A Dental Clinic Brings Laboratory Production In-House

Outsourcing transfers work, but it does not transfer accountability to the customer. When a patient-specific product arrives late or wrong, the clinic still owns the promise. Bringing a dental laboratory inside can restore control, yet it also makes the clinic responsible for design data, materials, machines, technicians, quality, capacity and every remake. Insourcing succeeds only when those elements become one governed production thread.

The source described a control crisis

On 1 March 2026, clinic founder Mikhail Agami described his decision to build an in-house laboratory on the Forbes Russia Expertise platform. He said outsourced laboratories became unreliable in 2022 as contractors disappeared or struggled with logistics and quality.

The account was a first-person business case, not an independently audited study. Its figures and assessments should be read as claims by the author about his clinic's experience.

The core problem was a mismatch of responsibility and control. Clinicians remained accountable to patients for diagnosis, plan, schedule, ordered construction and final assessment, while the manufacturing process sat beyond their direct influence.

Insourcing changed the boundary. The clinic gained access to work in progress, technical decisions and production evidence, but also inherited investment risk, staffing, maintenance, material control and utilisation.

Make or buy was a control-design decision

Outsourcing is useful when a specialist supplier aggregates expertise and equipment across customers. It converts fixed cost into variable cost and can offer technologies a single clinic cannot economically sustain.

It becomes fragile when lead time, communication, quality or supplier continuity is too uncertain for the customer promise. The right response is not automatically to bring everything inside. Leaders should identify which stages create strategic exposure and which remain better purchased.

A make-or-buy map can score each stage by clinical accountability, data sensitivity, turnaround need, volume, capital intensity, scarce skill, supplier depth, traceability and consequence of failure.

The result may be a hybrid. A clinic can own scanning, design review, urgent milling, inspection and remake data while retaining qualified external capacity for specialist materials, unusual processes or peaks.

The decision case needed explicit questions

1. Which failure modes harm the patient promise even when the supplier is contractually responsible?
2. What demand is stable enough to support people, equipment and material inventory?
3. Which capabilities must be available within hours rather than days?
4. Which processes require licences, evidence or specialist competence beyond the clinic's plan?
5. How will internal cost compare after maintenance, scrap, training and idle capacity are included?
6. What contingency remains when the in-house laboratory itself stops?

Insourcing should be approved against several demand scenarios. A laboratory sized only for a peak may carry expensive idle capacity; one sized for the average may immediately recreate an external queue during growth.

The author described a rapid internal launch

Agami said the clinic formed a separate investment fund and aimed to start production quickly without disrupting patient work. He did not disclose the investment amount, launch duration or patient volume.

The equipment set included scanners, three-dimensional printers, a milling machine and other digital systems. The clinic selected suppliers from China for availability, delivery speed and practical service.

The author said the quality was no lower, and sometimes higher, than European alternatives in his experience. That observation is not evidence about all Chinese or European equipment. Each application still requires its own acceptance, support and process results.

The initial team combined a chief technician with experienced specialists. The clinic then expanded through internal training, including people without previous laboratory experience, and established a separate ceramics direction.

Rapid launches create parallel risks. Machines arrive while layouts, recipes, documentation and roles are still forming. Experienced people may compensate through memory, hiding weaknesses that appear when new staff or more volume enters.

The product began as controlled information

A patient-specific construction cannot be separated from the data that define it. The source described one digital contour from three-dimensional scan and computed tomography to final construction and stored patient profile.

The production article does not assess diagnosis or treatment. Operationally, the point is that approved clinical inputs must cross a defined interface into manufacturing without ambiguity or uncontrolled reinterpretation.

An order package can include identity controls, authorised prescription, relevant geometry, material and shade requirements, design constraints, requested date, approvals and revision state. Missing information should stop release rather than trigger an informal guess.

File naming alone is unsafe. A system needs unique case identity, revision, source, timestamp and status. Every exported design and machine job should trace back to the approved order.

Access follows role. Clinical staff own the treatment decision and approval boundary; designers interpret the manufacturing data; technicians execute controlled processes; quality staff release evidence. The digital system should preserve who changed what and why.

Design became a manufacturing release

A digital model may look persuasive on a screen but remain difficult to make. Designers must account for material, minimum section, tool access, printer orientation, support removal, finishing allowance and the interfaces used during final work.

Automated suggestions do not remove review. Software settings, libraries and algorithms can change geometry. Version control should connect the design environment to the approved machine route.

Design approval is a gate. Once released, the exact file and parameters enter production. If a later change is needed, the old version remains visible and work in progress is contained.

Reusable templates improve speed only when their scope is defined. A template should not silently override case-specific requirements or carry an obsolete machine compensation.

Printing and milling required different controls

Additive production builds geometry layer by layer from a digital job. Output depends on material identity and condition, machine setup, exposure, orientation, supports, cleaning and post-processing.

A printed object is not complete when the platform rises. Residual material, support removal, washing, curing or other approved finishing steps determine the final state. The specific route depends on the material and intended use.

Milling removes material from a blank using tools and programmed paths. Blank identity, lot, mounting, tool condition, calibration, cooling or extraction and finishing influence dimensional result and surface.

Tool wear can create gradual drift before a visible failure. Replacement rules should combine usage, observed condition and process evidence. A new tool or blank supplier is a controlled change, not a purchasing substitution alone.

Neither route should be credited with generic medical performance. The laboratory must use materials, processes and release criteria appropriate to its authorised work and applicable requirements.

Ceramic finishing remained a skilled process

The source noted a separate ceramics direction because final appearance and quality require focused competence. Digital production does not eliminate manual judgement and craft.

Finishing can involve adjustment, layering, colouring, firing, polishing and inspection under controlled conditions, depending on the construction. Each repeated heat or material step needs an authorised route and recorded status.

A skilled technician can rescue a marginal upstream output, but repeated rescue hides process loss. The laboratory should distinguish normal finishing from correction caused by design, milling, printing or input errors.

Visual judgement benefits from standard lighting, reference conditions and documented approval. Screens, lamps and materials age, so the comparison environment also requires control.

Material genealogy protected the case

Personalised production uses small quantities across many unique orders, which makes informal material handling tempting. Traceability should connect each case to material type, lot, expiry or usable state, and relevant processing history.

Storage conditions matter. Humidity, light, temperature, contamination and opened-container time may affect different materials. The laboratory needs rules based on validated instructions, not one generic cabinet policy.

Substitution requires review. Similar colour, shape or supplier description does not establish technical equivalence. The change may affect design compensation, machine parameters, finishing, inspection or later recreation.

Inventory should balance continuity and obsolescence. Critical materials and machine consumables need protection against supply interruption, while excessive stock can expire or become incompatible with a changed process.

Quality release separated production from use

Quality control begins at the order and continues at every handoff. Final inspection cannot reliably discover all errors in identity, revision, material or hidden process history.

Release criteria can include correct case and design, dimensional and contact checks appropriate to the product, surface condition, absence of visible manufacturing damage, material and route records, and completion of required approvals.

Inspection equipment and reference artefacts need calibration or verification. A measurement result without known method capability creates false confidence.

Nonconforming work should be identified and physically or digitally contained. The team decides rework, remake or rejection under authorised rules, records the reason and protects the original evidence.

The remake loop was a learning system

The source said stored digital profiles allowed a construction to be recreated without repeating clinical stages when replacement was needed. Operationally, reliable recreation requires more than retaining a shape file.

The archive should include approved design revision, material, machine route, critical parameters, finishing state and release evidence. It must also account for changes in software, equipment and available material over time.

Every remake needs a reason code. Data capture error, design issue, manufacturing defect, damage in use, changed requirement and lost item lead to different corrective actions.

A rising remake count can reflect growth, not deterioration, so rates need a denominator and segmentation by stage, product family and cause. Leaders should review cost and lead time as well as frequency.

Corrective action closes the loop when evidence changes a template, training, machine condition, supplier control or order interface. Quietly remaking the item restores the case but leaves the system exposed.

Capacity depended on product mix

A printer or mill nameplate does not define laboratory capacity. Scan review, design, nesting, machine cycle, washing, finishing, ceramics, inspection and approval compete across cases with different complexity.

Planning should identify the constrained resource by period. Buying another machine does not help if design approval or ceramics is the queue. Adding technicians does not help if one mill or material route remains the limit.

Urgent work needs an explicit policy. Constantly inserting priority cases destabilises ordinary promises and hides insufficient capacity. A reserved window or defined escalation makes the trade-off visible.

Maintenance and training consume capacity but protect future output. Excluding them from the plan makes every normal intervention appear as an unexpected delay.

A compact operating scorecard included

• Orders released on time and complete at the manufacturing interface.
• First-pass acceptance and remakes by attributable cause.
• Lead time by stage, product family and priority class.
• Machine availability, queue, changeover and maintenance response.
• Material yield, scrap and expired inventory.
• Training coverage and cases requiring senior rescue.

The claimed gains needed careful reading

Agami claimed that lead time for fixed constructions fell from 10–14 days to three days, output rose 100%, and margin increased 30%. He also described lower logistics costs and fewer remakes.

The article did not disclose periods, definitions, volumes, capital base, depreciation, labour allocation or audited records. The figures demonstrate the author's reported experience, not a universal forecast for another clinic.

A sound business review compares full internal cost with an external benchmark. Internal cost includes equipment, space, finance, maintenance, software, materials, scrap, training, management, quality and idle capacity.

Benefits include avoided supplier margin and logistics, faster response, capacity, knowledge, data continuity and service differentiation. Some are financial; others reduce risk. They should not be counted twice.

Insourcing did not end supplier management

The laboratory still depends on equipment vendors, materials, software, spares and specialist services. It has shortened one chain while creating several new technical ones.

Supplier selection should test availability, service response, documentation, parts, training, software continuity and evidence for the intended process. Purchase price alone is a weak proxy for lifecycle control.

The company also needs a continuity plan for machine outage, cyber incident, staff absence or material shortage. A qualified external laboratory may remain a useful recovery route even after normal work moves inside.

The clinic took back the work

For service businesses in Russia, the case illustrates a broader rule: when accountability cannot be outsourced, critical production may deserve stronger direct control. But ownership only moves risk; it does not remove it.

The laboratory became strategic when clinical inputs, design, machines, materials, technicians, inspection and archive formed a traceable thread. Each handoff had an owner, and every remake could return evidence to the stage that created it.

The real choice was not simply external versus internal. It was an opaque promise versus a governed capability. Insourcing worked to the extent that the clinic learned to operate a small manufacturing system while preserving the judgement and responsibility of its service.