Per user per month, Microsoft is raising M365 Business Standard from $12.50 to $14 on July 1. You get Copilot Chat with AI agents for Word, Excel, and PowerPoint, whether you asked for them or not. Meanwhile, most teams are also paying for ChatGPT Plus, Claude Pro, or Perplexity on top. The question that follows: if your existing software keeps getting smarter, do you still need those separate AI subscriptions?
The answer comes down to three things that have nothing to do with technology.
Why is everyone bundling AI into existing software?
Software companies are worried their customers will migrate to standalone AI tools. That is the core of it. If you figure out that ChatGPT writes better email responses than Outlook's suggestions, and Claude summarizes contracts faster than Word can, why keep paying for the full M365 package?
The answer from Microsoft, Google, and Adobe is predictable: build in the same AI capabilities, raise the price slightly, and make it feel so natural that you stop asking the question. Microsoft is moving most aggressively. From July 1, 2026, every M365 Business plan includes Copilot Chat with AI agents that summarize Word documents, write Excel formulas, and build PowerPoint presentations. Business Basic rises from $6 to $7 per user per month, Business Standard from $12.50 to $14.
Think of it like mobile carriers bundling Netflix and Spotify starting in 2019. The standalone apps still exist. But many people just use what the carrier includes. That same consolidation is now happening with AI inside business software.
According to Eurostat's 2025 Digital Economy and Society report, 13.4% of EU enterprises have adopted AI, with adoption rising sharply at companies with 250-plus employees (above 40%) compared to micro-businesses (around 11%). The gap is not just about budget. Larger companies are more likely to already have enterprise software with AI built in. Smaller businesses tend to reach for standalone tools like ChatGPT first.
What can built-in AI do, and what can it not?
Built-in AI has one big advantage: context. Copilot in Excel knows your spreadsheet. Gemini in Gmail has access to your email history. Firefly in Photoshop understands your design file. You do not have to manually paste or upload anything.
But that advantage has a hard ceiling. Built-in AI works only within the walls of that one application. Copilot in Excel cannot analyze the customer interview notes sitting in a PDF. Gemini in Docs cannot cross-reference data from your CRM. The AI is trapped in the silo of the tool it lives in.
Standalone tools like ChatGPT and Claude work exactly the other way around. They do not know your applications, but they are flexible. You can send them any document, ask any question, hand over any problem. Think of a freelance consultant: no institutional knowledge of your company, but a fresh perspective on anything you bring in.
AI researcher Nathan Lambert from Interconnects.ai put it clearly in June 2026:
“Don’t be loyal to one provider. Each model must have a fairly high probability of success, and switching to a peer model regularly solves the task.”
Nathan Lambert, Interconnects.ai
His point: no single model is best at everything. The smart move is using multiple models for what each does best.
| Capability | Built-in AI (Copilot, Gemini) | Standalone plan (ChatGPT, Claude) |
|---|---|---|
| Knows your documents | Yes, within that one app | No, you upload them |
| Works across multiple apps | Limited | Yes, no boundaries |
| Deep reasoning and analysis | Basic | Advanced (thinking modes) |
| Creative writing | Functional | Stronger on tone and nuance |
| Code and technical work | Only within the tool | Broadly applicable |
| Extra cost | Included (price increase) | $20-25 per user per month |
When is built-in AI enough?
Three scenarios where teams do fine with what their software already includes.
Your work happens mostly inside one suite. A team that lives in M365 all day gets real value from Copilot Chat right there: summarizing documents, drafting email replies, capturing meeting notes. Those are exactly the tasks built-in AI is optimized for. The same applies to teams in Google Workspace, where Gemini summarizes your emails, suggests responses, and helps draft documents.
Your AI usage is reactive. You occasionally ask a question, get a text rewritten, or request a summary. As long as you use AI as a lightweight assistant for small tasks, what comes bundled with your software is probably enough. A standalone subscription only pays off when someone is using AI actively, daily, for tasks that go beyond what the bundled suite can handle.
Your team is small and your budget is tight. For a five-person company, M365 Business Standard at $14 per user per month is significantly cheaper than five separate ChatGPT Plus subscriptions at $20 each. That is $30 per month saved, or $360 per year. Not enormous, but real.
When do you actually need a standalone plan?
Built-in AI falls short in three specific situations.
You work with your own data that does not live in your standard software. If your team analyzes customer interviews, compares contracts, or does market research, you need a tool that can handle any document regardless of where it lives. Copilot in Word only knows the document you have open. Claude or ChatGPT can analyze ten documents simultaneously and find cross-references between them. That is the difference between an assistant who only works in the kitchen and one who can move through the whole house.
You use AI as a thinking partner for strategy. Built-in AI is optimized for tasks within the application: a formula in Excel, a paragraph in Word. A standalone model like Claude or ChatGPT can walk through a complete business plan, challenge your assumptions, and stress-test scenarios. That is a fundamentally different kind of use.
You are building workflows or automations. Companies using AI to automate processes, from customer service to content operations, need API flexibility. Built-in AI rarely provides that. The OpenAI and Anthropic APIs do, charged per token. According to McKinsey's 2025 State of AI report, companies that integrate AI into workflows rather than using it as a standalone tool see 3.5x higher productivity gains. That gap matters more than most SMBs realize.
What does it cost, side by side?
Here are the current monthly prices per user in USD for business customers.
| Tool | What you get | Price per user/month |
|---|---|---|
| M365 Business Standard (from July 1) | Copilot Chat + Word/Excel/PowerPoint agents | $14 |
| ChatGPT Plus | GPT-5.4, reasoning, code, plugins | $20 |
| Claude Pro | Claude Opus 4.8, long documents, code | $20 |
| Gemini Advanced | Gemini 3.5, Deep Research, 1M context | $21.99 |
| Perplexity Pro | AI search with citations | $20 |
A ten-person team on M365 Business Standard pays $140 per month for built-in AI after July 1. Add ten standalone ChatGPT Plus subscriptions and you are at $200 more. Total: $340 per month.
Here is the calculation you can run today: count how many hours per week each team member actually uses a standalone AI model for tasks the bundled suite cannot handle. Under two hours per person per week? Built-in is probably enough. Over five hours? A standalone plan pays for itself. Most teams sit somewhere in between. That middle ground is where the decision actually lives.
How do you decide for your team?
Avoid treating this as a technical decision. It is an organizational one. Three steps cut through it.
Step 1: map what your team actually uses AI for today. Not what they could use it for. What they actually do. Most teams use AI for a handful of tasks: rewriting text, summarizing, brainstorming, and searching. For those four, built-in AI covers most of it.
Step 2: find the exceptions. Is there someone on your team who regularly analyzes large document sets, writes code, or uses AI as a thinking partner for complex questions? That person probably needs a standalone plan. The rest may not.
Step 3: test before deciding. Give two or three team members a standalone subscription for one month alongside their existing software. After four weeks, ask one question: what did you use the standalone model for that you could not do in Copilot or Gemini? If the answer is "basically nothing," you save $20 per person per month going forward.
Here is the thing: most teams that take AI seriously end up with a combination. Built-in AI for daily tasks, a standalone plan for the two or three people who push it hardest. That is not a hedge. It is the rational split, because no single model is best at everything and the cost of running both is lower than the cost of asking the wrong tool for the wrong job.
What about your company's data?
The choice between built-in and standalone AI affects your data. Two things worth knowing.
Built-in AI in enterprise software (M365, Google Workspace) does not use your data to train models. Microsoft and Google have committed to that contractually for business accounts. Free tiers of ChatGPT and Claude are different. Research from TheAIDaily found that 44% of SMBs shared company data with a free AI tool last month, while only 28% knew those tools may use that data for training.
The EU AI Act requires transparency about AI use starting August 2, 2026. If your team sends AI-generated output to clients, you need to disclose it. That applies equally to built-in and standalone tools. A paid business subscription (ChatGPT Team, Claude for Business) gives you the same data guarantees as enterprise software: no training on your data, plus the processing agreement required for GDPR compliance.
The data protection difference is not between built-in and standalone. It is between free and paid.
What to do this week
Software vendors are pushing AI deeper into everything you already pay for. For most teams, that is good news: the tools you are already using are getting smarter without requiring an extra subscription.
Check two things this week. First, what are you paying for standalone AI subscriptions, and how much of that capacity is your team actually using? Second, what can the AI in your existing software already do? Many teams discover that Copilot in Excel or Gemini in Docs handles what they have been doing in ChatGPT. They just never tried it.
For a broader look at AI adoption by business size and sector, TheAIDaily's statistics section tracks the latest figures across Europe. The smartest move here is not all-or-nothing. It is knowing exactly where your current software stops and where you need more. Start there, and the rest follows naturally.
