Case Study · Live in Production · Built Solo by Directing AI Coding Agents
TIKIR — CONVERSATION IS THE INTERFACE
Tıkır is an AI-native business management platform. Its core idea is simple: instead of navigating panels
and filling forms, you tell the system what happened — and the data lands in the right place, correctly
processed, with the follow-on work automated. Hergele Mobility runs on it as the first customer. The tool
is company-internal, so no screenshots are published here.
Why Not Just Another CRM
In a traditional CRM, the operator serves the software: find the right panel, configure the right fields,
write descriptions, remember the conventions. Every one of those steps is friction — and friction is why
CRM data goes stale, why field notes never get entered, and why the "system of record" quietly stops
recording. Tıkır inverts the relationship:
Traditional CRM
The operator serves the software
Navigate panels to find the right screen
Configure fields and settings per record
Write descriptions by convention
Data entry is a chore → records go stale
Automation is a separate product to configure
Tıkır
The software serves the operator
Say what happened — typing or speaking
The assistant places the data in the right structure
— and, more importantly, processes it correctly
Follow-on work triggers automatically
Ask questions in plain language, get answers from live data
An Order, End to End
The clearest example is an order. In most companies, an order kicks off a relay of manual steps across
disconnected tools. In Tıkır, placing the order starts a chain that runs on its own:
When an order comes in, the vehicle's serial number and full specification are created automatically —
and because Tıkır connects to Wagger, the chain extends into production notes.
From there, every datum needed for after-sales service is retained on the same record. It is not a CRM
bolted onto a company; it is an end-to-end system: sales → production → after-sales, one identity.
What Conversation Unlocks
Voice-first data entry. The single biggest feature in practice: records can be created just by
speaking. In the field — in a warehouse aisle, next to a vehicle, mid-visit — nobody opens a laptop to
fill a form; they say it, and it's recorded correctly. Field data actually gets captured.
Smart stock, by asking. "Do we have these in stock?" or "An order just came in — can we make
the date?" are questions, not report-building sessions. The assistant answers from live inventory
and commitments.
Procurement on a shared platform. Suppliers work on the same platform, so purchase follow-ups
happen where the data lives instead of over scattered email threads.
Automation as the default. Most follow-on work — record creation, linking, notifications,
after-sales bookkeeping — is wired to happen automatically rather than being someone's checklist item.
Under the Hood
Multi-tenant architecture with row-level security — each organization sees only its own data.
AI assistant (Claude) that both answers questions and performs actions on live data.
Built solo by directing AI coding agents, on the same docs-as-code discipline as
Wagger: decision registry, single-source values, automated record gates.
Tıkır is currently a company-internal tool; screenshots and customer data are not published. This page
describes the system's approach and structure only.