Data Solutions

Our Story

Data Solutions grew out of delivery work, not a whiteboard. For years our engineers moved project truth through CAD, GIS, spreadsheets, PDFs, KMZ and SHP files, email attachments and field notebooks. The tools were useful. The problem was that the truth kept spreading across too many places.

One drafter writes a cable as ABC 4x50, another as 4x50 ABC, a third as A.B.C 4*50 mm2. A structure that should be S190 appears as 190S or S-190. Same cable. Five names. One avoidable delay, repeated across thousands of features and every revision.

We built Data Solutions by solving those problems in our own projects first, then spun it off from Engineering Data Solutions so that utilities, engineering firms, EPC contractors and agencies can run the same reproducible workflow. Governance and transparent data flow are the foundation. Encoded domain rules and network computations carry the predictable engineering on top: pole generation, service drops, load stamping, cable and phase rules, structure assignment, voltage drop, clearances and printing. The engineer sets the intent; the system carries the repeatable work.

One Record, Top to Bottom

Data Solutions is a data management system for engineering workflows, integrated vertically. Field capture, design, engineering checks, printing, project management and messaging all read and write the same governed project record through one API. Nothing is exported from one stage only to be re-imported, renamed and reconciled in the next.

Open at the Edges

Vertical integration does not mean a closed box. Data enters and leaves through controlled doors: GIS and CAD layers (SHP, KMZ/KML, GeoJSON, GeoPackage), spreadsheets, survey-grade observations, utility reference datasets and cloud geography, and two-way model exchange with the line-design software engineers already use. Results leave as KMZ for Google Earth, GIS packages, XLSX schedules and BOQs, PDF drawing sets and client templates, with record identifiers preserved. Form Factory defines what each record must carry before it is accepted; Style Center keeps one visual language from the phone screen to the printed sheet.

DS Field: Capture Facts Offline

DS Field is the mobile, offline-capable capture surface. Forms built in Form Factory can require field presence, minimum GPS accuracy, photos and controlled values before a record is accepted or flagged for review; survey-grade RTK/DGPS observations and their precision metadata travel with the record where the project standard demands them. When connectivity returns, records sync with their attributes, evidence, geometry and status intact.

DS GridDesign: LV Network Design

DS GridDesign is the design workstation for greenfield extension, fill-in, reinforcement, mini-grid and as-built conversion. It starts from a governed record: imported designs pass through an intake workspace that maps columns and layers into the domain model, renames inconsistent attributes and previews real rows before anything touches the live project. Bad names, missing values and duplicates are stopped at the door.

The engineer sets the intent and the rules carry the drafting: build the network, place and number poles, connect customers, assign service drops, size cables and transformers, and apply structure, assembly, stay and earthing rules aligned to local codes and client templates. A failed check points back to the feature that caused it, never to an anonymous drawing layer.

Voltage Drop: Every Customer Checked, Every Reinforcement Costed

Voltage drop is not a spreadsheet run at the end. Every processed LV transformer is solved as it is designed: a four-wire, phase-by-phase calculation from the transformer terminals to each customer's supply point, at design load.

Every customer, every section. Supply voltage and percentage drop for each customer, split between transformer, feeder and service cable; loading of every cable section and of the transformer against its rating; feeder length flagged for the area class. Results are stamped on the design itself, so the map can be coloured by drop.

The project's own standard. Limits come from the project's rule set: the national distribution standard where one is confirmed, IEC 60038 (±10 % of 230 V) where it is not. The report says which one it applied.

Reinforcement, not rejection. When a customer fails, the engine proposes the cheapest change that works: one conductor size up, span by span, where it removes the most drop per unit of cost; a new circuit cut where it leaves the fewest customers failing; or the next transformer size, never an implausible leap.

Year by year. Load growth to saturation turns the result into a schedule. Reinforcement needed at commissioning is flagged as required; later years are scheduled suggestions.

Line Engineering: MV and HV Lines on Real Terrain

DS Grid is the governed model for overhead lines: alignments, structures, tension sections, terrain, cables and design criteria in one versioned package. Every edit is a typed command applied to an exact revision. Revisions are immutable, can be compared entity by entity, and carry review decisions (draft, submitted, approved or rejected) with a reason, fenced on the revision the reviewer actually saw.

Engineering Maps, Printed by AI Agents

Drawing sets are where projects lose weeks: hundreds of LV transformer sheets, MV plan-and-profile booklets, key maps, schedules and legends, rebuilt by hand every time the design moves. In Data Solutions the drawing set is an output of the record. Layouts, styles and per-sheet exceptions are governed documents; an AI agent composes, renders and visually checks every sheet, and the engineer reviews the set before it is issued.

LV prints in two deliberate modes: on its own transformer sheet it carries the categorical engineering palette; inside an MV booklet it becomes bluish context under the red MV route. The two are never mixed.

• LV transformer sheets (A0/A3) with pole and customer schedules, quantity tables and legends;
• MV plan-and-profile booklets (A3) with cover, naming page and key map;
• voltage-drop reports (A4) and one-line-per-transformer overviews (A3);
• project atlases and custom-area maps, KMZ for Google Earth, GIS packages and XLSX BOQs aligned to client templates.

Printing runs on the engineer's desktop or a signed-in project server, close to the data, and it is the same command whether a person or an agent starts it.

Project Work and Messaging: the Work, the People, the Paper Trail

Engineering correctness is not enough. Scope moves while survey is running, packages wait on each other, instructions arrive by email and decisions by phone. Project Work extends the same governed record to the work itself, and messaging keeps people talking beside the record rather than instead of it.

1. CommentA reviewer comments on a transformer or a line structure.
2. TaskOne command turns the thread into a linked task, placed on the map at the structures it names.
3. AssignThe task is offered to a person, who accepts or declines; the project conversation shows it.
4. CorrectThe engineer, or an agent, applies a typed correction as a new revision.
5. CloseProgress rolls up, review and closeout follow, and the reply is filed as a record.

Dashboards read the same records: progress against the fleet, health, physical overlaps and who drew what. Nobody is asked to assemble a status report, because the status is already in the record.

Bring Your Own AI: Your Intent, Our Engineering

Engineering teams have spent years learning tools before they could trust their own outputs: hundreds of GIS functions, CAD menus, cleaning routines and export dialogs. Data Solutions ends that apprenticeship. Pair the AI assistant you already use (Claude Code, Codex, GitHub Copilot or any MCP host) with the product, say what you need in your own words, and the work runs on your project's data, under its standards and its governance.

No more years of tool mastery

• Ask in your own words. "Buffer around the line" finds the buffer command first. That an outsider's word for an operation leads to the right command is a tested property of the product, not luck.
• Complex inputs, fast. A client spreadsheet with unfamiliar columns, a contractor's shapefile or KMZ, a backup from line-design software: the agent inspects sheets, columns and coordinates, proposes the mapping into the domain model and stages a cleaned batch. The tedious part of intake becomes a review.
• GIS analysis without a GIS course. Corridors, lengths and areas, points along lines, crossings, and only the new line portions of a contractor's network against the authoritative base.
• Cloud geography with the price first. Parcels, customers and administrative units inside a transformer's area come from a read whose cost is planned before it runs.

Prefer the screen? The same flow runs in the DS GridDesign desktop with some learning, on the same commands. You can also work side by side: the AI opens a model in your Profile view and selects the structures it is talking about, you refine by hand, and you iterate together until every design decision meets the standard.

Under the conversation sits the ds command line: nearly 600 typed commands across 24 domains, served to agents through MCP with domain skills that ship with the desktop. The agent learns the product from the product:

$ ds capabilities --search "buffer around the line"
data.vector.buffer      Buffer each feature by a fixed distance into a polygon zone.
…
$ ds capabilities --search voltage
design.lv.voltage-drop  Check LV voltage drop by year and recommend compliant upgrades.
…
$ ds capabilities dsgrid.project.set-approval
dsgrid.project.set-approval  (contract v1)
Record a review decision on one model revision (needs --yes).
  effect global_write  authority headless_project  available

Human ownership, built in

DS Solar: Design, Finance and Reports

DS Solar plans generation from the same record. Demand starts from the design itself (customer categories, transformer schedules and network lengths), and weather and PV reference data arrive with a verified content digest, so every result traces back to its inputs.

Platform, Security and Scale

Data Solutions runs in the cloud and on the engineer's own machine. The desktop and the project server keep working offline and synchronise when connected; heavy computation runs in native code; dense networks stream and tile, so engineers pan, inspect and edit country-scale programmes without splitting the job into fragile files. Data is isolated per project, roles decide who reads, edits, approves and exports, and deployment patterns are available for clients with data-residency requirements.

The Team Behind the Product

Choose Your Adoption Path

Client references, named project evidence, screenshots and sample deliverables are shared in tender submissions and on request, under the confidentiality the procurement requires. They are not published on this page.

Contact

Company: Data Solutions
Website: datasolutions.rw
Email: magese@datasolutions.rw
Phone: +250 783 000 360
Location: Kigali, Rwanda
Product line: DS Field / DS GridDesign / DS Solar