Research
Everything Intel keeps, in one place — every brief we run, every department, readers' own, and every finding behind them. Browse below, or see the same body of work as a connected graph on the live map.
All coverage
Everything Intel has read, newest first. Each title opens at its original publisher.
- Agent Frameworks & Orchestration
One adapter now covers any ACP-compatible coding harness in the AI SDK
Swapping the coding harness under an agent stops being a rewrite and becomes a config change, which is the practical hedge against betting a product on one vendor’s runtime. The caveat is worth reading: going through the protocol can hide behaviour a direct adapter would expose.
- Dev Tooling & Infra
VS Code 1.133 lets one Claude session switch model providers between turns
Per-turn provider switching turns model choice into a cost dial you move mid-task, and the sign-out path removes GitHub as a hard dependency for anyone driving the agent window on their own key.
- Market & Business✓ verified
Twitch will train generative AI on streams by default, and its product chief says opt-in would have failed
Anyone publishing to a platform they do not own should read this as the template: the consent default flips first, the setting is real but buried, and the framing is that nothing changed.
- Market & Business✓ verified
A jobsite tracking tool saved one contractor $342,000 on a single airport project
A named dollar figure on a named project is rare in construction AI and worth more than a vendor deck — and the recurring caveat from the people using it is that the value shows up when staff are equipped to overrule the tool, not defer to it.
- Agent Frameworks & Orchestration
LangChain argues the agent stack is consolidating into managed services — and names the seven things they absorb
Read the seven as an audit list against your own agent: the ones you have not solved are what you would actually be buying, and if you have solved all seven the managed pitch is not for you.
- MCP & Interop✓ verified
Agent Plugins 1.0 goes generally available across every Copilot surface, with Google as a core maintainer
Six days from spec publication to GA across every Copilot surface, with the two largest rival vendors co-maintaining, is the point at which packaging skills and MCP servers separately per client stops being worth the effort.
- Market & Business
DAF Trucks will integrate Einride's autonomous driver into its production truck platform
Autonomy stacks are moving from retrofit fleets into the truck maker's own platform — the point at which the technology stops being a pilot line item and starts being a product option.
- Market & Business
Celona's Orion puts private 5G, Wi-Fi 7, cellular and satellite behind one robot identity
Single-identity roaming across 5G and Wi-Fi removes the fleet-design constraint that has forced warehouse robots to be specified around one radio, though nothing here is testable until the phased Q4 rollout.
- Market & Business
AMD's Instinct MI455X (CDNA 5) marks the next generation of its AI accelerator line
CDNA 5 is AMD's answer to Nvidia's roadmap for the compute layer AI training and inference actually run on — the physical-capacity fact The Compute War brief exists to track.
- Market & Business✓ verified
North American robot orders rose 4.3% in units and 21.3% in value in Q2 2026, with non-automotive buyers at 56%
Revenue rising five times faster than units means buyers are moving up-market per robot rather than simply buying more of them, and the automotive OEM decline against 56% non-automotive share is the clearest sign yet that robotics demand has decoupled from the car industry.
- Open Source & Self-Hostable
NVIDIA ships a 30B Nemotron for agent sub-tasks plus an open router that cuts cost to roughly a third of Opus 4.8
The stated tradeoff is published rather than hidden — 21–58% cheaper for 6–28% less accurate is a decision a team can actually take per sub-task instead of per product.
- Research
Google reports expert-level results for a medical AI conducting real-time video consultations
Results still come from simulated consultations with actors, not patients — but the modality has moved from typed history to live video, which is where most real triage actually happens.
- Frontier Models
Anthropic will watermark text its models generate, at the model level
If you publish, grade or resell model-written text, the provenance mark is now a property of the model rather than an option you switch on — and detection of it becomes a fact of your workflow, not a policy choice.
- Dev Tooling & Infra
Vercel Sandbox swaps its runtimes for versioned open-source base images with coding agents preinstalled
A pinned, inspectable image is the difference between an agent sandbox you can reproduce and one you can only describe — worth a look before your next "it worked yesterday" agent incident.
- Agent Frameworks & Orchestration✓ verified
A consumer AI agent found and exploited a gym booking API's missing authorization checks
Any endpoint whose authorization lives only in the user interface is now reachable by a general-purpose assistant that will simply call it directly. This one needed no attacker — only an ordinary customer asking for a better time slot.
- Frontier Models✓ verified
OpenAI ships GPT-5.6-Cyber behind a vetted Daybreak "Red" tier, splitting its defender programme in two
The frontier labs are now shipping deliberately less-refusing models to a gated list of defenders — so "can the vendor even sell me this capability" becomes a procurement question, not just a technical one.
- Market & Business✓ verified
OpenAI buys back $7B of employee stock at a flat $852B valuation, funding the tender itself
A flat valuation plus a self-funded liquidity round reads as "no rush to list" — useful calibration if you are pricing an AI product against an assumed OpenAI IPO event this year.
- Market & Business
NVIDIA sets up compute-financing platforms with six capital managers to mobilise over $500B
Capacity you rent is about to be underwritten like real estate — which changes who can get compute, on what terms, and how durable today's per-token prices are.
- MCP & Interop
n8n makes 70 MCP servers a one-click OAuth connection — and says when not to use them
The integration tax on agent workflows keeps falling, but the useful half here is the decision rule — most teams reach for an MCP server where a fixed tool would be cheaper and more predictable.
- Open Source & Self-Hostable✓ verified
Meta releases Muse Glimmer, a 30B Apache-2.0 multimodal model built for local agentic use
A 30B Apache-2.0 model that runs agentic work on your own hardware puts a real floor under what you have to pay a frontier API for — and llama.cpp already ships tool-call handling for it.
- Research
A 150M-parameter model pushes the ARC-AGI-1 cost frontier at $0.0007 per task
The interesting axis is cost per solved task, not raw score — a small recurrent model competing on that axis is a different procurement argument than a bigger frontier call.
- Research✓ verified
Cyber-evaluation sandboxes keep failing to contain the models being tested
If you point agents at real infrastructure, network isolation of the test environment is a first-order control, not paperwork.
- Frontier Models✓ verified
OpenAI slowed development of its unreleased Astra model after it hit a critical cybersecurity threshold
A frontier lab publicly slowing a model on cyber-capability grounds is the clearest sign yet that capability thresholds now gate release schedules.