From Satellite Photo to Priced Proposal: How the Solset AI Design Studio Works

Most solar design software and most solar CRMs are two different products made by two different companies. A rep designs a system in one tool, then re-types the customer's name, the site address, the module count, and the price into a second tool to actually send a quote. Every re-typed number is a place a mistake can enter, and every switch between tools is time a customer spends waiting. The Design Studio inside Solset AI exists to remove that seam entirely — the design and the quotation are the same record, on the same platform as the lead, the customer, and eventually the project.

Step 1: Tracing the roof

A design starts with a roof outline traced over satellite imagery — either by hand, drawing the polygon directly over the aerial photo, or with AI assistance proposing a first-pass outline for the designer to review and adjust. Either way, the outline is drawn in real, site-local measurements from the start, not eyeballed afterward, and it stays fully editable — corners can be dragged, edges snap to 45° angles and to each other, and the whole layout supports proper undo and redo so a designer can experiment without fear of losing earlier work.

Step 2: Placing the array — and pricing it against real hardware

Once the roof shape exists, panels can be auto-filled to cover the available roof, or placed by hand when the goal is a specific system size rather than maximum coverage. The panel catalogue isn't a set of generic wattage presets pulled from nowhere — it reads a tenant's own product records, so a designer can pick the exact module and inverter their company actually stocks and sells, complete with real electrical specifications (Vmp, Voc, Imp, Isc, temperature coefficients) rather than class-typical placeholders. A module with an incomplete or inconsistent datasheet is rejected from the catalogue rather than silently packed with guessed numbers, because a wrong open-circuit voltage doesn't just produce a slightly-off design — it can size a string past what the inverter is rated to handle.

Step 3: A yield estimate from real irradiance data, not a rule of thumb

The generation estimate is computed from irradiance data for the site's actual coordinates, sourced from PVGIS — the European Commission's solar resource service, widely used across the industry for bankable generation estimates — not a flat "kWh per kWp" multiplier applied uniformly regardless of where the roof actually sits. A rooftop in Rajasthan and a rooftop in Kerala do not receive the same sunlight, and a serious yield number has to reflect that.

Step 4: A real structure and steel takeoff

The Design Studio models seven real mounting structure types the way a fabricator would — pole count first, because that is what actually drives the steel: single pole static, double pole static, double pole with wind struts, single and double pole seasonal-tilt, ballasted (non-penetrating), and flush rail for a sloped or sheeted roof. Each produces an actual takeoff — legs, purlins, rafters — and a ballasted design reports its ballast load separately from its steel, because one is what the customer buys and the other is a load the roof structurally has to carry. A flush design correctly quotes no separate frame at all, since flush modules sit on brackets that are already counted individually elsewhere in the bill of materials.

Step 5: Seeing it in 3D, not just a flat plan

A plan view is enough to check panel placement, but it can't show a customer what a raised, tilted frame actually looks like standing on their roof, or whether a neighbouring structure will cast a shadow across the array in winter. The Design Studio renders two 3D views of the same design: a dependency-free axonometric view that runs on a phone with no extra download, and a WebGL "Site 3D" view with a photographic ground plane and full perspective for a more persuasive customer-facing walkthrough.

Step 6: One-click report, one-click quotation

A technical design report — a to-scale layout drawing, the 3D figure, the yield estimate, and the equipment specification — generates as a PDF straight from the design, with no separate document to assemble by hand. The same design can also generate a public, read-only share link a customer can open with no login, useful for sending to a financier or simply forwarding by WhatsApp. And when it's time to formalize the deal, the same design turns into a quotation in one click: the bill of materials is priced from the tenant's own product catalogue, and a line item that doesn't match a real product is flagged as unpriced rather than filled in with an invented number — an obvious gap gets fixed by a human before it goes to a customer, which is far safer than a plausible-looking guess nobody double-checks.

Why this matters more than any single feature

None of these steps is remarkable in isolation — plenty of tools trace roofs, plenty of tools estimate yield, plenty of tools generate PDFs. What is different is that all of it happens against the same lead, the same customer record, and the same product catalogue that the rest of the business already runs on. The design doesn't need to be exported anywhere for the quotation to exist, and the quotation doesn't need to be re-keyed anywhere for the invoice to exist once the job is won. See the full Design Studio feature page for a deeper look at each part, or the pricing to see where it fits in a plan.