$ proof

Proof, measured — not promised.

Synthetic Engineering is built by the same senior team as Synthetic Users — and before that, the product studio Impossible, which shipped named work for Roche, Google, Samsung and Fairphone. Every workflow carries a baseline and a delta.

Delivered for organisations including

01 / Case points

Described by mechanism. We do not publish invented numbers — specific, cleared figures and named references are shared under NDA at the intro call.

Documentation /0.1

Authoring that keeps pace with engineering

Industrial manufacturer · regulated technical content

Problem: A structured-authoring team maintained multiple document families across product variants and markets. Time-tracking in the diagnostic showed the largest single block of authoring hours went to cross-checking released engineering data — specifications, references, variant applicability — not to writing.
What we did: Six-week baseline of hours per document type across the delivery line. Then a two-week throwaway prototype: a drafting agent over an LLM API with retrieval on released engineering data, proposing updates with source citations. Two-week limited deployment with a handful of senior authors, every output behind a validation gate owned by the lead author.
How it's measured: Instrumented as hours per document type vs. the pre-agent baseline, plus a citation-accuracy audit on every gated output. Drafting cleared the gate and scaled to the full team; a second candidate — auto-translation of safety-critical passages — was killed at the two-week mark for unacceptable review burden.

Multilingual content /0.2

Multilingual content past release cadence

Global product company · dozens of markets

Problem: Release windows compressed from weeks to days while human review throughput stayed flat. The bottleneck wasn't machine translation quality — it was reviewers re-reading whole files because they couldn't trust which segments needed attention.
What we did: Mapped the data landscape first: translation memories, terminology and style assets per language tier. Staged an LLM post-edit pipeline inside the existing translation stack — no platform swap — starting with the highest-volume language pairs, with reviewer-in-the-loop sampling QA and per-segment confidence routing deciding what humans actually read.
How it's measured: Measured as reviewed words per reviewer-hour by language tier, with quality-audit error rates held against the pre-pipeline baseline. Top tiers cleared the gate and expanded tier by tier; the lowest-volume tiers stayed fully human — the delta never justified the review overhead.

Engineering change /0.3

Change impact, traced in minutes

Industrial group · ERP change records

Problem: Every engineering change order triggered a manual hunt: analysts traced affected documents, catalogues and training content by memory and spreadsheet, typically days per change, with misses surfacing months later as downstream defects.
What we did: Two-week prototype on off-the-shelf orchestration: an agent walking change records against product structures and content references, producing a ranked impact list with confidence per item. Kept deliberately human-gated — the analyst signs off the list; the agent never edits content.
How it's measured: Measured as analyst time-to-impact-list vs. baseline, and missed-impact rate audited against a set of historical changes with known outcomes. Cleared the gate on time; the miss-rate audit is what earned analyst trust and adoption.

Master data /0.4

One defensible dataset, many workflows

Equipment manufacturer · product & catalogue data

Problem: The same product carried conflicting attributes across the commercial catalogue, the ERP and a legacy product-data system. Every automation candidate downstream — search, cross-sell, documentation reuse — stalled on the same root cause.
What we did: Data-quality diagnostic first: duplicate and conflict rates quantified by attribute class across the three systems. Then the one custom build the gates justified — a curated golden-record dataset with a matching pipeline and a human adjudication queue for the ambiguous tail, instead of another agent on top of bad data.
How it's measured: Measured as records reconciled per adjudicator-hour and downstream error rate in the workflows the dataset feeds. The dataset is the moat: three previously-killed automation candidates were re-scored against it and two re-entered the build queue.

02 / Heritage · named work

A decade of shipped products.

Before Synthetic Users, this team ran Impossible — a product and engineering studio whose client work is public and shipped. The engineering credential behind the gates.

Roche Regulated healthcare · clinician-in-the-loop

NAVIFY™ Tumor Board

From 2015, with Roche Diagnostics: designed, tested and helped implement the digital platform that organises tumor-board cases — structured patient and treatment data, ready when the specialist panel sits. The collaboration's methodologies and results seeded what became Roche's Diagnostics Information Solutions unit.

Google Frontier engineering · record time

Tango (pre-ARCore)

Designed and engineered the Constructor and Explorer apps for Tango, Google ATAP's augmented-reality platform and the precursor to ARCore — capturing, displaying and exporting 3D models of real spaces from a phone. Small, senior teams shipping cutting-edge work for a company that writes the book on engineering.

Roche Pandemic-scale diagnostics

NAVIFY™ Pass

With Roche Information Solutions through COVID-19: integrated digital solutions and data capabilities for diagnostic labs, clinics and patients — then extended into a platform for self-test kits and future pandemics, shaping rollouts across EU, US, Australian and Asian markets.

Samsung Clinical-grade consumer health

HeartWise

Home-based cardiac rehabilitation on Samsung watches and phones plus a clinical portal: heart rate captured before, during and after prescribed exercise, symptoms and compliance shared securely with the physician's case manager.

Fairphone Longevity by design

Fairphone software

At the start of Fairphone's journey, built the Fairphone Updater — OS updates, alternative-OS installs and a simplified experience that kept the world's first ethically-sourced phone useful for longer.

“Impossible stands out because of their ability to quickly ramp up and parachute in, understand what needs to get done and execute it.”

Arjun Shrinath · Senior Business Product Manager, Roche · on working with Impossible

“Impossible was instrumental in the success of the movement.”

Bas van Abel · Founder & CEO, Fairphone · on working with Impossible

The studio also shipped ventures of its own — among them Synthetic Users, now the team's main company, and Nikabot, a time-tracking assistant whose DNA lives on in Baseline's time capture.

03 / Measurement philosophy

The analytics layer decides what productizes and what dies.

Baseline and delta on every workflow. Internal wins are proven first; only what clears the bar is scaled, and only what scales is ever considered for productization. Nothing ships on a hunch.

04 / Sectors & work types

Structured authoring & documentation
Change impact & product data
Multilingual content operations
Catalogue & master data
Knowledge & training content
Document-heavy operations

note: specific engagement detail and references available under NDA at the intro call.

Want the specifics?

We share cleared metrics and named references under NDA on the intro call.

Book an intro call