oil diffs = -1, gas diffs = -.25, fixed expenses of 2000/well for the first 6 months from FPD, and 1200/well for the remaining months, an NGL yield of 90, and the forecast we ran in the previous steps. Run economics when assembled.
Example 2:
Type Curve Analysis of asset in Delaware Basin
- Create a forecast of all wells in the project with at least 24 months of production data, called ‘Rep Wells’. Run a modified Arps forecast here using all wells data the max peak.
- Using the Rep Wells forecast, create a type curve for each group of wells where the Target Formation = WOLFCAMP A/B, WOLFCAMP C/D, and BONE SPRING. Run the type curve fit on all 3.
- Create 3 new wells in the project and add them to a forecast called ‘TC WELLS’.
- Apply each type curve to one of the wells based on the date 7/1/2026.
- Create a scenario with the 3 new wells, and use an As Of Date of 7/1/2026, the forecast TC WELLS, 100% WI & 75% NRI, the price decks as of 6/1/2026, a drilling cost of 5 million dollars 90 days before the FPD, a completion cost of 3 million dollars 45 days before the FPD, and an NGL Yield of 100. Run economics when assembled.
Example 3:
Create a new forecast of all the wells in the project called Proximity Time Series Evaluation. Follow the steps below based on the Months Produced header buckets defined below.
1. Split wells into 3 buckets:
- over 36 months of data
- 6–36 months of data
- <6 months of data
2. For wells with >36 months:
- Forecast using last 36 months
- Run Diagnostics for last 6 months of production data
- For wells with oil B-factor, oil EUR/ft, oil decline, or oil CUM diff fall outside P10-P90 ranges, move flagged wells to the edit bucket
3. For edit bucket wells:
- Run Proximity forecasting on these wells normalized on a 1:1 basis
- Do not match the proximity curve to current well data
- Use proximity EUR from this step as the EUR constraint on a second pass on these wells, using only the last 6 months of production data
4. For wells with 6–36 months:
- Run Proximity forecasting on these wells normalized on a 1:1 basis
- Use proximity EUR from this step as the EUR constraint on a second pass on these wells, using only the last 6 months of production data
5. For wells with <6 months:
- Use Proximity forecasting directly, using production data to inform the curve, and normalize on a 1:1 basis
Video Walkthrough