The grid has
a repo now.
The model repository for power-system planning. Models, scenarios and decisions stay connected, so engineers and agents validate, analyse and publish from one governed source.
See how it works →Live with National Grid DSO
Backed by Y Combinator · 20VC
Leap Forward Ventures · 33East
01our thesis
Every grid plan begins with a model. Almost none stay connected to it.
The engineering model lives in specialist tools — PowerFactory, PSS®E, IPSA, CYME, ETAP. The plan is assembled across queues, spreadsheets, studies and reports, and slowly detaches from the network that produced it. Squid keeps the model, assumptions, evidence and decisions connected — from first study to published plan.
The model is not an input to planning.
It is the foundation of it.
02the repository
Bring the models together.
Import power-system models and the context around them — PowerFactory, PSS®E, IPSA, CYME and ETAP exports, GIS layers, spreadsheets and studies. Squid resolves them into one governed model — every version, scenario and source kept, without overwriting the approved baseline.
↓ resolve · validate — CIM (CGMES)
kelsworth 33kv · 2 records → 1 object
v40 — v41 — v42 ● approved baseline
every object · every source attached
diff any two versions
03model control
Make the network trustworthy.
Resolve network assets across formats, standardise the model, and run topology and parameter checks — while scenarios branch, merge and stay traceable to the approved baseline.
See what was supplied, what Squid inferred, what changed and what still needs an engineer.
v38 — v39 — v40 — v41 — v42 ● approved baseline · network@v42
scenario / connections · gate 2 — open
scenario / reinforcement · option B
option B approved · merged → v42
+ second 33 kV circuit · 64 MVA3 queued projects unblocked
04model interfaces · exchange
Put the model to work.
Exchange approved models. Build scenarios. Query and analyse the network. Publish public, shared or private views. Give planning agents a controlled model and evidence layer to work from.
05the planning picture
What exists today. What asks to connect. What must be built. Why.
Squid connects the current network, connection queue, reinforcement plan and drivers of need, each tied to the constraints, assumptions and evidence behind it. Every team and every approved agent works from the same model and the same history.
- which model version and scenario produced this investment need?
- show the evidence behind the Kelsworth flexibility requirement
- what changed between the draft and approved plan for Netherford?
- what changes if the Netherford upgrade slips by two years?
- which option releases the most capacity per pound?
ask the model
network@v42
3 dependent projects: Kelsworth BESS, Brayfield Solar and Dunmore Housing. Option B (second 33 kV circuit) releases +64 MVA.
ask the model
network@v42
3 dependent projects: Kelsworth BESS, Brayfield Solar and Dunmore Housing. Option B (second 33 kV circuit) releases +64 MVA.
06power-system analysis
AI plans the study.
The solver does the physics.
Agents changed how engineers build software because code is structured language: versioned, diffable, testable. A power-system model is a language too. Squid makes it one agents can work in. AI models will keep changing; the foundation underneath them stays.
AI can only engineer what it can model.
Squid assembles the approved model and assumptions, prepares the study cases, calls the analysis engine, compares the outputs and returns a replayable result for engineer review. Corrections and approvals stay with the model, so the next investigation starts with more context.
> which reinforcement unlocks the most ready-to-connect capacity before 2030?
ask the model
- >what changed between v41 and v42?
- >which assets failed validation?
- >what constrains 24 MW at Kelsworth?
- >compare the least-regret options.
- >publish the evidence behind this flexibility requirement.
AI investigates · the solver computes · the engineer decides · the model remembers
07standards & benchmarks
Built on the common standard.
Squid contributes to the GB CIM Advisory Group on the common model standard, and runs its own programme benchmarking AI agents on real power-system engineering tasks.
standards
One shared way to describe the grid.
GB CIM Advisory Group: how GB networks describe and exchange power-system models
ai benchmarking
AI, measured on the real grid.
squid’s benchmark programme: what agents can really do on power-system engineering tasks, before they touch a plan
08proof
Launched with National Grid DSO.
“We’ve worked closely with our partners Squid to bring the data to life and unlock greater value for our stakeholders.”
National Grid DSO — Squid’s launch customer — used Squid to turn static flexibility reporting into a live, explorable stakeholder view backed by the network model.
Doerte Schneemann · Head of Flexibility Markets · National Grid DSO
Read the announcement →live with — National Grid DSO
launching with — Northern Powergrid + further European network operators
fit & deployment
Start with an export. Keep the stack.
Squid deploys alongside the tools and source systems utilities already run. Begin with the models teams already produce.
no rip and replace
Squid runs beside existing analysis tools, GIS, data platforms and databases. Nothing is replaced — approved models stay exchangeable.
live in days
Start from a model export, or connect the source directly. Squid understands power-system formats, so the first models are validated and explorable in days — not months.
expand on trust
Begin with one set of models and scenarios. Add teams, integrations and workflows as the model earns its place.
security-by-design
permissioned access · source lineage · version history · engineer approval · audit trail
EU + UK GDPR — in place
ISO 27001:2022 — certified
SOC 2 Type I — attested
SOC 2 Type II — audit in progress
Bring us a planning question.
Bring one model export and the question around it. We’ll show how Squid connects the model, scenario and evidence — without replacing the tools engineers already use.
Talk to us →