Frequently asked questions.

Tasq is an AI operations platform built for oil and gas. It finds failure patterns across your entire asset base before variance shows up in a report. The platform combines signal search, Build Your Own Model, automated downtime tracking, AI data entry, work management, and agentic workflows into a single system. Tasq connects 5+ data sources including SCADA, Artificial Lift, Production Accounting, Downtime, and WellView into one unified intelligence layer.

Tasq is the only zero code AI surveillance platform built specifically for oil and gas. Unlike generic anomaly detection tools that flag everything with high false positive rates, Tasq uses pattern specific detection with a feedback loop. Unlike low code platforms that still require data science support, Tasq lets any production engineer build and deploy models directly. Unlike SCADA dashboards that show what happened, Tasq shows what is about to happen. Platforms like Pi and Snowflake lack the signal search capabilities, while platforms like Palantir are great data connectors but lack the industry intelligence Tasq provides.

No. Tasq is built so that production engineers, field operators, and control room staff can build and deploy AI models directly from their own domain knowledge. In oil and gas, the real experts are domain specialists, not Python programmers. There is no SQL, no feature engineering, and no machine learning expertise required. Your domain expertise becomes the model. You are pointing at the pattern directly in the data and deploying it.

Tasq integrates with SCADA, WellView, Maximo, Pi Historian, Quorum, accounting systems, artificial lift controllers, and other operational data sources. Tasq connects to one or multiple sources simultaneously, aggregating data into a single intelligence layer. With Tasq v6, integration is even easier: connect to your database, spin up a Tasq agent, and access AI instantly through Microsoft Teams or the native interface.

Tasq delivers the highest accuracy timeseries pattern matching available for oil and gas. You can search 12 months of raw, high frequency SCADA data in 30 seconds per pattern. The search works directly on raw signals including casing pressure, tubing pressure, flowrate, and other channels. No SQL and no feature engineering needed. Every detection is automatically classified with root cause attached, and the positive/negative feedback loop sharpens match accuracy over time. With Tasq v6, accuracy jumped from 50% to 80% with 8x wider case coverage thanks to the agentic architecture where multiple agents validate, test, and refine results together.

Detection lead times depend on the failure type. Liquid loading is typically detected 31 hours before alarm, with 50 events detected per month. Paraffin buildup in tubing is detected up to 8 days before production variance, with 10 events per month. Plugged chokes are detected 40 hours before alarm, with 15 events per month. Manual surveillance misses 50 to 90% of events at scale. Tasq detects 4x more events than manual surveillance across every deployment.

Any pattern visible in SCADA data can become a model. Operations teams have deployed models across all issue types: plunger behavior change, circulating gas lift injection, injection setpoints variance, bad plunger setpoint change, slow building gas lift restriction, gas lift injection drift, heated separator burner failure, plugged choke, short cycle loading, pre liquid loading, compressor valve pre failure, offset frac interference, dump valve failure, tank leaks, and arrival sensor failures. If you can see it in the data, Tasq can search for it.

Build Your Own Model lets any production engineer turn a known SCADA pattern into a live AI model running across your entire asset base in under two minutes. The process has 4 steps: (1) Drag and drop over any timeseries pattern you recognize, (2) Thumbs up or down on the returned matches to refine accuracy, (3) Publish and deploy across your entire asset base with one click, (4) Tasq classifies root cause automatically and triggers downstream workflows. Every production engineer has seen liquid loading building, or knows how a plunger behaves before it fails, or recognizes the SCADA signature of paraffin weeks before variance shows up on a report. Build Your Own Model removes the data science bottleneck entirely.

After Tasq returns historical pattern matches, you give each one a thumbs up or thumbs down. Tasq recalibrates the model based on this feedback, sharpening accuracy with every iteration. In practice, initial deployment accuracy starts around 84% and improves to 94%+ within 30 days. False positive rates drop from 18% to around 3% over 6 weeks. The model recalibrates with each feedback signal and no retraining is needed.

Under two minutes from pattern to deployed model. One year of high frequency SCADA data is searched in 30 seconds. The feedback and refinement step takes about 60 seconds. Once you publish, the model runs on every well the moment you click deploy. In one operator case, three models were deployed in under 6 minutes total and now run continuously on every well across the basin.

Each detection triggers a full agentic workflow automatically. The event is classified with root cause attached. A work order is created and assigned to the correct field engineer. Vendor bids are surfaced inline for repair and intervention decisions. A reliability report is generated per event, per asset, per timeframe. Recommendations are assigned to the correct role. BOE recovery is tracked and the model feedback loop is updated. All of this happens without manual intervention, copy paste, or handoffs.

Tasq v6 introduces a fully agentic architecture. Multiple agents work together to bring the right results, embedding the right context in each prompt. Agents invoke models, graphs, maps, and recommendations, not just code. Integration is dramatically simpler: connect to your database, spin up a Tasq agent, and access it through Microsoft Teams. Accuracy improved from 50% to 80% with 8x wider case coverage. Build Your Own Model is now a core agent, enabling better search, better workflows, and more accurate recommendations. The agentic system validates, tests, and refines results before answering, eliminating the SQL hallucination problem.

Yes. Tasq handles data entry through multiple methods: automated audio entry where you speak your well report and Tasq transcribes and classifies it automatically, field data capture from mobile devices, PDF upload and transcription, and standard form entry. Tasq recognizes equipment failures, downtime events, and maintenance needs from voice input. AI recommendations are generated during entry based on well history, current conditions, and model outputs. The goal is to make data entry not the gatekeeper to actionability.

Tasq automates downtime hours, volume, and reason. Every downtime event is automatically detected, categorized, and analyzed without manual logging. In one major operator deployment, Tasq uncovered 2x more downtime events than what was being manually reported, and identified 4x more total production opportunities. For dump valve failures specifically, Tasq found between 1.5x and 4x more failure events than manual logging captured every month. That is not operator error. It is the fundamental limit of human surveillance at scale.

Tasq deploys several types of AI models. The Well Target Model dynamically updates each well's production targets using historical trends, real time data, physics models, and user feedback, uncovering 2x more production opportunity. The Bulk and Test Prediction Model projects gas and oil rates from pressure data when the well is not in test, uncovering 30% more deferment. The Setpoint Optimization Model recommends setpoint changes to optimize wells, delivering roughly 11% production increase on accepted recommendations. The Real Time Model catches underperformance live as it is happening, yielding a live prioritization list. On top of these, any user can build custom models through Build Your Own Model.

Tasq uses AI driven setpoint recommendations based on SCADA pattern analysis to optimize plunger wells. The system monitors plunger behavior changes, arrival sensor data, and cycle patterns in real time. When the system detects suboptimal performance, it recommends specific setpoint adjustments. For wells that accept recommendations, operators see approximately 11% production increase. The Build Your Own Model feature can also detect plunger specific issues like bad plunger setpoint changes and short cycle loading before they impact production.

Full platform deployment takes under two weeks. Tasq connects to your existing SCADA, WellView, Maximo, Pi Historian, Quorum, accounting systems, and other data sources. With Tasq v6, integration is even faster: connect to your database, a Tasq agent spins up automatically, and users get AI insights from the first minute of onboarding without any touchpoints needed. The old model of 3 to 6 month trials is a thing of the past. Each individual AI model deploys in under two minutes and runs continuously from the moment you publish.

Tasq delivers a proven 10x return on investment in the first month, scaling to sustained 100x ROI. At one major operator: $1M in new production revenue was generated in the first month. The organization internally reported a 2,000 bbl/day uplift equating to $36M per year. Downtime reporting improved by 2x, uncovering $75M per month in previously untracked opportunities. Downtime itself was reduced by 30%. Across just three Build Your Own Model deployments (liquid loading, paraffin, plugged choke), operators captured $4M in value over six months. Individual event values range from $3,255 to $16,640 depending on the failure type and BOE target.

Across deployments: 30% reduction in deferment. 200x faster than internal teams at building models. 2x more production opportunities uncovered. 86% reduction in operators time spent on manual surveillance. 60% reduction in cycle time. 4x more events detected vs manual surveillance. Operators no longer waste time manually entering downtime data or sifting through inconsistent SCADA dashboards. Engineers and control rooms focus on high impact actions while Tasq handles detection, classification, routing, and reporting.

The competitive landscape has shifted from applications to models. The traditional approach of comparing Product A vs Product B, dashboards stacked on dashboards, is outdated in an AI age. The value comes not from each model in isolation but from connecting them so your internal AI benefits from all of them together. Companies that stay locked into the application mindset will keep adding siloed products, limiting overarching intelligence. Companies that move to a model first mindset build internal AIs that are more powerful, flexible, and cost effective. With advances in MCPs and aggregation tools, this is now far easier than before.

The next generation of E&P companies will not be traditional energy firms with AI tools. They will be AI first companies that make better decisions, faster. Future operations will be run by AI agents that analyze data more accurately than humans, operate autonomously using reinforcement learning loops, and execute work itself. The first company to fully integrate AI into its decision making and operational workflows will be able to acquire and grow quicker than the rest and generate better cash flow and shareholder returns. AI is not going away. The best operations will come from AI driven workflows, and with Tasq, you can enable this in under 30 seconds.

An internal study revealed that 70% of features lacking intelligence failed. They added no real value and did not change how users worked. In contrast, features driven by AI that simplified decision making and transformed workflows received excellent feedback. This was the turning point. Tasq 5.0 was rebuilt from first principles around intelligence in every feature. The core problems solved: fragmented software layers, cumbersome model building, lack of scalability, and inefficient user workflows. By embedding search, analysis, model building, workflows, and data entry into one unified system, what used to take hours became a 10 second process.