The Future Is Orchestration

I round up the most relevant AI-in-finance news - the deals being done, who’s rolling out what, and what’s actually working on the front lines

…Throwing money at the wall and seeing what sticks. There's going to be a lot more of that over the next couple of years.

Why years of learning are about to compress into months. And the playbook for success in 2026 crystallises: orchestration from a central interface, structured repositories of information, and agents that actually execute tasks. That's the stack., a conversation I had over Christmas dinner that I can't stop thinking about.

In This Week’s Issue:

From The Trenches

The Future Is Orchestration

I've been spending a lot of time in Claude Code lately. Not asking it questions. Working inside it.

My email flows through Claude now. So does my Notion, my Drive, my CRM, my newsletter platform, my Slack, my ActiveCampaign. Everything talks to everything else. And I control it all from one place.

Here's what that actually looks like. I ask Claude to pull my emails. We triage together. I decide what needs a response. Claude drafts it, I approve, it sends. I never open Gmail. Opening Gmail now feels like using a fax machine.

But the real magic is the interconnectedness. When I update something in Notion, I can tag a teammate in Slack about it from the same interface. I can attach a memo to that Notion entry, or write the memo from scratch, or create an image to go with it. All without leaving Claude. The systems talk to each other through me, but I'm not the one clicking between tabs.

Here's the mental model I keep coming back to. You're going to hear this narrative a lot in 2026:

Central orchestration + Repositories of information + Agents to execute tasks.

Let me break that down.

Central orchestration is the command centre. One interface where you direct everything. You're not logging into five different platforms. You're not copying and pasting between tools. You're sitting in one place, giving instructions, and the systems respond.

Repositories of information are your structured data sources. Not a graveyard of files in Box with 50 versions of every model. Actual organised information that AI can reliably pull from. Your CRM. Your deal files. Your portfolio data. Structured, tagged, accessible.

Agents are the doers. They don't just answer questions. They take actions. Send that email. Update that record. Pull that data. Create that report. They operate the platforms on your behalf.

This is what we've been building towards at DealSage. The repository of deal information, structured and accessible, with agents on top that execute specific workflows. We've got some exciting updates coming soon with agents running directly in the platform. More on that soon 👀

Here's what I think is underappreciated about this model. The biggest barrier to new technology is always the same: "We have to learn a new platform" or "We have to migrate all our data." That's what kills adoption.

But if you can integrate everything and have agents operate the platforms, that hurdle disappears. You don't have to learn a new interface. The agents do the work on your behalf and surface the information you need. You're not clicking all over the place. You're just asking for what you want.

Think about what becomes possible:

All of that run by agents, without your direct involvement if you want.

This isn't some future state. This is possible now. Today.

The playbook for 2026 is clear. The firms that figure this out will have a structural advantage. The firms that don't will wonder why their competitors seem to have discovered time travel.

News Digest

Meta Acquires Manus for $2B+ in 10-Day Sprint

Meta closed its acquisition of Manus, the Singapore-based AI agent startup, for over $2 billion on December 30. The deal reportedly took just 10 days from initial discussions to close. Manus hit $100M+ ARR in just 8 months after launch, processed 147 trillion tokens, and created over 80 million virtual computers for users.

The company was founded by Chinese entrepreneurs and backed by Benchmark (at a $500M valuation), Tencent, and ZhenFund. As part of the deal, Meta is severing Manus's Chinese ownership ties and discontinuing China operations. Around 100 Manus staff are joining Meta, with CEO Xiao Hong reporting to COO Javier Olivan.

This is Meta's fifth AI acquisition of 2025, following PlayAI, WaveForms, Rivos, and Limitless. It signals Meta's broader pivot from open-source LLaMA models to closed-source alternatives (codenamed Avocado and Mango).

The details:

Why it matters: The speed tells the story. Ten days. For a $2B+ deal. Meta saw the agentic AI market forming and moved before anyone else could bid.

My take: This is the AI talent land grab in action. Meta has been vocal about feeling behind in the AI race. Their solution? Throw money at the problem. Manus was one of the only popular consumer-facing AI products that wasn't already locked up by investors or acquirers. It wasn't huge in the US, but it had genuine global traction. Meta saw the window closing and moved before anyone else could bid. Ten days from first call to close. That's not due diligence. That's a talent acquisition with a product attached.

Read more

What The Builders Are Saying

The cutting edge of AI isn't in press releases. It's on X/twitter. Here's what the people building this said this week.

@rakyll (Jaana Dogan, ex-Google engineer) on Claude Code vs. a year of Google engineering

Why this matters: This is what's so fun about the AI world right now. Everyone is here, developing, living in it as we speak. Google and Anthropic are direct competitors. But some advances are just too amazing not to give your flowers. A Google engineer publicly admitting that a competitor's tool replicated a year of internal work in an hour? That's the moment we're in.

@arohan (ex-Google and Meta engineer, Gemini co-author) on agentic coding

Why this matters: AI collapses the learning curve. It turns junior engineers into senior engineers dramatically fast. New-hire onboarding on large codebases shrinks from months to days. What used to take hours of Googling and Stack Overflow is now a single prompt. Agency is all you need now.

My take: I made this same point about finance 3-4 months ago and got a lot of polite nods and raised eyebrows. As I've been saying for a few weeks now, tech leads the rest of the world by about 18 months. Everything we're seeing there is coming to other industries pretty soon. The analyst learning curve will compress from 2 years to 2 months. The question isn't whether this happens. It's whether you're positioned for it when it does. Or whether you're the one training your replacement.

This Week in AI M&A:

SoftBank acquires DigitalBridge for $4B (Dec 29) - Part of SoftBank's AI infrastructure buildout supporting the OpenAI/Stargate project. DigitalBridge manages $80B+ in digital infrastructure assets globally. Same week SoftBank completed its final $22B payment to OpenAI, bringing total investment to ~$40B for >10% stake at $260B valuation. Masa isn't betting on AI. He's betting the entire company, his reputation, and probably his lunch money on it.

Other Cool Stuff I’ve Read of Seen This Week:

Axios: AI in 2026 is the "show me the money" year (Jan 1) - Menlo Ventures partner: "Enterprises will need to see real ROI in their spend." EY's tech leader: "Boards will stop counting tokens and pilots and start counting dollars." The vibes-based budgeting era is officially over.

TechCrunch: VCs predict AI coming for labor in 2026 (Dec 31) - Battery Ventures: "2026 will be the year of agents as software expands from making humans more productive to automating work itself." MIT study found 11.7% of jobs could already be automated. Somewhere, a middle manager just felt a chill.

Forrester: Half of AI layoffs will be quietly rehired (Dec 2025) - 55% of employers regret laying off workers for AI. Klarna replaced 700 employees with AI, quality declined, customers revolted, humans rehired. Plot twist: half of AI-attributed layoffs will be rehired offshore at lower salaries. The "efficiency gains" were the friends we made along the way.

Fortune: AI startups that could be acquisition targets in 2026 (Dec 24) - M13: "I predict attractive prices for assets like Factory, Codegen, Wrap." Foundation model companies will gobble up application layer companies with proven PMF. Cursor might stay independent. Everyone else should update their LinkedIn.

TechCrunch: In 2026, AI moves from hype to pragmatism (Jan 2) - AT&T chief data officer: fine-tuned small models will be the big trend. "If fine-tuned properly, they match larger models in accuracy and are superb in terms of cost and speed." The hangover has arrived, but the party wasn't a complete waste.

Acquisition Intelligence is a weekly newsletter on AI in M&A for finance professionals, private equity investors, investment bankers, corp dev teams, and deal-makers.

For questions, feedback, or to share what you're seeing in the market, reply to this email.

P.S. I'm Harry, co-founder of DealSage. We're building an AI-native deal intelligence platform to help professionals turn their institutional knowledge into better decisions. If you're curious what we're up to, check out dealsage.io or just reply here