Sandbox
Fast, evidence-backed feasibility
We previously built Sensat, an AI infrastructure SaaS company, from zero to eight-figure revenue deployed across $250bn worth of critical projects worldwideOlympics, Heathrow, HS2, Severn Trent, etc (background).

Now we are solving the single greatest bottleneck holding back trillions in global infrastructure investment – feasibilityInfrastructure feasibility is a preliminary, in-depth investigation that determines if a proposed project, such as data centres, fusion plants, roads, or utilities, is viable, sustainable, and worth the investment.

ProblemA $350bn
feasibility gap

Every year, ~10,000 infrastructure projects enter early-stage feasibility in the UK aloneOur beachhead market..

Teams spend between $25k–$250k$3.5tn global CAPEX / year, approximately 10% is spent on early stage feasibility investigating each site, yet 50-70% of these projects die here, wasting hundreds of millions in abortive capital.

Today:

The process is manual, sequential, and depends on fragmented consultant reports

Decisions rely on static opinions rather than a concurrent, evidence-based system

There is no canonical object that represents "what we know about this site"

Despite billions spent, the system never gets smarter.What's changed is tolerance. Capital is tighter, planning scrutiny is higher, and teams are under pressure to justify early decisions with far less margin for abortive spend. The cost of getting feasibility wrong has always existed; now it's visible, measured, and no longer acceptable. Each site resets the process, knowledge is discarded, and feasibility is treated as a one-off opinion rather than an accumulating state.

The feasibility bottleneck didn't suddenly appear.

A series of structural changes have shifted pressure earlier in the development process, making speculation untenable and manual feasibility unsustainable.

Regulatory gating has moved earlier

Speculation is no longer acceptable

Grid connections, planning, and other approvals now prioritise projects that are development-ready, not simply first in line.

Harder investment gates

Early decisions now face real scrutiny

Investment committees and asset owners increasingly require defensible answers upfront. Weak feasibility is challenged earlier, with less room for intuition or narrative.

Planning and consent slowdown

Delay has become structurally expensive

Planning timelines have lengthened while demand continues to rise. Carrying non-viable sites deeper into the process now creates material downstream risk.

Consultant bottlenecks

Manual feasibility can't keep up

Consultants remain necessary, but their speed and consistency no longer match the pace or pressure of early-stage decision-making across multiple sites.

Data without reasoning

Inputs exist but judgement hasn't scaled

Policy, constraints, precedents, and rules are digitised and accessible. What hasn't scaled is the ability to reason over them quickly, consistently, and explainably.

Infrastructure feasibility

Sandbox allows feasibility knowledge to compound over time

SolutionAI-native reasoning

Sandbox encodes how experienced infrastructure teams actually reason about feasibility and runs that reasoning instantly and consistently across sites.

We standardise feasibility and apply deterministic AI to remove manual, sequential analysis. This allows teams to reach defensible decisions early and eliminate non-viable sites before capital and effort are consumed.
Concretely, we:

Standardise feasibility into a canonical SAFESite Assessment and Feasibility Evaluation object, turning a one-off exercise into an accumulating asset

Apply deterministic AI across explicit rule systems (planning, engineering) to replace opinion with reasoning

Collapse sequential workflows into parallel ones so bad sites are identified early rather than late

The result is a clear, evidence-backed view that teams can trust, share, and defend (and fewer months spent validating sites that were never going to work).

Deterministic AI matters because feasibility decisions must be explainable, auditable, and defensible to planners, boards, and regulators, not probabilistic suggestions.

Right to winWe've lived this problem

Sandbox is the product that surfaced consistently while working through feasibility failures with infrastructure owners and operators.

We've met and interviewed 200+ owners and seen first-hand where today's infrastructure feasibility fails in production.
Sandbox applies our playbook for building a $100m+ company in this exact market to feasibility-grade AI, with direct access to early customers and design partners.
Sandbox feasibility record

Sandbox creates the canonical feasibility record that accumulates value through time

MilestonesPlan

30 days

Prove automated SAFESite Assessment and Feasibility Evaluation

Prove that an automated SAFESite Assessment and Feasibility Evaluation can replace first-pass consultant feasibility for a defined sector.

60 days

Change real decisions

Demonstrate that SAFESite Assessment and Feasibility Evaluation outputs materially change feasibility decisions and prevent external spend.

90 days

Default feasibility artefact

Establish SAFESite Assessment and Feasibility Evaluation as the default, repeatable feasibility artefact teams rely on across sites.

Ask$2.4M allocation

We're opening a $2.4m allocation to back a proven team to rapidly move design-partner pull into production deployments.
Feasibility is the first decision every infrastructure project makes, and today it's the least structured.

Sandbox exists to turn that decision into the default feasibility layer projects are organised around.
RoundPre-Seed
InstrumentSAFE/ASA
Cap$15M post-money
Target round size$2.4M