← All briefings

The briefing · 21 Aug 2026 · 105 tracked changes

AI Tools Get Price Adjustments and Advanced Autonomy Upgrades

Today\'s technology landscape sees a reshuffling of entry pricing alongside significant upgrades to autonomous agents and hardware cooling.

A Shift in Entry-Level Software Pricing

Several prominent AI and data platforms announced substantial updates to their entry-level pricing. ThoughtSpot Spotter saw its entry price rise from $0.1 to $25 per month, while writing assistant Jasper raised its starting tier from $59 to $69 per month, and Copy.ai increased its base plan from $24 to $29 per month. Conversely, educational assistant Khanmigo dropped its entry price to free, and IBM Quantum Platform lowered its entry tier from $2,880 to $48 per month.

Why it matters Fluctuating entry costs indicate that developers are still finding the right balance between software access and infrastructure cost.

Built from 5 tracked changes

Agents and Robotics Gain Practical Control

Software agents are gaining more direct physical and digital control. OpenAI\'s ChatGPT Agent added computer-use and voice capabilities, while Zapier Agents introduced direct task-based automation alongside thousands of application connections. In robotics, Skild Brain introduced omni-bodied navigation and grasping capabilities, while Boston Dynamics\' Atlas added intelligent autonomy to its physical repertoire.

Why it matters AI is transitioning from a conversational partner into an active assistant that can control computers and physical hardware directly.

Built from 4 tracked changes

Compute Infrastructure Focuses on Cooling and APIs

Next-generation processing hardware is leaning heavily into thermal management and custom infrastructure. Google\'s TPU7x (Ironwood) and NVIDIA\'s Blackwell chips both added liquid-cooled capabilities to manage intense computing heat, a feature also adopted by Tenstorrent\'s Blackhole. Additionally, SambaNova\'s SN40L launched an inference platform with an OpenAI-compatible API to ease the integration of its specialized chips.

Why it matters The physical demands of training and running AI models are forcing hardware makers to design more efficient cooling systems and easier software access points.

Built from 4 tracked changes

See the raw change log