Teklens
Teklens is a Swiss AI Product Manager for software teams. It reads the real code alongside Jira, Confluence and past decisions, then turns that context into specs, Jira tickets, code-grounded estimates and release notes, without replacing the existing tools.
What is Teklens?
Teklens presents itself as an AI Product Manager for software teams: not another issue tracker, not a coding agent, not one more roadmap tool, but a layer that takes over the coordination work around product decisions. Its technical core is the Context Engine, which builds a living knowledge graph from the code itself, Jira tickets, Confluence pages, Git history and product meetings (Beta). Four retrieval techniques are named: semantic search, knowledge graph, LSP traversal and cross-repo lookup. The argument repeated across the site is blunt — Teklens has read your code, generic tools are guessing.
The tool covers the full Discover, Define, Build and Operate cycle, plus a Fast Lane for small items, and surfaces where the user already works: inside the Jira ticket, or in its own web app at app.teklens.ai. Native connectors today are Jira, Confluence, GitHub and files; GitLab, Bitbucket, Slack, Teams, SharePoint and authenticated MCP connectors are announced as roadmap. Five functional blocks carry the product — Spec to Ticket, standup automation, role handover, Sprint-Memory and the Portfolio Cockpit — alongside a Skill Studio for reusable templates and checks and an AI spend management view that attributes model and token costs to teams, projects and workflows. More than 40 qualified use cases are documented with example prompts.
Model choice is a per-project setting under a Private AI heading: Claude, GPT or Gemini, with sensitive projects on EU-hosted models and strict requirements routed through a European proxy. The figures the site builds its case on are third-party studies it cites, not its own measurements: 73% of a PM’s time spent on coordination, 27% on product strategy, 52% on unplanned troubleshooting. Teklens positions itself explicitly against Jira Rovo, which stays inside Atlassian and does not know the code, against coding agents, which speed up execution but leave the upstream and downstream untouched, and against an in-house RAG, which gives full control at the price of a second product to maintain. Behind it: a team of seven founders and engineers in Zurich, and AI-readiness research run with ETH Zurich and HSG. The interface exists in German and English only.
What it does
- Reads and semantically indexes every code repository, then links code, tickets, docs and decisions into a living knowledge graph
- Turns a spec into a Jira epic or ticket, with acceptance criteria drawn from real code paths
- Estimates effort, risk and complexity from the actual code, with the reasoning behind each figure
- Prioritises by business impact and technical feasibility in an Impact & Feasibility Matrix
- Detects spec drift: the gaps between what Jira plans and what the code actually does
- Generates release notes, living documentation, PRDs, stakeholder reports and pull request summaries
- Triages bugs and incidents back to the commit most likely at fault and prepares the fix
When to use Teklens / When not to
A quick filter to help you decide if Teklens is the right fit.
When to use Teklens
- Product managers who lose most of their week to coordination, admin and support instead of product strategy — the site puts that share at 73%
- CPOs, CTOs and Heads or VPs of Product and Engineering, who each get a dedicated page on the site
- Teams already working in Jira, Confluence and GitHub that want an AI layer without migrating anything
- Engineering teams whose estimates land 50 to 200% off and who want effort, risk and complexity justified by the actual code
- Organisations running an AI due diligence: per-project model choice, audit logs, SSO on Enterprise and an announced Swiss and EU hosting
When not to use Teklens
- Individuals and freelancers: the terms of business address the “using company” and no individual plan is offered
- Teams that work outside Jira, Confluence and GitHub — the only native connectors today; GitLab, Bitbucket, Slack, Teams and SharePoint are announced as roadmap
- Anyone looking for a coding agent: the site states plainly that Teklens is “not another coding agent” and stays on the product-management side
- Teams that need a permanently free tier: there is only a 14-day trial, and free seats are limited to collaborators and reviewers
- Teams that need a public API, a mobile app, complete product documentation (/docs still reads “Docs in Arbeit”) or an interface in a language other than German or English
How to use Teklens
A typical end-to-end flow, from setup to results.
- Fill in the form on the demo page with a business email address, the company name and your main bottleneck — no sales call required
- Install Teklens in read-only mode on a public repository, a sample repository or your own; Jira is not needed to get started
- Get a first result in under ten minutes: spec drift, undocumented components, risky dependencies, stale Jira links and one improved, code-grounded ticket
- Treat week 1 as the activation window — the site defines activation as the first finding accepted or exported into the workflow, not as opening an account
- Connect Jira, Confluence and GitHub so the Context Engine can index the full picture
- Set, per project, which models are allowed (Claude, GPT or Gemini) and configure the team-wide privacy mode
- Supply your own model API key on the TEAM plan, since LLM tokens are billed by the model provider and not by Teklens
- Invite collaborators and reviewers — free and unlimited — to check the evidence and give the sign-offs
- Work either from Jira, where the answer lands in the ticket, or from the web app at app.teklens.ai
- Open an artefact in Claude Code, Cursor, VS Code or IntelliJ in one click, with branch, context and acceptance criteria already loaded; alternatively, book a 30-minute founder call split into context, live demo and next step
Pros & Cons
Pros
- Grounded in the real code rather than documents alone — the differentiator repeated across the whole site
- No migration: it connects to Jira, Confluence and GitHub, and Jira stays the system of record
- Low-friction entry: a 14-day read-only trial, a first finding in under ten minutes, with no Jira connection and no sales call
- Unlimited free collaborator and reviewer seats — only active users are billed
- Explicit AI governance: per-project model choice, announced Swiss and EU hosting, audit logs and SSO on Enterprise
- A clear commitment that customer data is not used to train models, stated in three places including the Service Agreement
- A single public TEAM price and monthly termination to the end of the month on both plans
Cons
- The terms of business declare themselves a draft to be legally reviewed “before the public launch” and point to an individually signed pilot contract, while the pricing page sells immediate self-service access
- The publisher is announced as a company in formation, “Teklens GmbH (i.G.)”, yet the imprint publishes a complete set of Swiss commercial-register identifiers, which the register attributes to another company, Contem GmbH
- EUR 42.00 is not the full cost of the TEAM plan: it is per active user per month, LLM tokens run on the customer’s own API key and are billed by the model provider, and the terms state prices exclusive of VAT, which the pricing page does not repeat. The ENTERPRISE price is not public
- Native connectors are limited to Jira, Confluence, GitHub and files; GitLab, Bitbucket, Slack, Teams and SharePoint remain on the roadmap
- Product documentation is incomplete — /docs reads “Docs in Arbeit” — and there is neither a public API nor a mobile app
- No customer reference is published: the site says names, figures and case studies are kept back pending explicit permission, and its headline gains are given as orders of magnitude from conversations, “to be proven in the pilot”
- No security certification is published (no SOC 2, no ISO 27001), no Article 27 GDPR representative is designated, no support or data protection email address exists, and the interface is limited to German and English
Pricing & Plans
There is no permanently free plan. Access starts with a 14-day trial, after which the entry point is the TEAM plan at EUR 42.00 — or CHF 40.00 — per active user per month, stated exclusive of VAT in the terms of business. Collaborator and reviewer seats are unlimited and free of charge, so only active users are billed. LLM tokens are not included on this plan: they run on the customer’s own model API key (Claude, GPT or Gemini) and are invoiced by that provider, not by Teklens. The ENTERPRISE plan is priced on request and is not published.
- immediate product access and a focused 14-day trial
- read-only analysis of a public or sample repository without Jira
- a first code-grounded finding in under ten minutes
- unlimited free collaborator and reviewer seats
- team skills and internal rules
- controlled LLM access and a team-wide privacy mode
- cloud agents and automations on shared team context
- central billing and administration
- everything in TEAM
- tokens included on a single invoice with no model configuration to handle
- an optional dedicated endpoint with its own AI governance and a lower per-user cost
- per-project model choice including EU and Swiss hosting for sensitive projects
- a dedicated account manager
- an SLA
- audit logs and SSO
- Both plans run on an open-ended contract
- billed monthly and terminable for the end of the current month
Data, GDPR & hosting
A consolidated view of how Teklens handles your data.
GDPR overview
The FAQ and the pricing page claim “GDPR-konform, Data Residency CH und EU”. The privacy policy (version 2.0, last updated September 2025) is concrete on several points: it separates Teklens’s role as controller for the website from its role as processor for the product; it cites the EU GDPR for special categories of data; Google Tag Manager and the associated analytics load only after consent for EU visitors; transfers outside Switzerland and the EU are covered by the European Commission’s standard contractual clauses; and terms section 08 provides for a separate processing agreement (AVV, i.e. a DPA) whenever Teklens processes data for a customer. Two gaps are explicit: no Article 27 representative is designated although the publisher sits in Switzerland, outside the EU, and no DPO or dedicated privacy address is named — only a postal address, with the Swiss FDPIC (EDÖB) cited as supervisory authority.
Who owns the data?
Under Service Agreement section 04, all rights to the customer’s code and business data remain exclusively with the customer, and customer-specific work products — reports, configurations, documents — pass to it in full once paid. Terms section 10 confirms the customer keeps every right to the content it brings in and grants Teklens only the usage right needed to deliver the service; the platform, its components, trade marks and logos stay with Teklens. The privacy policy adds that content from connected systems is processed solely to deliver the agreed service, that customer data is never used to train models, and that where Teklens processes data for a customer, that customer is the controller and Teklens the processor.
Reuse rights
The customer may reuse what it brings and what it pays for without asking permission: it keeps all rights to its own content, and customer-specific work products transfer in full on payment, so specs, tickets, reports and documentation can be used and circulated internally at will. The limits sit on the platform itself — terms section 10 and Service Agreement section 01 grant a non-exclusive, non-transferable licence for the customer’s internal purposes, valid for the term of the contract only. The customer also decides which systems are connected and at what level of access. A mutual confidentiality obligation applies with no time limit and survives the end of the contract, and the privacy policy states that personal data is not sold on to third parties.
Data retention & training
Hosting summary
Teklens advertises hosting in Switzerland and the EU on four pages — homepage, pricing, product and the Service Agreement — under the wording “Data Residency CH und EU”. On Enterprise, the model is a per-project setting including EU and Swiss hosting for sensitive projects, with an optional dedicated endpoint; strict requirements are routed through a European proxy. The privacy policy names four subprocessors: application data sits with Supabase, website delivery and edge operation with Cloudflare, email is sent through Google Gmail on the teklens.info domain, and audience measurement runs on Google Tag Manager — all four are US vendors. The same policy states that some of the providers used may process data outside Switzerland and the EU, with those transfers covered by the European Commission’s standard contractual clauses. The publisher is subject to Swiss law and cites the Swiss FDPIC (EDÖB) as supervisory authority. No hosting country other than Switzerland is named, “EU” remains a region with no data centre specified, and no third-party certification — SOC 2 or ISO 27001 — is published to back these commitments.
Things to keep in mind
Risks and trade-offs to weigh before adopting Teklens.
- Hosting says two different things. Four pages advertise “Hosting Schweiz und EU” and “Data Residency CH und EU”, while the privacy policy states that some of the providers used may process data outside Switzerland and the EU, in countries where privacy laws may differ, under standard contractual clauses — and the four subprocessors named in that same policy (Supabase, Cloudflare, Google Gmail, Google Tag Manager) are US vendors
- The commercial framework says two different things. The terms of business declare themselves a draft to be legally reviewed “before the public launch” and refer to an individually signed pilot contract, while the pricing page sells immediate self-service access with a 14-day trial
- The publisher’s identity says two different things. The imprint presents “Teklens GmbH (i.G.)” as a company in formation while publishing a UID, a CH-ID, an EHRA-ID, an “aktiv” status and a 2020 articles date; those identifiers belong to Contem GmbH, and the corporate purpose copied onto the imprint is that company’s — consulting and coaching, including psychotherapeutic services — unrelated to software publishing
- The real cost of the TEAM plan exceeds EUR 42.00 per user: LLM tokens are billed separately by Claude, GPT or Gemini on the customer’s own API key, and the terms state prices exclusive of VAT while the pricing page never mentions it
- A tool that writes specs, estimates and acceptance criteria from code invites teams to delegate product judgement to an automated synthesis and to lose the discussion that used to validate it; the “code-grounded” estimate can also convey false precision, and the site itself presents its headline gains as orders of magnitude “to be proven in the pilot”
- Connecting Jira, Confluence and a code repository hands over a team’s entire product and technical knowledge; scope and privacy mode need to be set per project from day one, all the more so as the product claims a persistent memory — a knowledge graph “read once, never forgotten” and a Sprint-Memory that outlives sessions, which has to be reconciled with internal retention rules
- No independent security certification is published, no customer reference is verifiable, and there is neither a support address nor a data protection address — nor an Article 27 GDPR representative, although the publisher is established in Switzerland, outside the EU
Setup & Integrations
Technical difficulty
Low to try, moderate to deploy. The trial needs no heavy technical work: a form, then a read-only analysis of a public or sample repository, with a first finding promised in under ten minutes and no Jira connection. Full rollout is more demanding: connecting Jira, Confluence and the code repositories assumes administrative rights on those tools and API tokens, and the Service Agreement makes the customer responsible for providing the accesses and naming a contact person. The TEAM plan also requires supplying your own model API key. Onboarding, enablement and tool connection are offered as a paid service.
Deployment
Integrations
Supported languages
Behind Teklens
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Teklens.
Frequently asked questions
What is Teklens?
Do we have to replace Jira or Confluence?
How much does it cost?
Is there a free trial or a free plan?
Which AI models are used?
Is our data used to train models?
Where is the data hosted?
Who owns the code and the data?
What is the state of the commercial and legal set-up?
Is there an API, a mobile app or full documentation?
Should you pick Teklens?
Teklens occupies a well-defined slot: product coordination, not coding. Its differentiator is legible in the site’s own discourse — the context graph includes the real code, where neighbouring tools stop at the ticket — and the entry cost is low: a 14-day read-only trial, a first finding in ten minutes, no sales call. Pricing is unusually readable for the category, a single public TEAM figure with unlimited free collaborator seats, provided you remember that EUR 42.00 is per active user per month, excluding VAT, and that LLM tokens are billed separately by the model provider on your own API key.
Maturity is where the picture needs qualifying, and the site itself supplies the qualifications. The terms of business are declared a draft pending legal review before the public launch and point to an individually signed pilot contract, while the pricing page sells immediate self-service access. The imprint presents the publisher as a company in formation, “Teklens GmbH (i.G.)”, yet publishes a full set of Swiss commercial-register identifiers. Swiss and EU data residency is advertised on four pages, while the privacy policy admits that some providers may process data outside Switzerland and the EU, and the four named subprocessors are US vendors. Add to that no published customer reference, no security certification, incomplete documentation and no public API.
The team is experienced and identifiable, with seven founders and engineers in Zurich and research ties claimed with ETH Zurich and HSG. The target audience is narrow but well served: product teams already working in Jira, Confluence and GitHub who want their specs, estimates and handovers anchored in the code. Teams outside that stack, or those needing a signed-off contractual framework today, should wait for the pilot phase to close.
- Choosing a selection results in a full page refresh.
- Opens in a new window.