Strawberry: what it does, pricing and alternatives

What it does

Strawberry is a desktop web browser for macOS and Windows with AI assistants, called companions, built in. According to its documentation, companions can read the page in front of them, operate websites, use connected apps, create files and carry out multi-step tasks in the background. Each one can have its own instructions, memory, files, skills and routines, which are tasks that run on a schedule or when something happens in a connected app.

The Stockholm company moved the product out of beta with version 1.0 on October 26, 2026. The Next Web reports that the release lets colleagues share a trained companion, lets companions start on a schedule or when a new email or lead arrives, and can split large jobs among parallel sub-agents. It can also record Zoom, Teams and Google Meet calls locally, without a bot joining, and turn them into notes and follow-up tasks.

The official site lists presets such as Sales Sally, Talent Theo, Executive Ellis, Marketing Martin and Consulting Cora, plus the option to build a custom one. Listed connections include Gmail, Slack, Google Calendar, Google Sheets, Notion, Linear, HubSpot and LinkedIn.

Who it is for

The Next Web describes the audience as non-coding workers in sales, marketing, recruitment and customer service who handle repetitive browser tasks. The official site also addresses founders, executives and consultants. The common thread is work spread across many logged-in websites, where copying information from one tab to another takes up the day.

It makes less sense for people whose work sits inside one system with good automation of its own, or for companies that require a managed browser. Strawberry's own Chrome comparison says Chrome may fit better if an organization requires it or if someone relies on an extension Strawberry doesn't support.

How people use it

A realistic job is following up on inbound sales leads. This example is assembled from the documentation and has not been tested first-hand. A sales rep installs Strawberry, imports bookmarks, passwords and Chrome extensions, and hires Sales Sally. They connect Gmail and HubSpot, then add instructions describing the ideal customer, the research wanted on each company and the tone of a first email.

Following the documentation's security advice, the rep starts with read and draft permissions only. They ask the companion to research five recent leads and draft a follow-up for each, then check the sources and the work trace, which shows what the companion did during the run. Clear column definitions matter here, because one hands-on reviewer found that vague requests returned partial results that needed cleanup.

Once the output looks right, the rep turns the task into a routine triggered by a new lead, deciding in advance what may happen automatically and what must wait for approval. When a draft is ready, an approval card appears. For email sends, the docs say the card lets the rep edit the recipients, subject and body before allowing it. When it works well, the rep shares the companion with the team, keeping in mind that everyone it is shared with can currently edit it.

Pricing

I checked the pricing page on October 30, 2026. It lists Free at "$0 /month" with 500 monthly credits, and says no credit card is needed. Intern costs $20/month for 6,000 monthly credits, Part-Time costs $85/month for 25,000, and Full-Time costs $250/month for 75,000. Full-Time adds Expert mode and priority access during peak hours.

Browsing is free, but any companion response uses credits, including ordinary chat replies. Plan credits refresh monthly and don't roll over. Top-ups cost $0.01 per credit and do roll over. The Team option includes 3 free seats, counting the owner's, and each paid seat has its own monthly allowance. Enterprise pricing isn't listed, though the page mentions pooled credits, rollover, invoicing and custom terms.

Budgeting is the hard part. The credit documentation names the factors that raise cost, such as agent runs, web research, model choice and request length, but gives no typical cost per task. Short-term rate-limit windows can also pause work while monthly credits remain. Teams have no automatic pay-as-you-go overage, so work stops when credits run out unless automatic refills are set up. One independent review cites 2,000 free credits a month, while the current pricing page says 500, so the pricing page is the figure to trust.

Strengths

The main appeal is that the work happens where the logins already are. The official site says companions can act on any website, including those behind logins, which reduces reliance on formal integrations. Credentials are handled conservatively, according to the company: passwords and browsing history stay on the device, and it says passwords are never exposed to it or to AI model providers.

Approval controls are more detailed than a single on-off switch. Connected apps can offer per-action settings of ask every time, always allow or disable, and the docs advise against always allowing anything that sends messages, changes records, deletes data, invites users or spends money. That advice is unusually direct for vendor documentation.

Early feedback is warm, if limited. Product Hunt shows 4.8 out of 5 from 13 reviews, with one user saying it works well for lead generation and scraping. Make Tech Easier called it the best AI browser its writer had used, though that was a November 2025 piece. Luca Berton's hands-on review found it credible on tasks run against public sources, although it reports no success rates.

Limits and things to watch

Speed is the most repeated complaint. Product Hunt reviews tag slow performance and a lack of smoothness compared with traditional browsers, and one reviewer from about six months ago said performance can suffer on heavier workflows. Make Tech Easier noted lag and higher resource use, and said the product was no reason to abandon a regular browser. Berton saw heavier tasks take from tens of seconds to a couple of minutes, and found that dynamic fields, multi-step wizards and anti-bot measures can stop the agent. Most of this feedback predates version 1.0, and I found no independent test of the new release.

Approval is not absolute. Saved permissions still apply in Ask Always mode, some actions without an approval control can run without a prompt, and an always-allow setting removes the check entirely. A team can gradually loosen its own safeguards without anyone deciding to.

Sharing also carries risk, because recipients get edit access to settings, memories and skills until sharing stops. The docs warn against putting sensitive files or private instructions in widely shared companions. The Next Web raises the larger open question of whether an assistant trained by one employee can work reliably for a whole team without adding more oversight than it saves.

Some things are claimed or left unsaid. Strawberry says it meets SOC 2 requirements and complies with GDPR, but those are company statements, and Make Tech Easier treated its data-training claim with skepticism. The transcription docs don't cover participant consent, so teams need their own recording policy. The routines docs don't say whether the computer must stay on. There is also no mobile app, according to Berton's review. The company is small too, with 12 people and $6 million raised, according to The Next Web.

When to choose something else

Lindy suits teams that would rather not change browsers. Its pricing page describes an assistant in Slack threads, email and iMessage, with 1,500+ integrations, and team plans from $29.99 per user per month for 3,000 credits, pooled across the team. Its Enterprise plan lists a signed HIPAA agreement, SSO and audit logs, which Strawberry's pricing page doesn't advertise in the same terms.

Bardeen fits teams running a fixed, repeatable pipeline rather than wanting an all-purpose assistant. Its example workflow scrapes results, validates emails, qualifies leads and exports to Google Sheets. Plans start at $10/month, and per-row credit costs are published, so spending is easier to predict.

Clay is the better choice when the job is prospect data at scale. It lists more than 150 data providers, buying signals and CRM sync, with a free tier of 500 actions and 100 data credits a month. Its paid pricing is complex, and its page shows inconsistent figures, including a Launch headline of $167/mo alongside lower selector prices, so buyers should confirm the numbers first.

Verdict

Strawberry is a sensible trial for a small team with repetitive, multi-site browser work and someone willing to supervise the results. The free plan keeps that trial cheap, and read-only permissions keep it safe. Scaling it across a team is a bigger decision. The evidence for version 1.0 is still mostly from the company, usage costs are hard to forecast, and approvals protect only as far as saved permissions allow. Teams with strict compliance needs or heavy data requirements should look at Lindy or Clay first, and revisit Strawberry when independent reports on 1.0 appear.