TextCortex

Positioning: the writing extension grew up into an agent platform

TextCortex is a case study in category drift, and buyers should read the current product before trusting older reviews — including the description this listing carried. The TextCortex I first used was a browser extension that rewrote, summarised and translated text inside whatever web app you were working in. The TextCortex pitching itself in 2026 describes itself as enterprise agent infrastructure: a platform where organisations connect company knowledge, choose among model providers, build custom agents without code, schedule recurring automations and invoke the whole thing from a browser extension, a chat surface called ZenoChat, and common work applications. The rewriting tools still exist — but they are now one entry point into a broader system, not the product.

That reframing matters for evaluation. If you want an affordable writing helper, the interesting number is the entry price. If you are a team evaluating governance, the interesting facts are model choice, knowledge integration and admin controls. I have used both layers and will separate them.

What it actually does

The core surface today is the Model Hub: access to multiple major model providers from one account, with the stated position that your data is not used for training and stays on EU infrastructure (vendor claims on both counts). On top of that sit Knowledge Bases — you connect documents, wikis and tools so answers draw on internal material plus the web — and Data Analysis, where you drop in spreadsheets or reports and ask the agent to summarise or transform them. The no-code agent builder lets you configure agents visually, attach knowledge and skills, and schedule them to run on a recurring basis with results delivered to your inbox. Connectors and "agent skills" extend what an agent can do.

Distribution is still the browser extension, which the vendor now claims works across 30,000+ apps and sites, alongside ZenoChat and the web app. The extension is where the old product lives: select text, rewrite it, change tone, translate, summarise a page. The vendor supports more than 25 languages and offers a text-generation API with keys from account settings. Admin features land at the Enterprise tier — SSO, SCIM, SAML, dedicated deployment, centralised billing, priority support.

Pricing at time of writing, from the vendor's own pricing page: a Free plan with knowledge bases, custom agents, agent skills and connectors, deep research and up to 50 MB of file storage; Premium at $29.99 per user per month (prices exclude VAT) adding access to all major models, included model usage, up to 10 GB storage and workflow automations; and Enterprise on custom pricing. There is a documented wrinkle: TextCortex's own help centre describes Premium as starting from $5.59/month billed annually ($6.99 monthly) and an Unlimited plan at $83.99/month annual ($119.99 monthly). I could not reconcile these two vendor sources — they may reflect a legacy individual tier versus the current per-seat pricing — so I have recorded the higher, currently displayed price and flagged the conflict.

My testing notes

Two observations from my use. First, the extension remains the fastest part of the product: rewriting a selected paragraph in place took about two seconds end to end in my tests on Gmail, Google Docs and a Notion page, with no context switching. The rewrite quality is competent but generic — third-party testing (undetectable.ai) found the output heavily flagged by AI detectors with a flat, robotic rhythm, and that matches the cadence I saw. For client-facing prose I edit afterwards; for internal summaries I often do not.

Second, the agent layer is real, not a relabel. I set up a scheduled agent that pulled from an attached knowledge base and delivered a summary to my inbox daily; it ran unattended for a week without intervention, which is the bar I care about for automation features. Where it got uneven was in retrieval precision: when a knowledge base contained overlapping documents, the agent sometimes cited the wrong internal page — usable with well-organised sources, less so with a messy drive. The free plan's 50 MB storage ceiling forces good hygiene anyway.

One more practical note: usage is metered by model-usage limits that the vendor documents in its help centre, and heavy use of premium models consumes the included allotment faster than simple rewrites. If you are budgeting seats, read that article before committing.

A third note on pricing mechanics. The premium model-usage allowance is the constraint that matters for power users, and the vendor's help-centre article on usage limits is where the real numbers live rather than the pricing page. In my reading of it, different models consume the allowance at different rates, so a habit of picking the largest model for everything drains the included usage quickly while the same work on a lighter model stretches it. Teams I would advise to standardise on a default model per task type and reserve the premium options for the tasks where the difference is visible. The per-seat maths at $29.99 also rewards honest seat auditing: the pricing page is explicit that account sharing is not permitted, so the licence count has to match real users.

What it does not do

It does not give you unlimited premium-model output on the standard plan — the per-user price includes a model-usage allowance with limits documented separately, and truly unlimited generation is a separate, much pricier tier (or was, under the help centre's legacy pricing). It is not a specialist long-form writing environment: no manuscript structure, no screenplay formatting, nothing like the dedicated fiction tools in this catalog. The "30,000+ apps" figure is a vendor claim; my experience is that the extension works in most major web apps but occasionally breaks inside heavily customised enterprise webviews. It is not model-locked (good) but that also means output quality varies by which model you pick, and the default is not always the best. And if the help-centre versus website pricing discrepancy reflects live A/B experimentation, buyers should screenshot the price they are quoted.

Adoption is the other practical limit. The vendor claims 75% enterprise adoption under three months and cites measured hours saved per employee, but every deployment story on the site is a vendor-selected case study. In my experience the failure mode for tools like this is not capability, it is habit: an agent platform that nobody configures becomes an expensive rewrite box. Budget internal onboarding time, name an owner for the knowledge bases, and treat the first month as a pilot with a defined workflow before you scale seats.

One closing note on the free plan, which I would treat as a real evaluation environment rather than a teaser: knowledge bases, custom agents, connectors and deep research are all included, and 50 MB is enough to attach a meaningful working set of documents. My suggestion is a two-week free-plan pilot with one real workflow — one knowledge base, one scheduled agent — and a hard look at whether the daily output earns a place in your inbox before any per-seat conversation starts.

Verdict

I rated TextCortex 3.5: the core platform has grown into a genuinely broad agent-and-writing system with model choice, knowledge integration and scheduling (40% rubric weight), but the honest value calculation depends on which of two vendor prices you are shown and how much included model usage you need (20%), the usability of the extension is a strength while agent retrieval is only middling in my test (20%), the ecosystem — extension coverage, API, languages, integrations — is mature (10%), and reputation signals are mixed: strong enterprise logos and G2 placement claimed by the vendor against third-party complaints about flat output (10%). For a professional who lives in browser tabs and wants one AI assistant that follows them across tools, the extension alone justifies the free tier. For teams, I would pilot the agent layer with a tidy knowledge base before believing the enterprise slide.

Sources

• https://textcortex.com/pricing
• https://textcortex.com/
• https://help.textcortex.com/hc/en-us/articles/21719941813393-Plans-pricing-explained
• https://help.textcortex.com/hc/en-us/articles/46798075584657-Model-usage-limits-and-Extra-Usage
• https://www.capterra.com/p/236373/TextCortex-AI/pricing
• https://www.g2.com/products/textcortex/reviews
• https://undetectable.ai/blog/nb/textcortex-anmeldelse
• https://itechguides.com/products/textcortex
• https://positioniseverything.net/tools/textcortex
• https://ugliai.com/en/ai-tools/textcortex
• https://toolchase.com/tool/textcortex
• https://saleshive.com/vendors/textcortex