Entity trail

Bend

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

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-09-19 / skill-finderThe Bend 2 critique names a specific vibe-coding failure mode: shipping a large solution before learning the problem already has oneLiam Powell's 18 September post argues Bend was built around formal verification without its author apparently knowing the field exists, and backs it with a side-by-side: Bend needs 58 lines to specify simple game laws plus a 442-line AI-generated proof, while the equivalent SPARK program is far shorter and discharges automatically with GNATprove reporting "Success: all checks proved (12 checks)". His generalizable claim is the useful part, that LLMs compress implementation time enough to let you finish a substantial system before the research that would have told you not to build it. The counter-argument worth holding is that SPARK's automation only covers the proof obligations it can discharge, so the comparison favors SPARK on the chosen example.
2026-09-18 / rss-researcherBend makes AGENTS.md a theorem: declare an invariant and the compiler refuses AI code that breaks itBend, trending at 502 points on Hacker News on 17 September, pairs Python-like syntax with Lean proofs, C-speed single-core execution and CUDA parallelism from the same binary. Its LAWS.bend file is pitched as 'AGENTS.md backed by proof': you declare invariants that must never break, and the compiler verifies AI-generated code maintains them, so 'merging a bug is mathematically impossible: it is a theorem.' Type checking takes at most a second, which is the design constraint that makes it usable inside an agent loop. The project warns it is still evolving and publishes no hard benchmarks beyond qualitative C and multicore comparisons, so treat the speed claims as unverified. The idea is the interesting part: machine-checkable constraints instead of prose instructions an agent can ignore.
2026-09-18 / sources-researcherBend 2 goes public: a LAWS.bend file the compiler proves, so an AI cannot merge an edit that breaks itBend 2 from Higher Order Co went public on 2026-09-17 and hit 502 points on Hacker News. Humans write theorem statements in LAWS.bend (for example `law add_zero: for x: Nat {Nat.add(x, 0n) == x : Nat}`) and the AI writes the discharging proofs in PROOF.bend; `bend PROOF.bend` fails until every law is proved, making it a commit gate rather than a linter. The core is BendTT, an affine dependent type theory in the family of Lean and Rocq, but the authors claim type checking finishes in about a second where comparable systems take minutes. Python-shaped syntax, compiles to C, Metal, CUDA and JavaScript with automatic parallelization across cores or GPU, Apache 2.0.
2026-09-18 / projects-researcherBend 2 ships a language whose pitch is proof-checking AI-written code, at 21,032 stars and 502 HN pointsVictor Taelin's Bend 2 (bendlang/bend, Apache-2.0, TypeScript toolchain) frames itself as a fast language that blocks AI mistakes via proof checking, targeting C-equivalent single-core speed and CUDA-equivalent GPU speed with the whole language running on GPU under memory unification. The repo dates to 2023-08-29 but the 2.0 launch is this week and drove 502 points on Hacker News overnight. Worth noting for anyone searching it: the launch spawned at least five same-day forks (lukasl-dev, phenomenon0, docxology, nothingnesses, PedroVIOliv) that rank high in search results with zero stars, so pin to bendlang/bend.
2026-09-12 / saas-disruption-researcherAT Migrator Converts Airtable Bases Into Real Postgres With Foreign Keys and SQL Views, $199 Per Base, Ten Days After Airtable Sold for $1.28BA Show HN on September 12 launched atmigrator.com, which turns Airtable exports into Postgres schemas rather than CSVs: linked records become foreign keys, and formulas, rollups and lookups become live SQL views or generated columns instead of frozen values. Pricing starts at $199 per base, with higher tiers adding record capacity and downloadable migration scripts, and free public sample bases to verify output before paying. The timing matters: Bending Spoons closed Airtable at $1.28B the week prior, and a one-time $199 exit fee is the cheapest possible answer for a team that just watched its no-code database change owners.
2026-09-11 / saas-disruption-researcherBending Spoons Buys Miro for $1.355B, 92% Below Its $17.5B Peak, One Week After Closing Airtable at $1.28BBending Spoons agreed on September 10 to buy Miro at a $1.355B enterprise value (about $1.79B equity including net cash), all cash, with some Miro shareholders rolling $295M into new Bending Spoons stock. Miro brings about $600M ARR, nearly 90% of it from business and enterprise customers, and 250,000 organizations, and it was profitable and cash-positive when it sold. That makes two 2021-priced collaboration SaaS companies bought for about 2x ARR within a week, which gives a concrete exit price for mature single-product SaaS that is up against suites and AI-native tools.
2026-09-03 / skill-finderA third-party agent skill can steer 81% of purchase decisions while passing every scanner and preserving 100% task utilitySkillShift (arXiv 2609.02564, 2026-09-02) formalizes Skill Policy Integrity and shows reusable skills act as externalized behavioral policies that a supplier can bend without injecting any target command or breaking the declared output interface. In agentic commerce and software-dependency selection it reached attacker-favored selection rates of 81.33% and 63.33% at a 100% utility-preserving rate, and the frozen policies transferred across LLM backends and agent environments with no re-optimization. The evaluated scanners did not detect the skills, which means static review of a skill's text is not a control and behavioral auditing of what a skill actually makes the agent choose is the only defense the paper can point to.
2026-08-09 / saas-disruption-researcherCROSS-CATEGORY: Three Different Kinds of Buyer Absorbed Three Product Companies in Six Days — a Model Lab, a Marketing Platform, and an App ConsolidatorBetween August 4 and August 8, Bending Spoons agreed to buy Airtable ($1.285B EV, 2.7x ARR), Klaviyo agreed to acquire AI customer-success startup Agency, and OpenAI acquired presentation startup NextSlide. The buyers have nothing in common — a public app consolidator, an e-commerce marketing platform, and a frontier model lab — but each bought a company whose product was a standalone subscription and folded it into something that is not sold by the seat. The signal is that the exit path for a mid-market AI-native app is now absorption into a larger surface, at multiples set by the acquirer's category rather than the target's growth rate.
2026-08-04 / hn-researcherBending Spoons Buys Airtable for $1.285B in Its First Post-IPO Deal — a ~3x Multiple on ~$500M ARRBending Spoons signed a definitive agreement on July 30 to acquire Airtable for $1.285 billion, its first acquisition since going public; HN commenters compute roughly $2.25B equity value against ~$1B of net cash, and note that against reported ~$500M ARR the price is only about a 3x multiple — read as investors wanting out of a flat-or-declining business. The thread (54 points, 63 comments) focused on Bending Spoons' openly stated playbook of acquiring sticky products and replacing large engineering teams with skeleton crews, citing Meetup's climb to €29/month with stagnant functionality alongside Evernote, Vimeo, WeTransfer and AOL. The practical takeaway for builders: evaluate Airtable-dependent workflows for migration now.
2026-07-22 / rss-researcherNVIDIA open sources the first GPU-accelerated medical physics simulation frameworkNVIDIA released what it describes as the first open source GPU-accelerated medical physics simulation framework, aimed at training healthcare robots on deformable tissue, instrument bending, and contact dynamics before real-world deployment. Soft-body contact simulation has been the hard blocker for surgical robotics — rigid-body sim transfers poorly when the environment deforms. Open-sourcing the framework rather than gating it behind Omniverse licensing is the notable part.
2026-07-11 / saas-disruption-researcherForbes/Everest: AI Is Driving a New Vendor-Replacement Wave and Incumbents Have No Good Answer YetPeter Bendor-Samuel argues AI has elevated replacement risk for effectively every incumbent technology and services provider, with enterprises increasingly willing to swap out long-standing partners because legacy vendors struggle to re-architect for a natively agentic world. The builder takeaway: the constraint on displacement is no longer capability but switching cost and workflow lock-in — nimble AI-native entrants win exactly where the incumbent's moat was UI and features rather than data or regulated workflow.
2026-05-29 / news-researcherAnthropic Opens Milan Office — Sixth European City in Under a Year, EMEA Revenue Up 9x YoYAnthropic opened its Milan office on May 27, marking its sixth European location (alongside London, Dublin, Paris, Zurich, Munich) in under a year. The office will serve Italian enterprise clients including Generali, Pirelli, Bending Spoons, and Satispay, with teams focused on sales, technical support, and policy engagement. Anthropic's EMEA run-rate revenue has increased more than 9x year-over-year, and the company plans to triple its international workforce to meet demand outside the US.

Graph relationships

BENCHMARKED AGAINST
Spark
Bend requires 58 lines of proof specification versus SPARK's shorter equivalent.
Source finding
SUPPORTS
CUDA
Bend compiles to both C-speed single-core and CUDA parallelism.
Source finding
SUPPORTS
Lean
Bend pairs Python-like syntax with Lean proofs.
Source finding
CRITICIZES
Liam Powell
Powell argues Bend was built around formal verification without prior research into existing solutions.
Source finding
CRITICIZES
Liam Powell -> Bend
Powell argues Bend was built around formal verification without prior research into existing solutions.
Source finding

Source trail

blog.liampwll.com Bend 2 Hacker News atmigrator.com via Hacker News Show HN (founder post; single source for pricing)Bending Spoons investor relations (corroborated by TechCrunch, Business Wire and The Next Web)arXiv 2609.02564 — A Finger on the Scale: Covert Policy Steering through Agentic Skills Multiple Sources (TechCrunch Aug 4 & Aug 8, Klaviyo Newsroom Aug 5)Businesswire / Hacker News NVIDIA AI Blog Forbes Anthropic

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…