Pricing repriced | Software Talks AI
June 2026 · Issue #3 Monthly SaaS executive newsletter

Pricing repriced

AI tooling will be more expensive in the short-term.

This month

Welcome to another edition of Software Talks AI, Monterro's monthly read on AI for B2B SaaS leaders. On the agenda this month: the token cost transition hitting AI tools across the board, the agent platform race converging on a recognisable harness shape, an EU AI Act update, and Kaisa stealing the show at Autosbuzz by landing Mobile.de as its first AI customer.

But first, let's talk token costs. Tokens are the world as AI models see it. They don't see words the way we do, they see tokens. Tokens are what we pay for, in energy or money, when we use AI. We've been living in a golden era of AI usage. AI now has immense agentic capabilities which we've been able to use at a pretty steep discount. That discount is starting to disappear, as AI tools are transitioning their payment models.

GitHub Copilot, for instance, transitioned from feature-based to token-based AI credits on June 1st, which will increase costs by as much as 5x for some teams. Cursor made the same move in June 2025, Windsurf in March 2026, Replit through effort-based credits, and OpenAI Codex on April 2, 2026. Headline subscription prices look stable at $10, $20, and $39, but those numbers now buy a pool of model spend rather than a fixed number of requests. As GitHub itself acknowledged, a handful of requests can now incur costs that exceed the entire monthly plan price.

Frontier model rates are not following the cheap-and-abundant narrative either. OpenAI doubled GPT-5.5 to $5 input and $30 output per million tokens in April. Anthropic kept Opus 4.7 at $5 and $25 but shipped a new tokenizer that produces up to 35 percent more tokens for the same input text, raising effective costs 20 to 35 percent with no change to the rate card. Google's Gemini 3.1 Pro is priced 60 percent higher than 2.5 Pro, and while the recently announced Gemini 3.5 Flash is more affordable than 3.1 Pro, it is significantly more expensive than past models in the series.

A clear $100 to $200 per-seat ceiling for power users has formed across the industry. Anthropic Claude Max, OpenAI ChatGPT Pro, Cursor Ultra, Windsurf Max, and Claude Code Max all converge at the same two price points.

In the short term, we need to be ready for a price hike in AI services, but there are many small adjustments we can make to improve token efficiency. When using AI services, we can save tokens by spending more time planning before building, and when implementing AI into products, we can use techniques such as prompt caching and batch APIs to reduce token costs by as much as 90%. In both cases, we can also consider using cheaper models. Regardless, we must be model fluid.

In three-to-five years, we believe inference will be drastically cheaper and abundant. Capacity gets built, open-source catches up on the second tier, and models become optimized. Plan for cheap and abundant. Operate as if it is not. Do not lock product pricing to today's inference COGS. And do not invest in defensibility that depends on which model you call rather than what you can do with cheap intelligence that competitors cannot. Tokens get cheaper. Proprietary data, customer workflows, integration depth, and trust do not. That is where capital should compound.

Top stories

What moved in AI this month, and what it means for B2B software leaders.

Story 01

The agent platform race keeps converging

We've talked a lot about Anthropic's Claude Cowork, and the company's competitors are certainly working hard to catch up to it. We've mentioned OpenAI's Workspace Agents recently, and now, Google have announced their alternative. Google's Gemini Spark is, what the company calls, a 24/7 personal agent. It has agentic capabilities with native Gmail, Docs, Sheets, and Slides integration. It's not yet available in the EU, but is another sure sign that Anthropic competitors are working hard to catch up.

In addition, The General Intelligence Company of New York just launched Cofounder 2, which is pitching itself as infrastructure for the one-person billion-dollar company. Cofounder 2 is a step towards AI that can run entire enterprises. The category has settled into a recognizable shape. An agent harness (application) with durable memory, file and app access, declarative goals, multi-agent orchestration, and a scheduling layer.

Read the full story →
Story 02

EU regulations update

On May 7, after nine hours of trilogue negotiations under the Cyprus presidency, EU governments and Parliament agreed to streamline the AI Act as part of the Digital Omnibus package. Legislation for high-risk standalone systems has been pushed forward. Machinery has been excluded from direct AI Act applicability, with the Commission getting delegated acts instead.

Some bits of regulation did get accelerated: the grace period for transparency and marking obligations for AI-generated content was cut from six months to three, so those obligations now bite from December 2, 2026.

Read the full story →
Story 03

Kaisa steals the show at Autosbuzz, lands Mobile.de as first AI customer

Kaisa is a software company in Monterro's portfolio that is working behind the scenes of Europe's automotive marketplaces, helping car dealers turn online enquiries into actual sales. Data from over 100,000 EU dealers flows through their platform, and the problem they solve is a critical one for dealers. Around 75% of car buyers purchase from the first dealer that responds to their enquiry, yet 42% of phone leads get no response at all and 65% of form leads wait more than 24 hours for a reply.

Kaisa built an AI agent to close that gap by handling lead responses automatically, around the clock, in a way that actually converts. The agent is a conversational AI that picks up the phone on behalf of dealers when the human seller is busy, out of network, or outside opening hours. Because the call routes through Kaisa's platform, the agent already knows which car the buyer is calling about and answers questions about the listing using the real data behind it. It then qualifies the lead by collecting a name, a phone number, and a sense of buying intent, and hands anything outside its scope back to the human seller.

Every conversation is transcribed and classified, which gives dealers a level of visibility into their phone channel that simply did not exist before. This is also where Kaisa pulls ahead of generic voice-AI providers, who would start every call cold and need Kaisa to feed them context in real time. Sitting on the platform means the agent starts every call already knowing what is going on.

Just the other week, the Kaisa team was in Barcelona for Autosbuzz, which is an annual gathering of the European automotive marketplace industry. It is a tight, senior crowd of around 350 C-level attendees from every major marketplace and technology player in the sector. One major attendee was Mobile.de, which is one of the largest automotive marketplaces in the world, with more than 42 000 active merchants on its platform.

The day before the event, after weeks of back-and-forth, Kaisa signed a contract with Mobile.de to be the launch partner for the AI agent. Ajay Bhatia, the CEO of Mobile.de and the keynote speaker of the gathering, decided that very night that he wanted to use his keynote slot to demo the Kaisa AI agent live on stage in front of the entire audience. The team spent the evening helping Mobile.de build and rehearse the flow until everyone was comfortable. The live demo worked perfectly, and the energy in the room was incredible. After this really exciting and energetic announcement, Kaisa is now set to go live with its AI feature, becoming the first company to introduce a dealer-focused AI product at scale.

Read the full story →

Key metrics

Numbers worth keeping.

Spend $100‑200/mo

The new normal for company-paid AI spend per power user each month

Cost ‑75%

DeepSeek cut the price of its AI models as token costs rise, pushing more users toward Chinese AI

Value ‑40%

The cheaper Gemini 3.5 Flash beats the pricier 3.1 Pro in some scenarios

Our take

Plan for cheap and abundant inference. Operate as if it is not.

In the short term, we need to be ready for a price hike in AI services, but there are many small adjustments we can make to improve token efficiency. When using AI services, we can save tokens by spending more time planning before building, and when implementing AI into products, we can use techniques such as prompt caching and batch APIs to reduce token costs by as much as 90%.

In both cases, we can also consider using cheaper models. Regardless, we must be model fluid. In three-to-five years, we believe inference will be drastically cheaper and abundant. Capacity gets built, open-source catches up on the second tier, and models become optimized.

Do not lock product pricing to today's inference COGS. And do not invest in defensibility that depends on which model you call rather than what you can do with cheap intelligence that competitors cannot. Tokens get cheaper. Proprietary data, customer workflows, integration depth, and trust do not. That is where capital should compound.

Try this

Put it into practice before the next issue.

1

Create a scheduled task in the AI tool of your choice that goes through your calendar and email every Monday morning and helps you plan your week.

2

Plan before building. Are you using AI to create PowerPoints? Spend two minutes planning the content in text before building it. A few minutes of planning will save you thirty minutes of revising. It's a token save too, as building PowerPoints is quite computationally intense.

3

Try Claude Design at claude.ai/design. It can design websites and make beautiful slides as well, and it has a separate token usage pool from other Claude features.

claude.ai/design →

Footnote: Before using Claude Cowork, check with your organisation that it complies with your internal policies. Avoid inputting sensitive or personal data into AI tools unless your organisation has given explicit approval.

Infuse AI

Case study: Nimblr, 3x growth in shipped code changes

Inspired by Monterro's Infuse AI programme, portfolio company Nimblr has in just four months fully embraced AI tools to accelerate development, internal productivity and customer delivery.

Read the full story →

“We have had a 3x growth in shipped code changes and significantly reduced a long-standing backlog. It allows us to be far more ambitious with our roadmap planning.”

Martin Prieto, CTO at Nimblr

Get Software Talks AI in your inbox

One email a month on AI for B2B SaaS leaders. Written by the Monterro team.