TURA WORKSPACES

Communications
  • Chat
  • Mail
  • Meetings
  • Calendar
Work
  • Tasks
  • Project
    Management
  • Forms
  • Flow
Collaborate
  • Vault
  • Notes
  • Whiteboard
  • Knowledge
    Base
Intelligence · one institutional memory, across every app
  • Memory
  • Assistant
  • Agents
Underneath every app
AI
  • Local AI · RAG
  • External LLM
Security
  • Zero Trust
  • RBAC
  • DLP
Governance
  • Audit trail
  • Oversight
  • KVKK & GDPR
Runs on
On-premise (Q1 2027)Cloud SaaS (Q3 2027)
See it in action

Your organisation's AI operating system.
It remembers.

Apps on top. Memory, permissions and audit underneath. All on your own infrastructure.

Tura Workspaces brings your conversations, meetings, tasks, projects and knowledge into one enterprise workspace, with private AI that turns them into institutional memory. It runs on your own infrastructure, and not a single bit leaves the building.

v1.25.0 · Under active development · V1 in development through the end of 2026 · Pilot target: early 2027

Built with an AI-Driven Development Lifecycle (AI-DLC)

Let's show the architecture to your team

Product

One workspace replaces the patchwork of chat apps, task boards, file shares and meeting tools. Every conversation, decision and document lands in a single data model, under one permission graph and one tamper-evident audit trail, and becomes institutional memory your teams can question in plain language.

Every part serves one purpose: to gather the value and keep it with the organisation.

Examples

A logistics companySix months after the operations manager moved on, her successor asks Tura.

Why did we change the cold-chain supplier for the Ankara warehouse?

Memory

“Two audits in the first quarter found temperature breaches in deliveries. The operations review agreed to move to the backup supplier from 1 May.”

  • MeetingOperations review · 11 April
  • VaultQ1 audit report
  • TasksSupplier switch-over

A retail bankA new compliance officer prepares for an audit and asks Tura.

What did we report to the regulator about the card-fraud incident in March?

Memory

“The incident report of 14 March: 212 cards affected, limits reset within four hours, root cause a third-party terminal update. Submitted on 15 March.”

  • VaultIncident report · 14 March
  • MeetingCrisis call · 14 March
  • Chat#fraud-ops

A construction groupA site manager joins a project halfway through and asks Tura.

Which subcontractor built the façade on the Bodrum project, and were there any defects?

Memory

“The façade was completed in 2025 by the subcontractor on contract F-12. Two water-leak defects were reported at handover and fixed under warranty in October.”

  • Project managementBodrum residences
  • VaultWarranty log
  • MeetingHandover review

A broadcasterThe new head of product asks Tura before the next season.

Why did we pause the 4K channel package in the eastern region?

Memory

“Bandwidth tests in three cities showed buffering above target. The product board agreed to relaunch after the CDN upgrade.”

  • MeetingProduct board · 2 June
  • NotesCDN test results
  • TasksCDN upgrade

A manufacturerA quality engineer on her first week asks Tura.

What did we change on line 3 after last year’s quality recall?

Memory

“A second torque check was added at station 7 and the bolt supplier was replaced. The recall rate fell from 1.2% to 0.3%.”

  • Notes8D report
  • TasksStation 7 change
  • WhiteboardLine 3 layout

A hospitalA radiology team lead covering a night shift asks Tura.

What is our agreed procedure when the MRI booking system is down?

Memory

“Bookings move to the shared radiology calendar and the front desk calls affected patients. Agreed in the November incident review.”

  • MeetingIncident review · November
  • NotesDowntime procedure
  • CalendarRadiology fallback

Answered only from what the person asking is allowed to open. Illustrative examples.

Value

What Tura Workspaces brings to your organisation

Each one rests on a mechanism in the product; those marked Next are on the roadmap.

KnowledgeWhat your organisation knows, and how it uses it

Institutional memory

Expertise stays when employees leave.

Unified search

One search across chat, files, notes and meetings.

Natural-language questions

Anyone, new hire or veteran, gets answers by asking, not digging.

Accountability

Every decision traceable to its reasoning.

Trust & controlWhere your data lives, and who can reach it

Data sovereignty

Your data stays where you put it.

Security

Encrypted with keys only you hold.

Compliance

Audit-ready by default, not by project.

Management visibility

Leaders see status without reading private messages.

OperationsRunning the business, day to day

Process tracking

Meeting action items become tasks on approval; nothing agreed gets lost.

One data model

No copy-paste between tools: every app works on the same record.

Continuity

Operations continue through any outside outage.

Cost

One licence replaces a stack of subscriptions.

IntelligencePeople and AI agents, working on the same memory, inside your organisation

Conflict detection Live

Duplicated work and contradicting facts across units surface before they cost you.

Assistant Next

The language model you choose, switched on by you, every call on the audit trail.

Agents Next

Department agents that act only on what that department may reach.

Tura Flow Next

When something happens, Tura acts, on rules you define once.

Apps

Every app on one data model

Screens from the running product. Select any image to enlarge it.

Home

Chat

Meetings

Tasks

Project management

Notes

Whiteboard

Vault

Memory

Oversight

Admin & compliance

Comparison

Where we match, and where we differ

Our reference is Mattermost. The difference sits in three places: one data model for every app, conflict detection across units, and an audit chain that cannot be rewritten.

Tura WorkspacesSlackMicrosoft TeamsMattermostRocket.ChatopenDesk
Runs on your own hardware
AI that runs on your own hardware
One data model for every app
Cross-unit conflict detection
Guest and vendor isolation in the data model
Tamper-evident audit chain
Customer holds the encryption key
Vendor jurisdictionTürkiyeUSUSUSUS / BrazilGermany

Built into the productPartly, or only as an add-on or higher tierNot offered

Based on each vendor's public documentation, October 2026.

Security & compliance

Built for the strictest audits, so every organisation gets the same guarantees

Four guarantees built into the design, mapped to the rules auditors ask about. Everything here describes an on-premise deployment, the default; the managed cloud option planned for 2027.3 will come with its own processing terms.

Data stays home

Every component, the AI included, runs on your hardware. Nothing leaves unless you switch it on.

Your keys

Encryption keys live in your own KMS. Switch it off and the stored files become unreadable.

Tamper-evident audit

Every access is chained to the one before it; alteration and deletion are refused by the database.

Isolation by design

Who sees what lives in the data model. A guest cannot even learn that a resource exists.

How the guarantees meet the rules

RegulationData stays homeYour keysTamper-evident auditIsolation by designWhat it means in practice
KVKKTürkiye · Law 6698No transfer abroad; technical measures in place, and breach scope comes from the audit log.
BRSA IT RegulationTürkiye · bankingSystems stay in the country; customer secrets stay confined to the group they belong to.
GDPREuropean UnionPrivacy by design and by default; the 72-hour breach scope becomes a query, not an estimate.
NIS2European UnionAccess control, MFA from your directory, cryptography and a countable supply chain.
DORAEU · financial sectorA shorter third-party ICT chain, encryption under your own key management, and a real exit strategy.
AI ActEuropean UnionHuman approval before action, answers with their sources, and auditable use of data.
CLOUD ActUnited StatesNo vendor holds your data, so there is nothing for a foreign order to compel.

The guarantee this rule relies onNot what this rule asks for

Three tests your own team can run

Every claim here can be verified by your own team during a pilot.

1Read your own network logs0 outbound connections

With external AI left off, install it on your network and watch the firewall.

2Recompute the audit chainEdits refused

Call the verification endpoint yourselves, then try to alter a record.

3Switch off your key storeFiles unreadable

The proof the key is yours: without it, the product is helpless.

Who is responsible

On-premise, your organisation is the controller and we never access the data, so we are not even a processor. Your obligations stay yours; we make them technically possible and produce the evidence an audit needs.

Certification
  • DoneInternal security pass: authorisation, secrets, headers
  • PlannedIndependent penetration test
  • PlannedISO 27001 and SOC 2

A full regulation-by-regulation map is shared with your security and compliance team in the first conversation.

About

What we focus on, who we work with, and where the product genuinely stands today.

Any organisation that values its knowledge · Data sovereignty and security · AI that runs inside, with any outbound link your decision

Status and roadmap

What is ready today, and what comes next

v1.25.0Tagged release, the 25th update since v1.0 in August
2,706Commits, 77% written by AI agents, behind human release gates
8,800+Automated tests, run on every change
1,703Locked decisions, each written down with its reasoning

Versions

On-premise workspace + local AI memory

Q4 2026 · In development through the end of the year; 25 releases so far since v1.0 in August 2026.

What's in it
Chat, tasks and projects, notes, a file vault, meetings with a whiteboard, and a memory that answers with its sources, organised around your own chart of units.
Built in
Compliance Mode, legal hold, retention, a data-loss ceiling and an audit trail that cannot be rewritten.
Where it runs
Your own servers; the web and the desktop from a single client, offline on the desktop, and guests let in without exposing the rest of the organisation.
Local AI
The memory runs on your own hardware: it reads what your teams write and answers with its sources, without anything leaving the building.

Server deployment + your own LLM, governed

Q1 2027 · From a running product to a production installation, with the AI layer switched on.

Deployment
A production-grade installation on your own servers, set up with a single command.
AI
The Assistant, connected to the language model you choose under your own rules, and model-written summaries on top of the memory.
Milestone
First pilot deployment at a customer site.

AI agents + automated workflows

Q2 2027 · AI that acts as well as answers, still inside your organisation.

Agents
Agents join your teams as colleagues: they take tasks, draft and summarise in the same capsules, with access to exactly what that department may reach. People approve.
Automation
Tura Flow: event-to-action rules that administrators define once and every unit can reuse.
Milestone
Agents in pilot use with a customer.

Managed cloud option + Kubernetes scale

Q3 2027 · On-premise stays the default; a managed edition for teams without their own data centre.

Managed cloud
A multi-tenant edition run for you, with each organisation isolated and its keys still in its own hands.
Scale
Kubernetes deployment for large installations, on your servers or ours.
Milestone
First managed-cloud tenants live.

Refinements from customer deployments

Q4 2027 · A release shaped by what our customers learn running Tura in production.

Refinements
Improvements and fixes prioritised from customer deployments and their feedback.
Deployments
Smoother installation and upgrade paths for new customer sites, on-premise and in the cloud.
Milestone
First long-term support release, kept on security updates for two years.

After V1, releases are named by year and quarter (2027.1, 2027.2, …). The last release of each year, starting with 2027.4, is a long-term support release, kept on security updates for two years.

How it is built

An AI-driven development lifecycle

Its core is discipline rather than speed: the gates stay with a person, the sessions have no memory.

AI-Driven Development Lifecycle

Human in the loop. AI orchestrates. Agents accelerate.

Human(in the loop)Scope · Approvals · Decisions
AI Orchestrator(Coordination & Control)Plans · Routes · Monitors · Learns
AI Agents(Specialized)Execute · Deliver · Report
1. THE SPRINT CYCLE

From idea to value, in structured sprints

1. PlanDefine · Analyze · Prepare
2. BuildDevelop · Integrate · Test
3. VerifyTest · Validate · Ensure quality
4. CloseRelease · Document · Handover
5. DeployMonitor · Support · Improve
Human LayerControl & Oversight
  • Define product scope & priorities
  • Answer functional requirement questions
  • Set acceptance criteria
AI LayerAI Orchestrator
Orchestrates agentsAssigns · Coordinates · Monitors
Manages dependenciesCode · Infra · Data · APIs
Handles deliveryBuilds · Tests · Releases
Ensures quality & securityChecks · Reports · Mitigates
Learns & improvesFrom feedback & metrics
2. SPECIALIZED AGENTS

Expert agents. Focused responsibilities. Working in their sprint phase.

LensUI/UX Agent
PLAN
  • Analyzes user needs & flows
  • Designs UI/UX and visual assets
  • Prepares prototypes & wireframes
  • Works with the AI layer on design decisions
Works in: Plan phase
KeystoneSoftware Architecture Agent
PLAN
  • Designs system architecture
  • Applies SOLID principles and design patterns
  • Defines the technology stack
  • Creates technical specifications
Works in: Plan phase
GuardSecurity Agent
PLAN
  • Reviews security architecture
  • Analyzes code for vulnerabilities
  • Checks dependencies
  • Reports and suggests fixes
Works in: Plan phase
AssayQA Agent
VERIFY
  • Designs test strategy
  • Runs automated & manual tests
  • Validates quality and performance
  • Reports results & defects
Works in: Verify phase
Agents work only in their assigned sprint phase. Their outputs are evaluated by the AI layer and the Human layer.
ContinuityKeeps the memory. Ensures traceability.
RepositoryCode · ArtifactsSprint filesPlans · DesignsDecision logChoices · RationaleModule specificationsDesign · TechnicalSessions have no memoryStatelessPersistent project artifacts ensure continuity.
The principles of the methodDecisions, checks, deployment, releases
Decisions are written down, append-only
Every architectural decision becomes a dated record, and no code quietly contradicts a locked one. When a decision changes the old entry is superseded rather than deleted — so why something is the way it is can still be read later.
Every criterion has a machine check
The evidence for an acceptance criterion cannot be "someone looks at it": each one is bound to a check that runs on every change. And a check is deliberately broken before it is trusted — a check that cannot fail is not the same as a check that passes.
Deployment stays with a person
Merge approval and the release tag belong to the product owner. Deployment is not automatic: a command is handed over, the install is run by hand, and the running version number is read back. A sprint closes only when the deployment reports the new number.
Every release is accounted for
Each release ships with a software bill of materials (SBOM) and signatures, and the container images it depends on are pinned by digest. If a pinned image changes upstream, the build stops and the change becomes a recorded decision rather than a silent update.

Investor relations

Where we stand, and the conversations we are open to

Tura Workspaces is growing on its own momentum, and we are open to investment and strategic partnership offers.

Stage

Early stage and founder-led, with the product in active development and a clear roadmap through 2027.

Built so far

About three months from the first decision in July 2026 to v1.25, with 77% of commits written by AI agents behind human release gates.

Where we are heading

Production deployments in 2027, AI agents and automated workflows, and a managed cloud edition for organisations of every size.

Ways to work together

Investment, a strategic partnership, regional distribution or white-label, or a partnership that brings enterprise customers.

Why now. Organisations want AI on their own knowledge, but rules such as KVKK, the GDPR and the EU AI Act, and the risk of company data leaking into outside AI services, keep that knowledge from leaving the building. Tura brings the AI to the data.