Xenobots

The Science

Computationally designed organisms were just invented in 2020 and yet they possess the power to transform fields such as human health, climate remediation, and bioengineering. A biobot, the general name of an agential biology, senses, decides, moves, and acts on its surroundings in a single sub-millimeter package — a combination available neither to passive materials nor to the cells it is built from.

• Sensing. Detect gradients, targets, and conditions in situ. Living cells arrive with a sensing array already built in.
• Navigation. Move through fluid and other complex surroundings, following pH, thermal, chemical, and light gradients.
• Signal amplification. Turn a small local signal into a large detectable one; each bot tells two friends.
• Payload delivery. Carry and release useful material precisely where a target is found.
• Sequestration. Capture, concentrate, and transport unwanted material for later removal.
• Transformation. Act chemically on a target — the difficult step conventional methods can’t reach.

Biobots are programmable on three levels. Body plan and population behavior produce functions that no constituent cell possesses; the capabilities are designed and collective.

• Cell function — what the cells do. Skin and cardiac cells arrive with sensing, contractility, and metabolism already built in. Synthetic biology circuitry extends them further.
• Body function — how they are arranged. Conformation determines locomotion, gradient-following, and object handling. The same cells in a different body plan do a different job.
• Swarm function — what populations do together. Signal amplification, aggregation, and pile-making. One bot detects; the population makes the detection visible.

The Technology

Affectionately known as MomBot, this AI-driven robot-scientist is a platform built to automate closed-loop experiments in developmental biology and synthetic morphology. Rather than navigating physical 3D space like a standard mobile robot, MomBot navigates the space of chemical, electrical, and environmental cues to study how cells self-organize into biological robots such as xenobots. Tufts University houses MomBot, and scientists there are using it to explore bioelectric patterns, anatomical memory, cellular regeneration, and motile capabilities.

Someday, we will be able to sit in front of this combination of software and robot scientist platform, specify the shape of an animal, plant, organ, or biobot that we want, and it will convert that shape specification into a set of stimuli that will be given to cells to get them to build exactly that shape (no matter how novel — total control over growth and form).

Mike Levin

MomBot is the first of a forthcoming class of agential translation interfaces that will enable human scientists to more deeply understand and communicate with the collective intelligence of xenobots. It is a prototype to the Anatomical Compiler, a future system representing the long-term endgame of the science of morphogenesis. It will be a platform to produce biobots at scale, uniting human ingenuity, the awesome computational power of AI, and biology’s limitless potential.

Fauna-Sponsored Experiments

Before any application, biobots must survive and move in applicable media. Experiments presently underway gate everything downstream, and they are where Fauna’s sponsored research at the Levin and Bongard labs is concentrated now.

• Seawater, for aquaculture. Cold tolerance below 60°F for deep-water salmon; sustained function above 80°F for tropical shrimp.
• Blood, for diagnostics. Viability at human body temperature and viscosity — the gate for sepsis work and everything else in human health.
• Hydroponic freshwater, for agriculture. Mid-60s to high-70s°F is well tolerated; the open question is pH 5.5–6.5 from nutrient solutions.
• Wet soil, for remediation. How wet must topsoil be to support biobots — and can we bring the contaminant to water instead?

Temperature and pH ranges are Fauna-internal experimental targets derived from partner operating conditions, not published results. Cell viability envelopes are established in the peer-reviewed literature (Blackiston et al., Sci. Robot. 2021); the operating conditions above are the targets those envelopes must be tested against.

Alongside this, sponsored research at the Levin and Bongard labs is characterizing sensing, motility, and state readout to the tolerances a real deployment imposes — how precisely a biobot detects, how reliably it navigates, and how legibly it reports what it found.

Anticipated Applications

Detection and identification comes first, but binary detection is not enough: every operator already knows something is there. The value is in which compound, which pathogen, at what concentration — the sensing and signal-amplification capabilities working together.

The market is not asking for a better sensor. It is asking for something that senses, follows complex rules, and then acts. The functions that follow are the ones that act on the physical environment.

• Absorption. Take up contaminants in situ and collect the bots later.
• Sequestration. Concentrate diffuse material into a designated area for removal.
• Transformation. Change a compound chemically rather than eliminate it — PFAS is the canonical case.
• Destruction. The endpoint for any remediation context.
• Environmental remediation. Detection, sequestration, and partial transformation of contaminants in water and soil, including PFAS at parts-per-trillion thresholds. Partial is often sufficient: a biobot need not complete a transformation, only perform the step conventional methods cannot reach economically.
• Human health diagnostics. Ex-vivo interrogation of biological samples for signals conventional assays cannot resolve.
• Precision agriculture. Telemetry on crop and ecosystem health, and targeted biological control in place of broad-spectrum chemistry.
• Synthetic biology and bioprocess. Deeper telemetry on the internal state of host cells and fermentation systems.

Bio wins where bio is uniquely capable — where adaptability, specificity, self-assembly, and gentle interaction justify the development cost. Where chemistry is adequate and cheap, chemistry will win. Finding the boundary between those two is the present work.

Foundational science papers