Stockholm startup Strawberry is back in the news on October 26, 2026, as coverage pairs its agent-driven browser with talk of a Series A round. The hard facts are older and narrower. Silicon Canals reported that the “self-driving” browser company raised €5.1M, the same round other outlets put at $6 million in seed funding, co-led by General Catalyst and EQT Ventures in October 2025. None of the sources we could verify names a Series A investor, amount or valuation.
That gap matters for anyone tracking the AI agent market. A browser that runs its own agents sits in the same race as workplace assistants such as the one covered in our look at Claude Duet and its 6,000 connected apps. This analysis separates what Strawberry has confirmed from what the headlines imply, then looks at the competition, the risks and what a real Series A would have to prove.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
Strawberry Browser at a Glance
Strawberry builds a browser where AI agents, which the company calls AI Companions, do the clicking, typing and copying that a person normally does across tabs. The company is based in Stockholm. Its founders are Charles Maddock, Arian Hanifi and Sebastian Thunman, and its public positioning is “a browser built for agents, not tabs.”
The pitch is simple. Most office work happens inside a browser anyway. If the browser itself carries the agent, the agent can act inside web apps where the user is already signed in, instead of asking for API keys or building a connector for each tool. AI Deck describes it as a self-driving browser with resident agents that handle research and CRM data entry.
| Item | Detail | Status |
|---|---|---|
| Company | Stockholm-based agentic browser startup | Confirmed |
| Founders | Charles Maddock, Arian Hanifi, Sebastian Thunman | Confirmed |
| Seed round | $6 million, October 2025 | Confirmed |
| Round leads | General Catalyst and EQT Ventures | Confirmed |
| Angel backers | Founders of Lovable, Hugging Face and Supabase | Confirmed |
| Beta | October 2025, 10,000 pre-registered users | Confirmed |
| Series A | Investors, amount, valuation, timing | Unconfirmed |
| “AI agent platform launch” | Exact official product-launch wording | Unconfirmed |
The Founders Behind the Stockholm Startup
Maddock, Hanifi and Thunman lead a company that pitched itself to investors on a narrow thesis: the browser is where knowledge workers spend their day, so it is the cheapest place to put an agent. Maddock has the public-facing role and has given interviews to Swedish press about strategy and fundraising. Thunman has spoken about the company’s reach beyond Sweden.
A Transatlantic Cap Table
The seed round pairs a Swedish backer, EQT Ventures, with a US one, General Catalyst. Thunman told Ny Teknik that the company has “support from both sides of the Atlantic, which is perfect as we are targeting international markets.” He added that investors had given practical advice on building a company, and that both firms carry international name recognition. That matters when a young company wants to hire in the US and sell to US teams.
Operators as Angels
The angel list reads like a map of the current AI tooling scene: founders of Lovable, Hugging Face and Supabase. Each runs a product that developers and builders already use daily. Their participation signals product credibility more than it signals cash, since the round was only $6 million in total.
How AI Companions Work Inside the Browser
Strawberry’s AI Companions are specialized agents that run inside the browser and carry out multi-step tasks. The named examples are research, data extraction, lead generation, email management and workflow automation. Reported integrations include LinkedIn, HubSpot, Gmail, Bullhorn and internal dashboards.
Look at that list and a pattern shows up. LinkedIn feeds prospect data, HubSpot holds the CRM record, Gmail carries outreach, and Bullhorn is applicant-tracking software used by staffing firms. A single recruiter or sales rep can bounce between all four in an hour. An agent that sits in the browser and sees the same logged-in sessions can chain those steps without a bespoke integration for each vendor.
Reports also say Strawberry coordinates several agents through its own orchestration layer, so one agent can gather data while another drafts messages. We could not verify the architecture beyond that description, and the company has not published benchmarks for task success rates. Treat any accuracy claim without a published test as marketing until someone measures it.
The $6 Million Seed Round and Its Investors
Strawberry raised $6 million in October 2025. General Catalyst and EQT Ventures co-led, with the Lovable, Hugging Face and Supabase founders joining. Yuri Sagalov, a partner at General Catalyst, framed the bet in one line: “The browser is fast becoming the front line for AI, and the Strawberry team is at the centre of that shift,” according to Startups Magazine.
Maddock’s own statement in the same announcement was about access. He said the browser is a tool most people use daily and know intimately, and that shipping powerful AI features out of the box makes automation accessible to everyone. In his words, the funding helps the company “help modern workers love what they do and skip their busywork.”
A $6 million seed does not stretch far in a field where rivals ship their own models and run large inference bills. It buys a small engineering team, a year or so of runway and the right to talk to larger investors. That is the context in which a Series A rumor would surface around month twelve.
Series A Status: What Is Confirmed and What Is Not
Here is the cleanest reading of the evidence. The seed is confirmed. A beta is confirmed. A Series A is not confirmed by any verified source we reviewed. We found no named lead, no target size, no valuation, no term sheet and no closing date. Several outlets are linking Strawberry’s current agent push to a coming round, and that framing may prove right, but it is reporting about intent, not a closed deal.
Maddock’s earlier comments point the same way. Speaking to Breakit, he said the company would raise money “to accelerate faster, when we feel that the product is in place and we are ready to scale up the team.” That is a conditional, not a calendar date. It tells you the founders see fundraising as a response to product readiness, not a fixed milestone.
The same logic applies to the headline wording about an “AI agent platform” launch. The sources confirm a browser with built-in agents and a beta. They do not confirm a separate, officially named platform release on a specific date. We use the phrase only as shorthand for the browser and its agent layer.
What the 10,000-User Beta Tells Us
The October 2025 beta admitted 10,000 users who had signed up in advance. That number is a waitlist conversion, not a paying base. It shows people were curious enough to register for an unreleased browser, which is a real if modest demand signal. It says nothing about retention, task completion or revenue.
Computerwoche reported the Swedish AI browser moving into an open beta, and other reports say downloads later opened without pre-registration, with a credit-based model for usage. We could not verify the pricing details, so check the company’s own pricing page before relying on them. Credit models matter here because agent runs cost real inference money, which pushes every agent startup toward metered usage.
Pricing pressure is already visible across the developer side of the market. Our breakdown of Claude Code, Codex and Gemini CLI pricing shows plans stretching from $20 to $500, and agent products for office workers will likely land on similar tiers.
Strawberry vs the Agentic Browser Field
Strawberry is not alone. Perplexity, OpenAI, Opera and The Browser Company all ship browsers with agent features. The comparison below uses reported details from agentic-browser roundups, including agentmarketcap.ai, which varies in editorial reliability. Verify each line against the vendor’s release notes before you act on it.
| Browser | Maker | Reported angle | Reported update |
|---|---|---|---|
| Strawberry | Strawberry (Stockholm) | AI Companions for CRM, email, research | Beta since October 2025 |
| Comet | Perplexity | Search-led assistant browser | Free Mac and Windows access in October 2025, free iOS in March 2026 |
| ChatGPT Atlas | OpenAI | ChatGPT built into the browser | macOS launch October 21, 2025, merged with ChatGPT and Codex in March 2026 |
| Neon | Opera | Agentic browsing with subscription features | MCP Connector added March 31, 2026 |
| Dia | The Browser Company | AI-first browser with app integrations | Slack, Notion and Gmail added in March 2026, Dia Pro reported at $20 a month |
The table shows where Strawberry is different. Its named targets are sales, recruiting and operations work, with Bullhorn and HubSpot on the integration list. Rivals lean toward general consumer assistance. A vertical focus is a defensible wedge for a startup with $6 million, because it does not need to beat Google or OpenAI at everything. It needs to beat them at filling in a recruiter’s candidate record.
The weakness is distribution. OpenAI can push Atlas through ChatGPT’s user base, and Perplexity gave Comet away free. Strawberry has to win users one download at a time, with a team a fraction of the size.
Market Impact: Why Investors Chase the Browser Layer
Venture money is flowing to whoever owns the surface where agents act. A browser is that surface for web software. If agents handle more clicking, the company that controls the browser controls which sites the agent visits, which accounts it can touch and which models it calls. That gives it leverage over both software vendors and model providers.
It also explains the defensive moves. When agents begin working on behalf of customers, sites and institutions respond. We covered one example when five big US banks moved to block third-party AI agents. A browser agent that logs into a bank, a CRM or a recruiting system looks, to that system, like an automated client. Some vendors will welcome it. Others will treat it as a policy violation.
For Strawberry, the commercial outcome hinges on that tension. Its integration list covers tools whose terms of service and rate limits the company does not control. LinkedIn’s stance on automated access is a known pressure point for lead-generation tools. A Series A investor will ask how durable the product is if a platform changes its rules.
Historical Context: From Browser Wars to Agent Wars
Browsers have been a battleground since the 1990s, and Chrome’s 2008 launch reset the field by making speed and simplicity the main selling points. For years afterward, challengers struggled because Chrome was good enough and came bundled with Google’s services. New entrants mostly competed on privacy or design.
The agent era changes the question. The browser no longer competes on how fast it renders a page. It competes on how much work it can finish without the user. That opens space for newcomers, because the incumbent advantage in rendering engines matters less than the quality of the agent loop, the permissions model and the integrations.
Strawberry’s Swedish pitch against Chrome drew local coverage as soon as the beta opened. Ny Teknik ran a feature where the founders argued their AI browser is better than Chrome. We cannot judge that claim without independent testing, but the framing shows how the company sees its opponent. It is aiming at the default, not at the other startups.
Security and Trust Risks of Logged-In Agents
An agent that works inside your logged-in sessions can read your inbox, edit CRM records and send messages as you. That power creates the main risk of the category. A malicious web page can hide instructions in its text and try to steer the agent, a class of attack security researchers call prompt injection. Every vendor in this table has to defend against it, and none can claim the problem is solved.
Sensible buyers should ask five questions before giving any agent browser access to work accounts:
- Which accounts and sessions can the agent see by default?
- Does it ask for approval before sending email or editing records?
- Where do page contents go, and which model provider processes them?
- Can an administrator restrict sites and actions centrally?
- Is there a log of every action the agent took?
Mistakes also scale. We looked at how far a persistent bot can go in our story on Grok sending 187 messages while chasing a £350 refund. An agent with access to email and a CRM can repeat a bad action hundreds of times before anyone notices. Strawberry has not published a security audit, so buyers should treat any pilot as a limited trial with low-risk accounts.
Enterprise Adoption and the Shadow AI Problem
Browser agents often enter companies through individual employees, not IT. A salesperson downloads a free beta, connects it to HubSpot and starts saving an hour a day. That path worries security teams, and it fits the pattern we described in our report that 45% of workers use shadow AI as their employers push official tools.
For Strawberry, bottom-up adoption is both a growth engine and a ceiling. Individual users give it product feedback and word of mouth. To reach a Series A story built on revenue, it will need team plans, admin controls and a pitch that a procurement department can sign. Staffing firms using Bullhorn and sales teams on HubSpot are the logical first buyers, since the integrations match their daily tools.
Our wider AI and machine learning coverage tracks how these agent products move from individual trials to company contracts.
Reading the Quotes: What Founders and Investors Say
Four statements from the record frame the company’s story. Maddock stressed accessibility and a worker-focused mission. Sagalov of General Catalyst called the browser the front line for AI. Thunman emphasized international ambitions backed by both Swedish and US investors. Maddock, on fundraising, tied any new raise to product readiness and team scaling.
Put together, those comments describe a company that wants to grow into a bigger round, not rush one. That reading cuts against the idea of an imminent Series A, though a year of beta data and an agent upgrade could change the founders’ view of readiness. Founders often say they will raise when ready, then raise when an investor makes an offer too good to ignore.
One more detail deserves caution. A post on Strawberry’s X account dated October 23 reportedly promoted a major upgrade to its agents. We found it only through a search result and could not confirm the upgrade’s contents, so we do not treat it as established.
Five Predictions for the Next 12 Months
These are our judgments, not reported facts, and each could be wrong.
- A Series A, if it comes, will be led by a returning investor. General Catalyst and EQT Ventures know the product and can move fast, which makes an insider-led round the most likely route.
- Strawberry will push harder into recruiting and sales. Its integration list already points there, and a narrow vertical gives investors a clearer revenue story than a general assistant.
- Admin and security features will arrive before any big enterprise push. Buyers will not approve a logged-in agent without central controls and action logs.
- Platforms will tighten rules on automation. Expect more rate limits and terms changes from services that agents touch, which will force browser vendors to negotiate or work around them.
- At least one larger rival will copy the “companion” model. Specialized agents for named jobs are easy to imitate once they prove popular, so Strawberry’s lead depends on execution speed.
What to Watch Next
A few concrete events would turn the rumor into a story. This table lists them, along with the evidence that would count.
| Signal | Evidence that would confirm it |
|---|---|
| Series A closes | Company announcement or investor blog post naming lead and amount |
| Paying customers | Published pricing, team plans or customer names |
| Product maturity | Exit from beta or a dated release with notes |
| Security posture | Third-party audit or published prompt-injection defenses |
| Platform tolerance | LinkedIn, HubSpot or Bullhorn comments on agent access |
| Task reliability | Public benchmarks or independent hands-on tests |
Developers see a similar reliability question in coding agents. Our guide to catching edit drift in Cursor Plan Mode shows how much checking an agent’s output still needs. Browser agents will need equivalent review habits.
The takeaway: Strawberry is a real company with a real seed round and a working beta, and it targets a believable niche. The Series A story is not yet a fact. Until a named investor or the company confirms it, treat it as talk.
FAQ: Strawberry Browser and the Series A Question
What is the Strawberry browser?
It is an AI-powered, agentic browser from a Stockholm startup. Built-in AI Companions run multi-step tasks such as research, data extraction, lead generation and email management inside the browser.
Who founded Strawberry?
Charles Maddock, Arian Hanifi and Sebastian Thunman founded the company. Maddock is the co-founder and CEO, and Thunman is a co-founder.
How much funding has Strawberry raised?
The confirmed figure is $6 million in seed funding from October 2025, co-led by General Catalyst and EQT Ventures. Silicon Canals reported the same round as €5.1M.
Has Strawberry announced a Series A?
No verified source we reviewed confirms one. There is no named lead investor, amount, valuation or closing date. Reports of a coming round remain unconfirmed.
Is the Strawberry browser available now?
A beta launched in October 2025 with 10,000 pre-registered users. Reports say access later widened, but check the company’s own site for current availability and pricing.
How does Strawberry differ from Comet, Atlas, Neon and Dia?
Strawberry focuses on specialized agents for sales, recruiting and operations work, with integrations such as LinkedIn, HubSpot, Gmail and Bullhorn. The larger rivals come from bigger companies and lean toward general assistance.
Is it safe to give a browser agent access to work accounts?
Treat it as a risk to manage. Agents acting in logged-in sessions can be steered by hostile web content, so start with low-risk accounts, require approvals for sending and editing, and review action logs.
