Strawberry AI agent platform 1.0 launch for non-technical teams

Direct answer: Stockholm startup Strawberry has launched version 1.0 of its AI agent platform, aimed squarely at non-technical teams such as sales, recruiting, customer support and marketing. The product is free to try, and it lets teams create their own virtual colleagues, called AI companions, that work inside the browser where a team’s context, workflows and logins already live. The company, which has raised $6 million from EQT and General Catalyst, says it plans to follow the launch with a Series A funding round. For a UK small business, the interesting claim is not that Strawberry can answer a prompt, but that a companion can remember how a team works and keep working alongside people over time — a promise that is genuinely useful if it holds up, and genuinely risky if it is taken on trust.

What happened

Strawberry announced the 1.0 launch of its AI agent platform on 26 October 2026, according to Retail Technology Innovation Hub’s report of the launch. The Stockholm-based company describes the release as a platform built for non-technical teams, and it is positioning the product against a familiar frustration: most people have now tried a chatbot, but few teams have managed to turn that experience into a reliable, shared way of working.

The distinction Strawberry draws is deliberate. Chatbots such as Claude and ChatGPT, in Strawberry’s framing, take a prompt and complete a task. A companion, by contrast, is meant to work like a real colleague, alongside people rather than instead of them. That is a strong claim, and it is worth unpacking slowly, because the difference between a tool that answers and a colleague that works is the whole bet this company is making.

The platform operates as a browser. That choice matters more than it first sounds. The browser is where most office work already happens: the CRM tab, the job board, the inbox, the supplier portal, the spreadsheet, the support helpdesk. By building the agent into the browser, Strawberry is saying that the agent should meet the team where the work already is, rather than asking the team to learn a new application, copy data into it, and then copy results back out again. A companion, Strawberry says, remembers how a team works and gets better the more it works with them, because it is connected to that context.

The launch comes with named backers. Strawberry has $6 million in funding from EQT and General Catalyst, two firms whose involvement will be read by other investors as a signal, even though the amount itself is modest by the standards of the current AI funding market. The company says it plans to follow this launch with a Series A round. No amount, lead investor, or closing date for that Series A has been announced, and it would be wrong to treat a planned round as a completed one. What can be said is that a 1.0 launch is often timed to give a company something concrete to show prospective Series A investors: real users, real workflows, and evidence that non-technical teams will actually adopt the product without a developer in the room.

The chief executive’s framing is about adoption, not capability. Charles Maddock, Strawberry’s CEO and co-founder, said: “The technology is already here: agents can now work for weeks in a row. Yet, we see so many companies struggle with adoption. We built a product so delightful that teams adopt it naturally, with companions you bring into your team like colleagues.” That sentence contains the company’s whole diagnosis. The bottleneck, in this view, is not whether an AI model is clever enough. It is whether ordinary teams will use it, trust it, and keep using it after the novelty wears off.

Yuri Sagalov, a partner at General Catalyst, added a second point that small-business owners should notice: “Today the person who is great with Claude keeps all of it on their own laptop. In Strawberry you train a companion once, and everyone on the team gets it, with everything it has learned. The browser is fast becoming the front line for AI, and the Strawberry team is at the centre of that shift.” In other words, the value is supposed to be shared, not individual. One person trains the companion, the whole team benefits, and the knowledge does not walk out of the door when one employee is on holiday or leaves.

That is an attractive pitch for a small firm, where one or two people often hold most of the operational knowledge in their heads. It is also exactly the kind of pitch that needs to be tested against the realities of permissions, data protection, and mistakes, which we return to below.

How the platform actually works

Strawberry’s model can be broken down into four ideas, each of which deserves a plain-English explanation.

First, the companion. A companion is a named AI helper that a team creates for a recurring kind of work. Think of it less like a search box and more like a new team member with a job description: this companion handles lead research, that one handles first-pass candidate screening, another handles routine support replies. The companion is not a general-purpose chatbot waiting for a question. It is set up for a workflow, and it is expected to be reused.

Second, the browser as the workplace. Because Strawberry operates as a browser, the companion works inside the tools a team already uses. It can move between tabs, use the logins the team already has, and draw on the context of how a team works. For a sales team, that might mean checking a prospect against the CRM, then drafting outreach. For a recruiter, it might mean gathering public information about a candidate and organising it in a consistent format. For a support team, it might mean pulling together the relevant order history before a human replies. The practical benefit is fewer copy-and-paste steps. The practical risk is the same thing viewed from the other side: an agent with access to live systems can also act in live systems, and that needs controls.

Third, shared memory. Strawberry’s claim is that a companion remembers how a team works and improves with use. In practice, this means the companion builds up knowledge about a team’s preferences, formats, and routines, so the team does not have to re-explain everything on every task. The General Catalyst comment makes the point explicitly: train once, share with everyone. For a small business, this is the difference between an AI tool that helps one keen employee and an AI capability that belongs to the business. It is also why the questions about who owns that trained knowledge, what happens to it if the subscription ends, and how it can be corrected when it learns a bad habit are so important. A shared memory that learns the wrong process will repeat the wrong process efficiently, for everyone.

Fourth, working alongside people. Strawberry stresses that companions work alongside people, not instead of them. That is a design philosophy as much as a feature list. The companion is meant to take on the repetitive, multi-step parts of a job — the gathering, the checking, the formatting, the first draft — while a human keeps the judgement calls: which lead is worth pursuing, which candidate is genuinely a fit, which customer complaint needs a personal touch. Whether any given deployment lives up to that philosophy will depend on how the team configures it. A tool that can draft, submit, and send will do exactly that if it is allowed to. The responsibility for setting those boundaries stays with the business.

None of this, on the evidence publicly available at launch, comes with published benchmarks, independent accuracy tests, or a detailed security audit. That is normal for a 1.0 launch from a small startup, but it is a reason to run a controlled trial rather than to roll the product out across a whole company on day one.

How Strawberry compares with the alternatives

It helps to place Strawberry against the three kinds of tool a UK small business is most likely to already know.

Against a general chatbot, Strawberry’s pitch is persistence and sharing. A chatbot conversation starts fresh, or nearly fresh, each time. A companion is supposed to carry context forward and be available to the whole team. The chatbot is brilliant for a one-off question. The companion is aimed at a repeated job. The trade-off is setup: a chatbot needs no setup, while a companion needs to be trained, configured, and checked before it can be trusted with real work.

Against a traditional automation tool, such as a workflow builder that connects apps through fixed rules, Strawberry’s pitch is flexibility. A rule-based automation does exactly what it is told, every time, and breaks when a website changes or a process varies. A browser agent can, in principle, adapt to a changed page, read an unfamiliar form, and handle variations. The trade-off is predictability. A fixed rule is easy to audit: it either ran or it did not. An agent’s behaviour is harder to audit, because it makes small decisions along the way. For regulated or high-stakes steps, that difference matters a great deal.

Against hiring or outsourcing, Strawberry’s pitch is cost and speed. A small business that cannot afford a full-time hire for lead research or inbox triage might use a companion to cover the repetitive part of that work. The honest comparison, though, is not companion versus employee. It is companion plus human oversight versus the current way the work gets done. If the companion saves three hours a week but creates one hour of checking and one embarrassing mistake a month, the arithmetic is less flattering than the headline suggests. The only way to know is to measure it on real tasks, with real error rates, over several weeks.

There is also a wider market context. Browser-based agents are one of the most crowded corners of AI in 2026, with large model makers, browser makers, and startups all pushing in the same direction. Strawberry’s bet is that being built for non-technical teams, with a delightful product that people adopt without being forced to, is a better wedge than being the most powerful agent for developers. That is a plausible strategy for a small company. It is also a strategy that larger rivals can copy quickly if the idea proves popular, which is why the planned Series A matters: Strawberry will need capital to keep building, supporting, and differentiating the product while bigger players move into the same space.

What it means for UK small businesses

For a UK small business, the Strawberry launch is worth attention for practical reasons, not because of the funding headlines. Here is how the story translates into decisions a small firm can actually make.

Start with the jobs that fit. The teams Strawberry names — sales, recruiting, customer support, marketing — map neatly onto the repetitive, browser-based work that eats time in a small business. Lead research is a good example. A small firm might spend hours each week finding companies, checking they fit, and noting down contact details. A companion that does the gathering and leaves the judgement to a human could be genuinely useful. The same logic applies to first-pass candidate organisation, support ticket triage, and marketing research. The jobs that do not fit are the ones where a mistake is expensive or hard to undo: sending a final offer, refunding a customer, submitting a regulatory form, or contacting a client with a binding commitment. Those steps should stay behind a human approval, whatever the tool promises.

Think about the free trial as a test, not a commitment. Strawberry says the platform is free to try. That is the right way to approach it: pick one narrow workflow, one companion, and one owner inside the business. Define what success looks like before the trial starts. For example, a lead-research companion might be judged on how many of its shortlisted leads are actually worth contacting, how long checking takes, and how many errors slip through. Run it for two to four weeks, on real work, with a human checking every output at first. Only widen access when the error rate is known and acceptable.

Treat shared knowledge as a business asset, and protect it accordingly. If a companion learns how your team works, that knowledge has value. Ask, before committing, how it can be exported, corrected, and deleted. Ask what happens to the trained companion if you stop paying, if an employee leaves, or if you need to remove a client’s data. These are not abstract questions. A small business that builds its processes into a single vendor’s companion, with no way to take that knowledge elsewhere, has created a dependency that will be painful to unwind later. The same applies to logins: a browser agent that uses team logins needs to be governed like any other holder of credentials, with least-privilege access, separate accounts where possible, and a clear way to revoke access.

Get the data protection basics right. Under UK GDPR, a business that lets an AI agent process personal data — candidate details, customer messages, lead contact information — remains responsible for that processing. That means knowing what data the companion touches, where it is stored, and whether it is used to train models beyond your business. It means having a lawful basis, telling people where appropriate, and being able to respond to a deletion request. A small business does not need a legal department to do this, but it does need to read the vendor’s data terms, not just its marketing page, and to avoid feeding sensitive personal data into a trial before those terms are understood. Employee data deserves the same care: a companion that watches how a team works can easily drift into monitoring, and staff should know what is being recorded and why.

Budget for the human time, not just the subscription. The most common mistake with AI tools in small businesses is counting the subscription cost and forgetting the oversight cost. A companion that needs an hour of checking for every three hours it saves is still a win, but it is a smaller win than the sales pitch suggests, and it needs a named owner. The businesses that get value from agents are usually the ones that treat them like junior staff: trained, supervised, given clear boundaries, and reviewed regularly. The businesses that get burned are the ones that treat them like magic.

Finally, watch the funding story without being swayed by it. A planned Series A is a sign that a company intends to grow, not a guarantee that it will survive. A small business adopting a startup’s product should always have a fallback: work that can continue without the tool, data that can be exported, and no single process that collapses if the vendor changes direction. That is not pessimism. It is the same common sense that applies to any supplier a small business depends on.

Timeline and key facts

The timeline matters because it shows a company moving from product to scale-up. A 1.0 launch is a statement that the product is ready for wider use. A planned Series A is a statement that the company wants capital to grow. Neither, on its own, proves that the product works well for any particular business. Both together explain why this story is worth watching, and why a careful trial is a better response than either instant adoption or instant dismissal.

Honest limits and what to watch

There are several honest limits to this story that a reader should keep in mind.

First, the evidence is thin. The public launch coverage is short, and it relies largely on the company’s own description and investor comment. There are no published customer names, case studies with measured results, pricing tiers, or independent tests in the launch report. That does not mean the product is weak. It means a reader cannot verify its quality from the announcement alone. A trial on your own tasks is the verification.

Second, the Series A has not happened yet. Reports of a planned round are not the same as a closed round. Until a lead investor and an amount are announced, the company’s runway, hiring plans, and ability to support a growing user base remain open questions. A small business should not make a long-term bet on a vendor’s future funding. It should make a short-term bet on a trial, with an exit plan.

Third, browser agents raise security questions that chatbots do not. An agent that can click, fill forms, and submit information on a team’s behalf can be tricked by malicious content on a web page, can misread a form, and can act on outdated or incorrect context. Vendors in this space talk about guardrails, approval steps, and audit logs — an area where agent login security matters as much as the agent itself, and those features matter enormously. A buyer should ask for them explicitly: what approvals are required before an external action, what log shows what the agent did, and how a mistaken action can be undone. If those answers are vague, the product is not ready for live customer or financial workflows, however good its demo looks.

Fourth, the adoption claim cuts both ways. Strawberry says teams adopt the product naturally because it is delightful. That is a good sign if it means people genuinely find it useful. It is a warning sign if it means people start using it informally, with personal logins and customer data, before the business has set any rules. Shadow AI use — employees using AI tools the business has not approved — is already a problem for many small firms. A delightful browser agent could make that problem worse unless the business sets a simple policy first: which tools are approved, which data may be used, and who approves a new companion.

Fifth, the competitive risk is real. If browser companions become a standard feature of major browsers and AI assistants, a small startup will need a clear reason to keep existing. Strawberry’s reason is its focus on non-technical teams and shared team knowledge. Whether that is enough will become clear over the next year, as larger rivals ship similar features and as customers decide whether they want a specialist tool or a feature inside something they already pay for.

What to watch, then, is concrete. Watch for the Series A announcement, and for who leads it. Watch for named customers and measured results, not just testimonials. Watch for published pricing, data terms, and security documentation. Watch for whether the product adds approval workflows and audit logs that make it safe for external actions. And watch for how the company handles the first public mistake, because every agent product eventually makes one, and the response tells you more about the vendor than the launch does.

For readers who want to follow the agent story more broadly, our earlier coverage of AI agents living in text messages shows how agents are moving into everyday channels, and our look at Anthropic’s agent app explores a larger rival’s approach to the same problem. Both are useful context for judging whether Strawberry’s team-focused pitch is genuinely different or simply a smaller version of a much larger trend.

FAQs

What is Strawberry’s AI agent platform?

Strawberry is a Stockholm startup’s AI agent platform, version 1.0, that lets non-technical teams create AI companions. A companion is a virtual colleague that works inside the browser, where a team’s context, workflows and logins already live, and is meant to work alongside people on recurring tasks.

Which teams is Strawberry built for?

The launch positions the platform for sales, recruiting, customer support and marketing teams. These are teams whose work is largely browser-based, repetitive, and multi-step, which makes them a natural fit for a companion that gathers, organises, and drafts before a human decides.

How is a companion different from a chatbot?

According to Strawberry, a chatbot takes a prompt and completes a task, while a companion works like a colleague, alongside people and over time. A companion is set up for a recurring workflow, remembers how a team works, and is shared across the team, rather than living in one person’s chat history.

How much funding has Strawberry raised?

Strawberry has $6 million in funding from EQT and General Catalyst. The company says it plans to follow the 1.0 launch with a Series A funding round, but no amount, lead investor, or closing date has been announced.

Is Strawberry free?

The platform is free to try at launch. Full pricing tiers have not been detailed in the launch coverage, so a business should confirm what is included in a trial, what happens when a trial ends, and what a paid plan costs before building a core process around it.

Should a UK small business adopt Strawberry now?

A controlled trial is a sensible next step, but a company-wide rollout is premature on announcement evidence alone. Pick one narrow workflow, measure accuracy and checking time, confirm the data and security terms, keep human approval on external actions, and keep an exit plan so the business is not stranded if the vendor changes direction.

Sources: Retail Technology Innovation Hub’s report of the Strawberry 1.0 launch (26 October 2026), including quotes from CEO and co-founder Charles Maddock and General Catalyst partner Yuri Sagalov; company background on its Stockholm base and $6 million funding from EQT and General Catalyst as reported in the same launch coverage.

CTA: Get a Free AI Consultation