Your AI
in silico designer.

Any computational biology task — we haven't yet hit a ceiling. Common ones: primer and probe design, multiplex assay design (qPCR / RT-qPCR / ddPCR), CRISPR guide design across modalities (KO / KI / base / prime editing / CRISPRa-i), cloning and construct design, codon optimization, gene expression and RNA-seq, WES / WGS, variant calling and ACMG classification, cryptic splice analysis, microbiome analysis, conservation and phylogenetics, literature reviews. If your task isn't on this list, bring it — the pilot is the right place to find out together.

Yes — any genome with public reference data, and we routinely add custom genomes when a project needs one. The agent has run designs and analyses across mammalian systems (human, mouse, rat, CHO, NHP), bacteria (E. coli, Salmonella, Pseudomonas, Mycobacterium, Clostridium, Lactobacillus and others), yeast and fungi (S. cerevisiae, S. pombe, P. pastoris, A. niger), viruses (influenza, SARS-CoV-2, RSV, herpesviruses, HIV, arboviruses), plants (Arabidopsis, rice, tomato, maize, soybean), and synthetic / engineered chassis. If your target organism isn't in standard databases, send the reference and the agent will work against it directly.

ChatGPT can answer questions about biology. It cannot pull sequences from NCBI, run alignments, design assays against conserved regions, BLAST-check specificity, call variants, score splice sites, build phylogenies, and ship a publication-grade QC'd report. Our agent works in a permanent, persistent computing environment — its own Linux workspace with real bioinformatics tools (BWA, SAMtools, GATK, MUSCLE, and dozens more), live access to genomic databases (Ensembl, UCSC, NCBI, gnomAD), and the ability to browse and operate web tools (CRISPOR, CHOPCHOP, UCSC In-Silico PCR, BLAST). And critically: the agent is embedded and customized specifically for your lab — your tools, your reagents, your conventions, your output formats — and it learns and adapts to your preferences with every task.

Every output is built on a chain of verifiable steps, not generative recall. Before the agent embarks on a task, it reads the relevant literature and current consensus, so it has full comprehension of the problem before any pipeline runs. It then executes real tools on real data, retrieves real references, and ships a full audit trail: numbered scripts, intermediate alignments, BLAST outputs, JSON checkpoints, tool versions, and timestamped database queries. After the primary pipeline completes, independent reviewer agents re-check the methods, the cited literature, and the numerical claims. Every report includes a mandatory Caveats & Limitations section where the agent flags what it could not resolve — assumptions made, low-confidence calls, missing data. During the pilot month, a qualified human reviews every report and configures the agent specifically for your needs before delivery.

Three layers of catch: (1) the agent self-reviews and flags limitations in every report; (2) an independent reviewer agent re-checks methods, claims, and literature before delivery; (3) during the pilot month, a qualified human scientist reviews every output before it reaches you. Senior-scientist human review remains available beyond the pilot as part of the full license — during the pilot it's the default on every task. If something still slips through, re-runs are free — the full reproducibility folder makes it trivial to iterate with fresh inputs or tightened constraints.

Each engagement runs in an isolated environment. Your data is never shared with other clients, never used for model training, and never retained after delivery beyond the project lifecycle. Our pilot agreement explicitly covers confidentiality, and the obligation survives termination. For especially sensitive workloads (clinical genomics, unpublished discoveries, regulated environments), we can run forward-deployed on infrastructure you control.

You do — fully. All client inputs and all outputs the agent produces (reports, figures, scripts, data files, intermediate files) are your sole property under the pilot agreement. We retain no rights to your data or derived outputs. Our underlying AI systems and proprietary methods are not transferred under the agreement, but everything generated for your project is yours to use, publish, and share.

Standard turnaround is 24 hours per task. Most analyses are processed by the agent within 30 minutes to a few hours; the rest of the window is reserved for human review and delivery formatting. Work that would typically take a wet-lab scientist days or weeks to do manually.

Onboarding starts with a kickoff call to capture your conventions, target organisms, preferred tools, and output formats — these get encoded into a per-team preferences file that the agent reads on every task. You submit tasks via the portal or email; each one comes back as a publication-ready deliverable with a full reproducibility folder. A qualified human reviews every output before delivery. At the end of 30 days, either party can decline to continue with no further obligation.

Every engagement starts with a free pilot task so you can evaluate the quality risk-free. After that, a fixed €500 / month pilot trial gives your whole team unlimited task submissions for 30 days. Team licenses (annual or monthly) are tailored to team size and usage. Book a call to discuss what fits your team.