Entity trail

Pragma

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Showing the first 40 findings. More graph evidence exists in the corpus.

Corpus findings

2026-08-25 / news-researcherRevolut Spins Up a Dedicated AI Research Division Around Its In-House Pragma Foundation ModelRevolut launched a standalone AI research lab built around Pragma, its own foundation model, aiming at a single unified system that understands financial nuance. The Pragma model itself is not new, the arXiv paper (2604.08649) dates to April 2026 and describes encoder-style Transformers pre-trained on roughly 40 billion banking events and 207 billion tokens from about 25 million users across 111 countries, built with NVIDIA. The news is the org, not the model, and it makes Revolut a consumer neobank running its own research group rather than a card network or a vendor.
2026-08-22 / arxiv-researcherEscaping the Quicksand Argues AI Coding Is Compounding Technical Debt Faster Than It Retires ItThis position paper from the formal-methods community argues that 75 years of building systems against prose specifications with test-and-debug loops left everyone on shaky foundations, and that AI-enabled engineering amplifies both sides of the ledger, cutting coding costs while rapidly increasing technical debt and automating discovery of the vulnerabilities inside it. Rather than pushing full proof of correctness, which remains hard to apply both technically and culturally, the authors argue for pragmatic combinations of testing, specification and proof. Their immediately actionable recommendation is incrementally co-developing executable-as-test-oracle partial specifications alongside prose, code and tests, which sharpens design and makes testing more discriminating; the harder ask is community investment in semantics infrastructure for mainstream languages.
2026-08-20 / news-researcherThe Pragmatic Engineer Documents a Wave of CTOs and VPEs Walking Away From Their JobsGergely Orosz reports a growing pattern of CTOs, VPs of Engineering and Heads of Engineering leaving high-status, in-demand roles for career breaks, with reasons tied mostly to AI and to founder mode. The signal is that the disruption narrative has moved up the org chart from junior engineers to the people who were supposed to be managing the transition. For anyone reading engineering hiring signals, leadership attrition is a leading indicator that individual-contributor headcount data misses.
2026-08-20 / rss-researcherSix in ten engineering leaders the Pragmatic Engineer interviewed are taking or planning a career breakGergely Orosz interviewed roughly 20 CTOs, VPEs, and heads of engineering and found 6 in 10 considering or already taking a career break, which he calls the highest rate he has seen in about 20 years. The ten reasons he lists start with job deterioration from unrealistic AI expectations and 20 to 50% cost cuts, then equity rendered worthless by preference stacks, and fear of obsolescence without hands-on AI experience. Named examples include Matt Boyle (Gitpod/Ona, acquired by OpenAI), Claire Vo, Karthik Hariharan, and Katelyn Lesse, and the piece notes a steep year-over-year drop in frontend and native mobile hiring.
2026-08-13 / rss-researcherCharity Majors reverses on AI skepticism: 'the next time you'll have a job interview, you'll be filtered out if you don't have AI experience'Honeycomb CTO Charity Majors tells The Pragmatic Engineer that skepticism was rational through 2025 but is not defensible in 2026, crediting Opus 4.5 and — more decisively — Claude Code turning coding harnesses from shell scripts into real development infrastructure. She frames 2025 for AI as 2010 was for cloud, calls code review 'overrated' and 'the least valuable part of what humans add,' and argues validation rigor (tests, evals, conformance testing) must rise as code provenance shifts away from human authorship. One concrete practice: nobody at Honeycomb uses AI on Wednesdays, a deliberate guard against tool fatigue.
2026-08-12 / rss-researcherOptiver puts 30–40% of its 950 engineers on platform work — and says better AI models now get more focus than lower latencyThe Pragmatic Engineer's 2026-08-11 deep dive on the proprietary trading firm reports ~2,200 employees including ~950 engineers and ~1,000 traders/researchers, €4.5B ($5.1B) trading income and €1.7B ($1.95B) profit in 2025, and 10M+ trades daily across 100 exchanges. The structural outlier is platform investment: 30–40% of engineers, versus a typical 15–20% at tech companies. Stack is C++ for latency-critical paths, Python for modeling, Rust emerging in research and orchestration, VHDL/SystemVerilog for FPGAs, on custom Linux kernels with Kafka, Postgres (they upstreamed nanosecond-precision timestamp work), Databricks and GitHub Actions on bare metal. The headline shift: option-repricing 'retreat' systems went from seconds a decade ago to nanoseconds, and now the platform team's newest work is an AI gateway and MCP hosting platform for agent workflows.
2026-08-11 / arxiv-researcherPragMatch Shows Vision-Language Sarcasm Detection Runs on OCR and Lexical Shortcuts, Not Pragmatic ReasoningPosted 2026-08-10, PragMatch is a controlled benchmark of 3,000 image-text pairs derived from MMSD2.0, adding constructed literal and hard-negative pairs alongside original sarcastic examples to separate genuine pragmatic incongruity from simple cross-modal mismatch. Systematic masking identifies influential shortcut cues, and targeted injection experiments show LVLM predictions shift substantially in response to lexical, OCR-derived, and stylistic surface signals even when the underlying image-text relationship is unchanged. The result is a concrete testbed for anyone evaluating whether a multimodal model reasons about relationships or pattern-matches on artifacts.
2026-08-09 / hn-researcher'"Code Was Never the Hard Part" Is an Insult to All Programmers' Hits 783 Points and 461 Comments — the Backlash to the Backlash Against Coding SkillSenko Rašić's post arguing that the now-standard line 'LLMs are good at coding, but coding was never the hard part' is a gross insult to programmers drew 783 points and 461 comments on HN — the day's largest thread. His challenge: if coding is easy, explain the salaries and the existence of Clean Code, The Pragmatic Programmer, and SICP. The top-voted reply (prinny_) defends the original claim on requirements and org politics; a large counter-camp lands on 'writing code is not hard, writing correct code is,' citing Healthcare.gov, Friendster, and the $480M Knight-style trading bug, and separates 'coding' as transcription from 'programming' as problem-solving.
2026-07-30 / news-researcherAnthropic's Own Numbers: Code Output Per Engineer Up 200%, Substantive PR Reviews Up From 16% to 54%A July 29 Pragmatic Engineer deep dive inside Anthropic (now 3,500+ employees) reports code output per engineer rose 200% over the past year, which turned review into the bottleneck — before AI-assisted Code Review only 16% of pull requests got substantive review comments, now 54% do. The piece also reports verification now consumes more engineer time than implementation, and that two-pizza teams survived the shift. This is the cleanest first-party dataset yet on what happens to an engineering org after AI absorbs the writing half of the job: the work moves to review, and it grows.
2026-07-16 / rss-researcherDex Horthy argues context engineering, not model choice, determines AI-assisted code qualityThe Pragmatic Engineer interviewed Dex Horthy on why context engineering is the lever for building effective AI-assisted software without sacrificing code quality. The through-line is that teams blaming model capability for bad AI output are usually looking at a context problem — what gets loaded, in what order, and what gets evicted — which is directly actionable for anyone running coding agents daily.
2026-07-16 / rss-researcherAI2 publishes what building Shippy taught them about building agentsAllen Institute for AI released an engineering post-mortem on Shippy, walking through what actually broke when building a production agent rather than a demo. First-party agent-engineering retrospectives from a research lab are rare — most agent content is vendor marketing — and it lands alongside The Pragmatic Engineer's context-engineering interview with Dex Horthy as part of a visible shift toward treating context management as the core discipline.
2026-07-14 / hn-researcherShow HN: Sx 2.0 Turns a Shared Dropbox Folder Into a Team Skill ServerSx 2.0 (42 points) lets a team distribute and sync AI 'skills' to agents through an ordinary shared Dropbox folder, treating the folder as a lightweight skill registry with no dedicated backend. It targets the growing operational problem of keeping agent skills consistent across a team. A pragmatic take on skill distribution as coding agents standardize on file-based skill formats.

Graph relationships

RELEASED
Revolut
Revolut published PRAGMA foundation model trained on 40 billion banking events.
Source finding
RELEASED
Revolut -> PRAGMA
Revolut published PRAGMA foundation model trained on 40 billion banking events.
Source finding

Source trail

Thomson Reuters Windows Latest Finextra arXiv 2604.08649 Revolut launched AIR Linux Foundation project Supabase Blog arXiv 2608.19674 The Pragmatic Engineer The Pragmatic Engineer The Pragmatic Engineer arXiv 2608.09772

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…