AI News — July 24, 2026

AI-assisted SDLC

• [2026-07-22] OpenAI — Presence packages guardrails, evals, and a Codex-driven improvement loop for production agents. OpenAI introduced Presence, an enterprise platform for deploying customer-facing voice and chat agents — bundling policies and standard operating procedures, approved actions, guardrails, simulations, and evaluation tools so a team can connect company systems, define agent behavior, test it, and manage changes after launch. The distinguishing piece for developers is the Codex-powered improvement process: OpenAI’s Codex coding agent reviews production interactions, identifies weaknesses, and proposes behavioral changes that staff must test and approve before they go live — a closed AI-in-the-loop maintenance cycle rather than a one-shot deployment. OpenAI backs it with its own numbers: Presence runs its English-language phone support line, where it resolves 75% of inbound calls without a human, and the Codex feedback loop cut human handoffs by 15 percentage points within 10 days of deployment. It ships as a limited general-availability program (not self-service), and OpenAI has not disclosed pricing (VentureBeat says it asked twice without an answer). It matters because it reframes OpenAI’s enterprise pitch from raw model access to a managed system for building, governing, and continuously improving production agents — with a coding agent, not humans, driving the iteration. (official, source)

  Presence is gated — limited GA, not self-service, no published pricing — so for most teams the takeaway isn’t the product but the pattern it productizes: a coding agent reading real production transcripts and proposing behavior fixes for humans to approve. Whether that closed loop generalizes beyond OpenAI’s own stack is the thing to watch, and the 75%-resolution and 15-point-handoff figures come from OpenAI’s own support line and are self-reported, not independently audited.

Watch list

• Kimi K3 open weights (July 27) — Moonshot has K3 live via API (kimi-k3, $3/$15 per million) but says the 2.8T weights go public July 27, reportedly under a modified-MIT-style license (unconfirmed until the model card ships). Watch for the model card, the license’s rumored monthly-active-user clause, whether third-party inference providers stand up endpoints undercutting Moonshot’s own pricing, and the first independent, non-arena evals. (prior coverage)

  With Moonshot hosting K3 at $3/$15 while planning to open the weights, July 27 sets up an unusual test of its own pricing: if third-party providers stand up cheaper endpoints on the released weights within days, Moonshot ends up competing against its own model — the first real read on whether an open-weights drop and a paid hosted-API business can coexist at these token prices.

• Gemini 3.5 Pro — still undated even as the Flash line ships. With 3.6 Flash and its siblings out (July 21) while 3.5 Pro remains without a date or model card (“coming soon” on the Gemini model page), the “Flash ships first” tell the last several briefings tracked has played out — the pipeline is clearly moving, but the flagship is still missing. The headline specs (2M-token context, Deep Think, pricing) remain reported, never confirmed. The signal worth revising a plan around is a Google post carrying an actual 3.5 Pro date and a published model card, not another leaked target. (unconfirmed) (report, prior coverage)

  For anyone who’s been holding a model decision for 3.5 Pro, the useful shift now is to stop treating it as imminent: with Flash shipped and Pro still dateless, the planning-safe assumption is that it isn’t arriving this month, so pick from what’s actually available — 3.6 Flash or a competitor’s flagship — and treat a real Pro date as upside rather than a milestone worth blocking on.

• MCP spec finalization (July 28) — the 2026-07-28 release candidate (stateless core, Extensions framework, Tasks, MCP Apps, authorization hardening) is in its final validation window, with the Python/TypeScript/Go/C# SDKs updating against it in beta (Python mcp 2.0.0b1; TypeScript split into new @modelcontextprotocol/server and @modelcontextprotocol/client 2.0.0-beta.1 packages; Go v1.7.0-pre.1; C# 2.0.0-preview.1). Security analysts continue to flag the stateless rework’s new surface — client-held state manipulation and a “hit-and-run” async-task DoS — on top of the Mcp-Method/Mcp-Name routing headers. Watch which of the four SDKs reaches parity last and any final header changes before the RC locks. (official, prior coverage)

  The detail that bites TypeScript users specifically is the package split: the single @modelcontextprotocol package is becoming separate @modelcontextprotocol/server and @modelcontextprotocol/client 2.0.0-beta.1 packages, so upgrading isn’t a version bump but an import rewrite — worth pre-staging on a branch against the beta now rather than discovering the two-package layout the day the RC locks.