← Back to Archive · Homepage - Newsletter

Web archive · this issue of MCC AI Weekly was sent by email on Thursday, August 27, 2026.

MCC AI Weekly · Thursday, 27 August 2026

Structure beats reaction, in AI and in leadership

MCC AI Weekly — issue cover illustration

This week's insights share a common warning against piecemeal fixes: whether it's a CIO's tech roadmap, a COO's workflow design, or a retention plan, treating symptoms case by case leaves the underlying structure untouched. The short news backs this up with hard numbers on agentic AI's rapid uptake. Read on for what deliberate architecture looks like in practice.

Insights

CIOs at top firms shift from IT management to strategy

McKinsey surveyed over 600 technology and business leaders for its Global Tech Agenda 2026. Nearly two-thirds of top-performing companies say their tech leaders are "very involved" in crafting enterprise strategy, versus 52 percent elsewhere. These firms are also replacing annual budget planning with continuous, capability-led practices, using agentic automation and data productization to drive higher EBITDA growth.

So what

Move CIOs beyond annual budget cycles toward continuous, capability-led planning tied to specific value pools like growth and margin. Where tech leadership stays confined to modernizing infrastructure, it will lag firms that treat technology as a strategy function.

McKinsey →

Agentic AI Turns COOs Into Enterprise Orchestrators

Boston Consulting Group finds agentic AI is moving operations from periodic optimization to continuous orchestration, redefining the COO's role. Leading deployments improve on-time-in-full service by 2-3%, cut shipment logistics costs by 15-20%, and lift inventory performance by 10-25%. The gains come from connecting previously disconnected point solutions, such as forecasting engines and logistics optimizers, so the whole system responds as one.

So what

COOs should redesign workflows end-to-end around AI agents, not layer them onto existing point solutions. The value comes from connecting forecasting, procurement, and logistics into one system, not from adopting tools in isolation.

Boston Consulting Group →

China's tech model: entrepreneurs first, state scales second

Baiguan founder Robert Wu rejects the idea that China's tech industries are built through top-down state planning. Instead, he describes a hybrid sequence: entrepreneurs discover an opportunity first, and the state later supplies policy support and capital to scale it. Wu uses EVs and DeepSeek as examples of this pattern.

So what

Leaders assessing Chinese competitors or partners should track where entrepreneurial activity is already emerging, not where state policy is announced. Policy and capital tend to follow proven opportunities, rather than create them.

AI Proem (Grace Shao interview with Robert Wu, Baiguan) →

Retention Needs a Framework, Not Counteroffers

Researchers writing in the Journal of Management find companies typically respond to attrition risk with one-off measures — counteroffers, culture initiatives, noncompetes — that treat symptoms, not causes. OpenAI's roughly $1.5 million average stock-based compensation per employee in 2025 still could not stop defections to rivals. The authors argue what's missing is a systematic architecture of employee mobility barriers, not more pay.

So what

Leaders should build a deliberate architecture of mobility barriers rather than reacting with pay increases or noncompetes case by case. The article's point is that episodic fixes address symptoms, leaving the underlying causes of departure untouched.

MIT Sloan Management Review →

Weekly Short News

Google wins Spirit Airlines data auction for $10 million

Google won a bankruptcy auction for Spirit Airlines' internal business data, bidding $10 million. The deal beat a $7.5 million offer from AI recruiting firm Mercor.io. The trove includes about 100 million deidentified emails, 500 million Microsoft Teams messages, source code and operational records. A federal bankruptcy judge must still approve the sale.

Bloomberg Law →

Goldman Sachs research on AI and jobs not available

No article content on Goldman Sachs' AI employment research was provided in the source material retrieved. The available text consists of an index of unrelated Neuron newsletter headlines from August 2026.

The Neuron →

Top-tier model usage flat at 6% of business tokens

Exponential View reports that usage of Fable 5, a frontier model, has stayed flat among businesses at 6% of all tokens and 11% of spend. Separately, the top 10% of companies using OpenAI's products consume 8.3 times more tokens than the typical firm.

Exponential View →

Invesco CEO on AI strategy and trust

McKinsey published a podcast episode featuring senior partner Eric Kutcher interviewing Invesco President and CEO Andrew Schlossberg. Schlossberg discusses putting AI directly into employees' hands rather than dictating its use from the top. He says asset managers must move faster on AI without eroding the client trust built over decades.

McKinsey →

Agent token use grows 14x since February, Exponential View finds

Exponential View reports AI agents began using more tokens than humans in February this year. Since then, agent token usage has grown 14x, while human token usage grew 2.8x over the same period.

Exponential View →

AI Player of the Week

Wispr Flow

Wispr Flow is a voice-to-text application that cleans up spoken input, removing filler words, fixing false starts and adding punctuation, then delivers polished text into whatever app the cursor is in, from Slack to a code editor. The company reports users working at roughly 220 words per minute versus 45 typed, and raised $81M to extend the approach into a broader "Voice OS". It suits professionals who draft constantly, such as managers writing messages, notes and emails throughout the day, less so anyone whose work rarely involves text output.

Wispr Flow →

Tip of the Week

Stop Re-Briefing the Model Every Chat

A new chat starts from zero: role, team, writing style and recurring context must be retyped, or the output drifts generic. A Project, or custom instructions, saves that briefing once and applies it to every conversation inside it, including reference files the whole team needs. For a manager running the same three or four analyses each week, set the Project up once. Then pick up where last week's session left off.

Found this useful? Forward it to a colleague — they can subscribe here.

You are receiving this email because you confirmed your subscription to MCC AI Weekly. Questions? Just reply to this email — it reaches us directly.

Imprint · Privacy Policy