AVP AI Publications
Materials from the AVP AI team on applying AI in oil and gas: articles and analysis published on external platforms. The collection below shows how the platform performs on real upstream tasks — from screening license blocks to automating interpretation workflows.
AI agents for business: how major companies are rolling them out in 2026
May alone brought four major industry headlines on AI agents — strong models are available to everyone now, and the winners are the teams that assemble a working system around them. A walkthrough of how the industry is moving from pilots to multi-agent teams with role separation and oversight, and why AVP AI runs six specialized agents to perform investment assessment on license assets. (Original in Russian.)
Read on Tambur Pub →$38B in US oil and gas M&A in Q1 2026 — what's shifting in asset valuation
The strongest quarter for US oil and gas M&A in two years closed at $38B in transacted assets. The bigger story isn't deal volume — it's a shift in how operators are valuing assets. AVP AI breaks down the three core changes in approach and which of them apply to Russian license auctions. (Original in Russian.)
Read on Telegram →Why 95% of AI projects in oil and gas die before they pay back
A breakdown of the credibility crisis around AI solutions after a wave of failed industry projects: startups and operators are stuck in a feedback loop — one side needs production data to prove value, the other waits for a turnkey solution instead of co-developing it as partners. (Original in Russian.)
Read on VC.ru →Kuyazinsky license block — full investment assessment of a Rosnedra lot
An end-to-end run of the AVP AI platform on an active Rosnedra license lot: license screening, infrastructure context, analog-based resource estimation (P90–P10), and 63 drilling-strategy combinations evaluated across three development scenarios. (Original in Russian.)
Read on LinkedIn →Halliburton: first fully automated drilling run offshore Guyana
Halliburton, ExxonMobil, Sekal, Noble and Wells Alliance completed the industry's first closed-loop geosteering drilling run offshore Guyana. Results: 15% reduction in scheduled time, 33% less tripping time, and 470 m of lateral reservoir section drilled fully autonomously. (Original in Russian.)
Read on LinkedIn →AI startups raise 9x more capital than traditional SaaS
The average AI startup round in Silicon Valley reaches $137M versus $15M for classic New York SaaS companies. AI teams operate at 40–60% EBITDA margins compared with 20–40% for traditional SaaS — premium valuations earned through operational efficiency and proven outcome metrics. (Original in Russian.)
Read on LinkedIn →Oil and gas shifts from AI pilots to enterprise-scale deployment
Half of all AI budgets in the industry now flow into process optimization: algorithms tune drilling parameters in real time and predict equipment failures. Documented case studies show 140 hours of avoided downtime, a 40% drop in breakdowns and $10M in annual savings. (Original in Russian.)
Read on LinkedIn →Is AI the best hire of the year?
88% of companies have already "hired" AI, but 94% see no measurable returns. Organizations that capture real impact are 2.8x more likely to redesign processes around the technology instead of bolting it onto legacy workflows. With market data from Gazprom Neft, Sber, Severstal, Norilsk Nickel, Saudi Aramco and JPMorgan. (Original in Russian.)
Read on Telegram →SLB handed seismic interpretation to an AI agent inside Petrel
The world's largest oilfield services company released a live demo of an AI agent running interpretation workflows inside Petrel. AVP AI unpacks one of the most telling industry milestones of the year — how a sector that held on to proven methods for decades started moving faster than anyone expected. (Original in Russian.)
Read on Telegram →Geoscience finally gets its own foundation models
In 2025 the first specialized foundation models for geoscience appeared — trained on millions of real-world data points, they cover several interpretation tasks at once: well logs, seismic, lithology. A walkthrough of what is already working in production and how it slots into existing workflows. (Original in Russian.)
Read on Telegram →We will run the platform on one of your assets
Share a license or coordinates — we will pull the geology, estimate reserves, production profile and NPV. No NDA, no commitment — just a result you can compare with your own numbers.